international comparisons of administrative costs in ...studies are international comparisons of...

International Comparisons ofAdministrative Costs in Health Care

September 1994

OTA-BP-H-135NTIS order #PB95-109682

GPO stock #052-003-01398-3

Recommended Citation: U.S. Congress, Office of Technology Assessment, InternationalComparisons of Administrative Costs in Health/ Care, BP-H-135 (Washington, DC: U.S.Government Printing Office, September 1994).

For sale by the U.S. Government Printing OfficeSuperintendent of Document, Mail Stop: SSOP, Washington, DC 20402-9328

ISBN 0-16 -045320-8

Foreword

The complexity of the U.S. health care system has become an issuein the debate over health care reform. In recent years, researchershave published studies examining whether the adoption of a Ca-nadian-style, single-payer system in the United States would

substantially reduce the administrative expenses. At the heart of thesestudies are international comparisons of administrative spending.

This background paper examines what is known about administrativecosts in the health care systems of the United States and several othercountries. In addition to exploring the types of activities that constitutehealth care administration, it reviews studies that measure and comparethese activities in different countries, and it explores the potential useful-ness of such comparisons. Although a Canadian-style system in theUnited States might indeed result in significant administrative savings,international comparisons of administration in countries other than Can-ada may also be helpful under a multiple-payer system by identifyinghow to achieve more modest savings or efficiencies in the way we ad-minister our health care system.

The background paper is part of a larger project, International Differ-ences in Health Care Technology and Costs. One other background pa-per, International Health Statistics: What the Number.v Mean for theUnited States, was published in November 1993. The remaining back-ground papers in the series will examine international differences inspending for physician and hospital services, and health care technologyand its assessment in eight countries. The House Committee on Waysand Means, under Chairman Dan Rostenkowski, asked OTA to under-take this assessment.

Preparation of this background paper was greatly assisted by an advi-sory panel, chaired by Rosemary Stevens of the University of Pennsyl-vania. In addition, many other individuals provided information and re-viewed drafts of the paper. OTA gratefully acknowledges thecontribution of each of these individuals. As with all OTA documents,the final responsibility for the content of the assessment rests with OTA.

ROGER C. HERDMANDirector

Ill

Advisory Panel

Rosemary Stevens, Chair Louis P. Garrison, Jr.University of Pennsylvania Syntex Development ResearchPhiladelphia, Pennsylvania Palo Alto, California

Stuart Altman Annetine GelijnsBrandeis University Columbia UniversityWaltham, Massachusetts New York, New York

Jan E. Blanpain John IglehartLeuven University, Belgium Health AffairsLeuven, Belgium Bethesda, Maryland

Harry P. Cain II Ellen ImmergutBlue Cross and Blue Shield Massachusetts Institute of

Association TechnologyWashington, District of Columbia Boston, Massachusetts

Thomas W. Chapman Lynn E. JensenThe Greater Southeast Healthcare American Medical Association

System Chicago, IllinoisWashington, District of Columbia

Bengt JonssonStockholm School of EconomicsStockholm. Sweden

Kenneth G. MantonDuke UniversityDurham, North Carolina

Edward NeuschlerHealth Insurance Association of

AmericaWashington, District of Columbia

Jean-Pierre PoullierOrganisation for Economic

Co-operation and DevelopmentParis, France

Mark SchlesingerYale UniversityNew Haven, Connecticut

Note: OTA appreciates and is grateful for the valuable assistance and thoughtful critiques provided by the advisory panel members.The panel does not, however, necessarily approve, disapprove, or endorse this is background paper. OTA assumes full responsibility forthe background paper and the accuracy of its contents.

iv

Clyde J. BehneyAssistant Director, OTA

Sean R. TunisHealth Program Director

Hellen GelbandProject Director for International

Differences in Health CareTechnology and Costs

PRINCIPAL STAFFMICHAEL L. GLUCKStudy Director

David Kaufmana

Research Assistant

Laura Esslingerb

Summer Intern

Romulo Colindresc

Research Assistant

ADMINISTRATIVE STAFFBeckie EricksonOffice Administrator

Daniel B. CarsonPC Specialist

Carolyn MartinWord Processing Specialist

Carolyn SwarmPC Specialist

Preject Staff

CONTRACTORSWilliam GlaserNew School for Social ResearchNew York, New York

Steffie WoolhandlerHarvard Medical SchoolCambridge, Massachusetts

James HahnUniversity of North CarolinaChapel Hill, North Carolina

Norbert PaquelCANOPE ConsultingParis, France

Nancy HenesonEditorial ConsultantBaltimore, Maryland

a Until June 1993.b From May 1994 through August 1994.c From August 1994.

v

1 Summary and Policy Implications 1Implications for Policy 1What Is Health Care Administration? 2Aggregate National Estimates of

Administrative Costs 3

Comparisons Between the United States andCanada 3

Personnel as a Measure of Administration 4Technology To Simplify Administration 5

C o n c l u s i o n 6

2 Defining Administrative Costs 7A Typology of Administrative Costs by Function 8

An Enumeration of Administrative Activities 9

3 Measuring Administration 15Estimates of Administrative Costs and International

Comparisons 16

Personnel as a Measure of Administration 39

4 Technology To SimplifyAdministration 47Standardization and Automation of Insurance

Claims 48Health Cards 48

APPENDIXES

A Acknowledgments 57

c ontents

B Comparison of Health CareAdministration Found In FourCountries 59

vii

C Methods Used in Himmelstein andColleagues’ Analysis of U.S. andCanadian Health Care Labor Forces 66

D Acronyms and Abbreviations 70

REFERENCES 71

Summaryand

PolicyImplications 1

International comparisons of administrative costs are one re-sult of the debate over health care reform in the UnitedStates. Advocates of a single-payer health care system (inwhich a single organization reimburses health care provid-

ers for all health services provided to patients) have compared theadministrative costs of the United States with those of countrieslike Canada to support their contention that the administrativesimplicity of a single-payer approach would yield savings thatcould offset the cost of universal coverage.

This background paper examines administrative costs in thehealth care systems of the United States and other countries. Inaddition to exploring the types of activities that constitute admin-istration in the health care systems of several developed coun-tries, it reviews attempts to measure and compare these activities,and it explores the potential usefulness of such comparisons.

IMPLICATIONS FOR POLICYOTA’s analysis suggests several conclusions for public policy:

Most of the empirical literature on administrative costscompare the U.S. and Canadian health care systems. Thesestudies indicate that administering the Canadian system con-sumes a substantially smaller proportion of health care spend-ing than does the U.S. system. Imposition of a Canadian-stylesystem in the United States would substantially reduce admin-istrative costs, although estimates of those savings range wide-ly (from $47 billion to $98 billion in 1991 U.S. dollars).Analyses of the administrative costs in countries other thanCanada suggest that health care systems with more than asingle payer, entailing a choice of insurance plans along withdecentralized cost control measures and payment of providers, I I

2 I International Comparisons of Administrative Costs in Health Care

involve higher administrative expenditures thandoes a single-payer system.International comparisons of specific adminis-trative activities may suggest ways in which theUnited States can achieve worthwhile but moremodest savings or greater efficiency in the wayit manages its health care system without mov-ing to a single-payer system. For example, aselectronic technologies are used more exten-sively to administer the health care system, theexperience of other countries may help theUnited States manage those technologies moreappropriately or cost-effectively. Unlike theU.S.-Canadian comparisons that have domi-nated the empirical literature to date, this ap-proach to international comparisons would fo-cus on how well administrative investmentsachieve their goals, rather than just tallying thecosts.Qualitative and quantitative evidence indicatesthat among developed countries with pluralis-tic, multiple-payer health care systems. theUnited States invests a greater proportion of itshealth care expenditures in administration.Little information exists on which to judgewhether any extra benefits accrue in the U.S.system from these additional expenditures.International comparisons of administrationcan be useful in understanding the detailedmanagement of other countries’ health caresystems, how individual patients and providersinteract with that system on a day-to-day basis,and differences in the numbers and types ofworkers who administer different countries’health care systems.The experience of U.S.-Canadian comparisonsunderscores the robustness of overall estimatesof administrative costs using imperfect datagathered for other purposes, especially whencomparing single-payer and multiple-payerhealth care systems. While primary data collec-tion to study administrative costs might yieldmore accurate estimates, the added confidencein the results is probably not worth the addedcost and logistical difficulties of carrying outsuch efforts. For detailed looks at specific com-ponents of health care administration, however,

a bottom-up approach may be necessary to un-derstand why costs differ among systems thatare more similar, and to identify potential mod-est administrative cost-savings or efficienciesfor the U.S. health care system.

WHAT IS HEALTH CAREADMINISTRATION?Although most people understand administrationto include the paperwork necessary to run a healthcare system, more comprehensive and precise def-initions are needed to measure and compare ad-ministration internationally. Thorpe (38) has sug-gested for the United States a classification ofadministrative costs according to the functionsthey serve and the type of individual or organiza-t ion performing these functions. This scheme con-siders administrative expenses as investments thathelp deliver medical services more efficiently orequitably.

However, for the purposes of internationalcomparisons, this typology alone is not sufficient.It does not include the many functions found inhealth care systems outside the United States,such as the setting of budgets, the negotiation ofreimbursement rates with providers, and the proc-ess for deciding whether to purchase expensivemedical equipment. It also does not take into ac-count that different countries might use differenttypes of staff or technology or face different pricesin carrying out the same administrative functions.Finally, it is not detailed enough to guide research-ers in the direct measurement of administrativeexpenses.

Glaser (15) has developed a detailed protocolfor a bottom up measurement of administrativeexpenses in any country’s health care system. As apractical matter, however, gathering data from dif-ferent countries following this approach wouldentail enormous expense, time, and logistical dif-ficulties (if, indeed, it is even possible). To date, ithas not been done. Furthermore, development of aconsensus about the precise definition of adminis-tration may be only of academic interest at thistime. More useful analyses might look at specificadministrative functions in different countries to

Chapter 1 Summary and Policy Implications 13

identify aspects that might be adopted in theUnited States to improve efficiency in the healthcare system.

AGGREGATE NATIONAL ESTIMATES OFADMINISTRATIVE COSTSGlaser ( 15) has applied his general protocol formeasuring administrative costs to make qualita-tive, descriptive estimates of the nature and mag-nitude of expected administrative expenses in thehealth care systems of the United States, and ofthree countries often pointed to by proponents ofU.S. health care rcform: Canada, the United King-dom, and Germany. Even without numbers, hisanalysis suggests that the U.S. health care systemrequires a more complicated administrative appa-ratus than do other systems. However, the magni-tude of many specific administrative activities canvary from country to country. For example, theGerman system relies heavily on negotiationsamong payers and providers to allocate health careresources, while U.S. payers increasingly attemptto control costs by scrutinizing the appropriate-ness of medical services prescribed. Nevertheless,Glaser’s analysis provides useful insights into theday-to-day management of these countries’ healthcare systems.

The Organization for Economic Cooperationand Development (OECD) annually publishesdata on health expenditures and outcomes, includ-ing administrative spending, collected from itsmember countries. Relying on a definition devel-oped by the U.S. Health Care Financing Adminis-tration (HCFA), the OECD includes only the ad-ministrative cost of public and private insurance,leaving out the administrative costs of hospitals,other providers, expenses borne by consumers,health services research, and the share of generalgovernmental administration or tax collection de-voted to health. In addition, not every OECDcountry has provided data on health administra-tion, and the comparability of data from reportingcountries varies,

Even with these limitations, the OECD data doprovide some insights into the administrative bur-dens of member countries’ health care systems.

Administrative expenditures vary substantially,between 1 and 7 percent of total health expendi-tures. Countries like the United States, Germany.and the Netherlands with multiple. segmentedsources of health insurance tend to spend more oftheir health budgets on administration. And trendsin administrative costs tend to reflect changes innations’ health care systems. All else being equal,the per-unit administrative costs have tended. onaverage, to decline over time due to economies ofscale and technological changes. Data from the1980s on the entire health care systems of Swedenand Australia and the public sector insurance pro-grams of Canada, the United Kingdom, and theUnited States are consistent with this trend. On theother hand, new insurance benefits, increased pa-tient coinsurance payments. and other cost-con-tainment measures tend to raise administrativeburdens, as evidenced in France in recent years.

COMPARISONS BETWEEN THE UNITEDSTATES AND CANADAIn recent years a literature has emerged comparingthe magnitude of health administration in theUnited States and Canada. All use various exist-ing data sources to estimate the administrativecosts of the insurance, hospital, and physician sec-tors of the U.S. and Canadian systems. Thesestudies extrapolate their estimates of Canadian ad-ministrative costs to estimate the potential admin-istrative savings of adopting a Canadian style sys-tem in the United States.

Himmelstein and Wool handler offered the firstquantitative comparison using 1983 data (20) andupdated their analysis using 1987 data (54). TheU.S. General Accounting Office (GAO) (43,44)and Sheils and Young (36,37) have offered theirown studies, using similar approaches, but differ-ing in some data sources and assumptions. Takentogether, these comparisons suggest that a Cana-dian-style system in the United States could havereduced administrative costs by between $47 bil-lion (36,37) and $98 billion (54) in 1991. anamount equal to between 6 and 13 percent of totalhealth expenditures in the United States that year.Although this range is wide. the conclusion that,


all else being equal, adoption of a Canadian-stylesystem in the United States could yield substantialadministrative savings is robust.l

Although the data used in all of these studiesare imperfect, they remain a reasonable approxi-mation of reality. Furthermore, the estimated dif-ferences between Canada and the United Statesare large enough to conclude that substantial dif-ferences in administrative costs exist between thetwo nations. It is less clear whether the Canadianexperience is predictive of administrative costs inthe United States under a single-payer plan. Forexample, there could be a general cultural tenden-cy in the United States towards more complex ad-ministrative structures leading to higher adminis-trative costs, even if the United States adopted aCanadian-style system.

In a more general critique of these U.S.-Cana-dian comparisons, Danzon (6) argues that the in-surance overhead figures for the United States in-clude significant expenses such as premium taxes,investors’ return on capital, and investment in-come that are not really administrative, makingthe U.S. data not comparable to the administrativedata for Canada’s public insurance programs. Inaddition, she suggests that the Canadian systemhas unmeasured costs associated with excessivepatient waiting time and the loss in overall eco-nomic productivity as employers and consumerschange their behavior to avoid activities that aretaxed to finance the country’s health care system.Furthermore, she points out that strict compari-sons of administrative cost data do not capture thebenefits of the U.S. system associated with con-sumers’ ability to choose providers and insurers.

Critics of Danzon’s approach suggest that shedoes not measure costs in the U.S. systemassociated with consumers trying to understandand evaluate the benefits, costs, and complex re-imbursement rules of alternative health insuranceplans, workers locked into jobs for fear of losinghealth insurance, and employers who must man-

age their employees’ insurance benefits and whomay avoid hiring employees they believe may becostly users of health services. Other critics havealso questioned whether medically significantqueues actually exist for health services in Canada.

PERSONNEL AS A MEASURE OFADMINISTRATIONA significant component of a country’s health careexpenditures are personnel costs, including thesalaries of people who carry out administrativeduties. In work commissioned by the Office ofTechnology Assessment (OTA), Himmelstein andcolleagues have attempted to use occupationaldata from national censuses and surveys to inves-tigate trends and differences in the U.S. and Cana-dian health care systems. For the United Statesthey calculated “full-time equivalents” (FTEs)employed in the health care sector between 1968and 1992 using the U.S. Census Bureau’s CurrentPopulation Survey (CPS), an annual survey of60,000 households representative of the civilian,noninstitutionalized population. Data on Cana-dian health care workers come from the 1971 and1986 Canadian censuses.

Between 1968 and 1991, the number of healthcare workers in the United States grew from 3.98million to 9.79 million (about one and one-halftimes), although the number of administrativeworkers grew at a much faster rate—from 718,000to 2.60 million (more than two and one-halftimes).

Comparisons with Canada show significant di-vergences over time. In 1971 the United Statesemployed 22,000 FTEs per million population,while Canada employed 26,565. By 1986, the to-tal number of U.S. health FTEs had grown 53 per-cent, while Canada’s had grown only 19 percent.Nearly all of the U.S. excess in health personnel ascompared to Canada is attributable to the greaternumber of managers and support personnel in the

l~ese ~$timates Of cost savings do not tie into account the cost of increased utilization by insured consumers who would use more health

services as their out-of-pocket expenses decreased under a Canadian-style system, a complex issue beyond the scope of this paper.

Chapter 1 Summary and Policy Implications 15

United States. In 1971 the two countries were al-most identicail in the number of administrativepersonnel per capita, but in 1986 the United Statesemployed 8.226 administrative health personnelper million population, versus Canada’s5,807—that is, the United States had 42 percentmore administrative personnel per capita in 1986than did Canada. Among other categories ofhealth workers in 1986, the United States hadmore technologists and technicians (2,423 vs.1,988), and more licensed practical nurses ( 1,333vs. 1,002), but fewer registered nurses per millionpopulation (5.41 9 vs. 6.948).

This analysis provides policy makers with auseful means of examining trends in the Canadianand U.S. health care systems. Its results are con-sistent with other studies finding that the UnitedStates spends more on measurable health care ad-ministration than does Canada. However, laborforce analyses such as this one do have limita-tions. They do not offer a solution to the problemof the potential] y unmeasured costs of public] y fi-nanced systems suggested by Danzon. In addi-tion, the CPS data used by Himmelstein and col-leagues cannot be used to identify non-medicalpersonnel in the United States who perform healthcare duties in nonhealth care settings, such as ad-ministrative personnel in private firms who ad-minister their employees’ health insurance bene-fits and management consultants. Inclusion ofthese workers would only increase the disparity inthe number of administrative workers between theUnited States and Canada. The analysis also ex-cludes private insurance employees in the UnitedStates and government employees in both coun-tries because of the difficulty in distinguishingthose workers who administer health insurancefrom those who perform other functions in theseorganizations.

TECHNOLOGY TO SIMPLIFYADMINISTRATIONStandardization of insurance claims forms, elec-tronic submission and payment of insuranceclaims, and the use of card technology to store ad-ministrative and medical information are threetechnological innovations that may have the po-tential to reduce administrative costs in the U.S.health care system. Estimates of potential savingsfrom standardization and computerization of in-surance claims vary widely, but in the case of cardtechnology, it is possible to examine the experi-ence of other countries to help understand theirpotential implications for the United States.

Health cards can take several forms, includingsimple paper or plastic cards, cards with magneticstrips (like automated bank teller cards in theUnited States), or smart cards, which embed a sili-con microchip within a plastic, wallet-sized card.2

These cards can have several uses: health insur-ance cards that include information about pa-tients’ health insurance coverage to simplifyclaims and reimbursement procedures or hospitaladmittance; medical cards to store limited patient

Technologies with the potential to simplify the administrationof health care include smart cards that can store and processadministrative and rnedical infromation

2SCJ cr:il lc\\ commonly uwd ctird technologies also exist including optical cards similar to compact di~ks, cards with embedded holo-~r:irT1\, ~lrlL] ~:lr(]$ ~]cflgne(] t. fit irl[() \[ln(iard17ed SIO[\ ~rl personal ~ornpu[crs. several of these technologies can be combined in a single card.


medical records: emergency cards that includeessential medical information for medical emer-gencies; and health professional cards that limitaccess to confidential, computerized records toauthorized personnel only.

OTA commissioned a study of several healthcard systems used in France. This analysis pointedout that health cards are only one piece of an over-all system for administering health care and main-taining records. The decision to use cards, or tochoose a specific type of card technology, is de-pendent on the intended application, the intendedusers, and the cost. In France, implementation ofcard systems was hindered by concerns over theconfidentiality of medical information and diffi-culties in getting physicians, administrators, andpatients to keep information on cards or othercomputerized medical records. These issues arelikely to arise in the United States should a cardsystem be implemented.3 However, concerns aris-ing from French physicians’ tradition of not shar-ing diagnostic or therapeutic information withother health professionals or payers should notcause problems in the United States. The Frenchexperience suggests that protection of such priva-cy has less to do with the choice of magnetic stripor smart card technology than the privacy safe-guards built into the overall computer system.Any kind of system has the potential to limit the

amount of information in the system and access toit (29).

Although recent estimates suggest that stan-dardization and automation of the insuranceclaims process would lead to cost savings afterinitial investments, no estimates exist for the costimplications of health card applications by them-selves in the United States. The French experienceindicates that health card systems involve signifi-cant start-up costs, but that standardization of thetechnologies used for different health care ap-plications offer opportunities for economies ofscale since several applications can use much ofthe same infrastructure.

CONCLUSIONThe recent debate over health care reform has re-volved, in part, around the desire to control costsand to find resources to cover the uninsured. If areformed system were cheaper to run, moneywould be freed for other purposes. It appears thatonly by a dramatic change to a single-payer sys-tem can great savings be realized. But even in theabsence of a single-payer approach, it may be pos-sible to achieve modest, yet worthwhile savingsand more efficient means of providing health cov-erage and services. The search for these savingsand efficiencies may be aided by the study of ad-ministration in other countries.

Sic Clinton A~IlllnlS[ratlon TS ~ropse~ He~]~ securl(y A(I (S. ] 757) WOUld issue every American citizen and leg~l resident a Health Secu-

rity Curd, ulthough the Administration has not suggested that use of such a card would necessarily reduce administrative costs.

Administrativecosts 2

A s potential reform of the U.S. health care system has gar-nered more attention, so too has the perceived complex-ity of the current system compared with that of othercountries (12). Analysts have associated that complexity

with the administrative apparatus employed to manage the healthcare system, making estimates of administrative costs relevant tothe debate over health care reform. At issue is whether or not areformed U.S. system can realize administrative savings that canbe used to pay for extended coverage, new benefits, or overallspending reductions.

In the literature, administrative costs often are equated withwasted resources that could be turned to more productive use.Many advocates of single-payer health care systems and analystswho measure health care costs for national accounting purposes(20,31,54,55) believe this is so. Other analysts, with a view to-wards macroeconomic theory and health care management, focuson administrative expenditures as inputs to the production ofhealth (4,5, 18,38). Seen in this light, administrative expendituresare an investment in people and services that have (often unmea-sured) benefits such as making the health care system more equi-table, less costly, or more cost-effective. Because such invest-ment tends to be greater and easier to identify in a multipayersystem, the notion of administration as an investment is common-ly supported by advocates of managed competition or other re-form plans that preserve multiple payers (6).

This background paper explores administrative costs in theUnited States and other countries and conceptual issues such asthe one described above. It reviews actual attempts to measureand compare the administrative burdens of different countries’health care systems, and it examines whether international com-

-23

17


parisons offer any insights into how various ap-proaches to health care reform may alter adminis-trative spending in the United States.

The word “administration” conjures up imagesof paperwork, clerks, and managers. One politicalscientist has suggested a more formal definition:those activities that regulate or control the behav-ior of individuals in an organization, enablingthem to implement policy decisions and achievegoals (15). In health care, administration is gener-ally understood to include nonclinical activities,however, this simple definition may not bedescriptive enough to allow one to measure ad-ministration in the United States and compare itwith that in other countries. For example, biomed-ical research, classroom medical education, andhospital food services are nonclinical but not ad-ministrative. The sections that follow reviewmore detailed attempts to define and classifyadministrative activities in health care and con-sider their usefulness in trying to measure themagnitude of health care administration.

A TYPOLOGY OF ADMINISTRATIVECOSTS BY FUNCTIONThorpe (38) classifies administrative activitiesand their associated costs according to the func-tion they serve and the type of individual or orga-

nization performing them. He shares the view thatthese cost are “inputs” to the production of admin-istrative services that help insure against illnessand deliver medical care. In his scheme, total ad-ministrative spending equals the sum of 1 ) “trans-action-related” costs, 2) benefits managementcosts, 3) the costs of marketing and selling of in-surance, and 4) the costs of regulation and com-pliance. Health insurers, hospitals, nursinghomes, physicians, employers, and individualsand other consumers are the various actors per-forming each of these activities (see table 2-1).

Thorpe stresses the fact that administrativecosts produce outputs, and that in comparingcosts, one must control for the type and level ofservices produced. In addition, Thorpe points outthat in the case of health insurers in the UnitedStates, not only does administrative spendingvary across insurers, the insurance product itselfdiffers among plans, making straight comparisonsof their administrative costs meaningless. Hence,it is fallacious to conclude that the health plan orthe country spending the most on administrationmust be the most wasteful.

Because Thorpe developed his classification todescribe the U.S. health care system, its useful-ness in comparing administrative costs acrosscountries is limited. In one critique, Hahn sug-

function/component

Transaction -related

Benefitsmanagement

Selling andmarketing

Regulatory/

compliance

Health Nursing Consumers/insurance Hospitals homes Physicians Firms individuals—

Claims processing Admiting, Admiting, Billing Tracking em- Submitting claimsbilling billing ployee hires/ter-

minations

Statistical analysis, Management Management Management Internal analyses Tracking ex-quality assurance, information information Information penses eliglbleplan design systems systems systems for reimburse-

ment

Underwriting, risk Strategic Strategic Advertising Flexible benefit Search costspremiums, adver- planning, planning programstising advertising

Premium taxes, re- Waste Discharge Licensing Filing summary Mandated benefitserve requirements management planning requirements plan descriptions, laws

COBRA obliga-tionsa

a COBRA IS the Consolidated Omnibus Budget Reconciliation Act of 1985, which Includes provisions for continuation of coverage when an employeeleaves a firm

SOURCE K E Thorpe, " Inside the Black Box of Administrative Costs, Health Affairs 11:2 (summer 1992) 41-55

Chapter 2 Defining Administrative Costs 19

gests two modifications of Thorpe’s scheme to-make it applicable to other countries (1 8). Hewould add a fifth administrative function called"oversight” that includes services associated withcalculating and setting global budgets and rate in-creases, evaluating capital expenditures, and ne-gotiating rates with providers. For example, hepoints to Canada and Germany as countries inwhich marketing to attract patients is relatively in-significant since all patients have some coverage.In its place is a bargaining component in whichphysicians or the associations to which they be-long negotiate with the government or insurers fortheir fees.

Hahn also suggests supplementing Thorpe’sscheme with a consideration of the differences incountries’ “production functions” for medical ser-vices. For example, one country may use clinicalstaff, such as physicians and nurses, to perform agiven administrative function while another coun-try may use clerical staff instead. Furthermore.even if two countries use the same type of staff andtechnology to perform a given administrativefunction, the salaries and prices of other inputs toproducing that function may differ between thetwo countries’ leading to different levels of totaladministrative costs. Hence, a true comparisonmust take account of differences in both inputprices and means of carrying out administrativefunctions.

AN ENUMERATION OF ADMINISTRATIVEACTIVITIESAlthough Thorpe’s scheme may be used to con-ceptualize different types of administrative activi -ties in health care, it is still not detailed enough toserve as a protocol for accurately and comprehen-sively measuring the amount of administration ina nation’s health care system. Recent work byGlaser, commissioned by OTA as part of this proj-ect ( 15) and based. in part on earlier research(1 3,1 4). would be the first step in a bottom-up ap-proach to actually comparing the magnitude of ad-ministrative expenditures among nations.

Glaser distinguishes his definitions from at-tempts to group administrative activities accord-ing to their functions (e.g.. transaction-related

costs, regulatory compliance, coordination). Heargues that while the classifications are useful inaggregating and analyzing administrative data,

his definitions are designed to help researcherscollect original data at the grass-roots level. Glas-er does not measure the outputs of administrativeactivities—i.e., the extent to which such activitiesaccomplish their goals: attempts only to providean exhaustive enumeration of the inputs of admin-istration.

Differences in the organization and financingof health care imply that some of the activitiesidentified by Glaser will not exist in every countryand that the relative magnitude of other adminis-trative activities will also vary. For each activityidentified, Glaser suggests that researchers collectdata on the total number of full-time equivalentemployees (FTEs) and total expenditures devotedto that activity. However. difficulties in gatheringdata, discussed later in this background paper,may inhibit researchers’ ability to measure andcompare the administrative apparatus of healthcare across national borders.

I Specific Administrative ActivitiesTable 2-21 lists all of the activities related to heathcare that Glaser identities as administrative in na-ture, primarily classified according to the orga-nizations in which they occur. Unless otherwisenoted, none of the substance of the work in theseorganizations counts as administrative-only theexpenditures for activities necessary to supportthat work. In legislatures and other governmentagencies responsible for health policy. resourcesexpended in making policy decisions would notbe considered administrative. The major excep-tions to this generalization are:●

■

Ministries and other public agencies that im-plement health policies (table 2-2, item 3). Tothe extent that such agencies are devoted tohealth, the entire budget can be counted as ad-ministrative except for expenditures for directclinical and public health services and policy-making.Insurers who pay providers (table 2-2, item9). All of their activities, except for the value of

Organization

1 Public and private organizations that collect vital statisticshealth-related lifestyles, health care financing data, health per-sonnel, and related Information for private organizations, thepublic, and policy makers

2. Legislatures and other organizations that make healthpolicy—prorated share of total administrative costs devoted tohealth

3. Ministries and other public agencies that Implement health poli-cies (Does not Include government agencies to reimburse pro-viders for health care sevices) To the extent agency IS devotedto health, entire staffing and budget minus expenditures for di-rect clinical services, direct public health work, and policy mak-ing

4 Organizations that deliver health care (hospitals, nursing homes,community health centers, home health care agencies, etc.). Forclinicians within such organizations, Includes prorated share oftheir time devoted to administrative functions

0

Organization

5 Individuals who provide care doctors, dentists, midwives, self-employed home visitors, dispensers of alternate medicine, etc.

6 Associations of providers national, provincial, and local offices

7 Organizations that supply health care providers with pharma-ceuticals, equipment, construction, and other materials

8 Government agencies that pay all or some providers Suchagencies can be national, provincial, local, a special fund thatdistributes government grants, or two or more of these together

Administrative activities

Organizational management (as distinct from clinical direction), internal financial work,clerical work communications, regulatory compliance, acquisition distribution andstorage of resources for the facility and clinical operations, personnel management,infrastructural operations

Calculating bills for patient care, billing payers, collections

Medical records the work of clinicians and office staff, transmitting them to other provid-ers, payers, utilization review monitors, etc.

Communication with Iiability Insurers

Litigation of disputes

Management, Internal financial work, clerical work

Communication with payers in negotiation over reimbursement and work rules, commu-nications with regulators, and communications with members explaining reimbursement,regulations, work rules, and clinical Innovations publications and public relations

Organizational management, internal financial work, clerical work, personnel manage-ment

Communications with health care providers and others, marketing and public relations

Negotiating orders, calculating bills, collections

Record-keeping required by price and quality regulators, communications with regula-tors

Communications with insurers, Iitigation

Management of operations, financing, and personnel in the several public agencies thatwrite budgets, process grants, and pay providers Shares attributed to health adminis-tration must be prorated, since some of these agencies deal with sectors outside health

Communications within government--for example, between the Cabinet and the legisla-ture, between the Ministry of Health and the Ministry of Budget--over past costs andfuture needs

Management of the flow of money from tax collectors to the payment agencies

Communications between the payment agencies and the providers Making the pay-ments themselves Collect Ion and audits about costs and performance

Reports to the Ministry's and the paying agencies superiors in government concerninghow the money was spent Reports to the legislature Preparation for special audits

Work of the auditing agency inl health

(Continued)

Organization

9 Insurers who pay some or all providers Payers can be publiccorporations, mutual aid associations, union-affiliated funds,mutual Insurance companies, or for-profit Insurance companiesNearly everything they do (minus the value of paid claims)constitutes administration.

Administrative activities

Organizational management, Internal financial work, clerical work, personnel manage-ment

Communications with subscribers and their payers, marketing, underwriting, negotiatingand writing contracts.

Communicating with regulators who set rules for paying providers

Negotiating with providers and provider associations over practice and reimbursementrules

Receiving, reviewing, and paying claims Utilization review

Communicating with regulatory agencies that review each insurer’s financial accounts,

Reports to government and to associations of insurers concerning the agency’s share ofhealth work and health finance Aggregation of these reports by government and theassociations of insurers Publication

Administrative activities imposed on outside organizations (such as the subscriber’semployer or trade union) in the administration of enrollments and disenrollments, admln-istration of benefits and claims, payment of providers

10. Organizations that conduct research on the organization, man-agement, and financing of the health care system All such workwithin these organizations may be counted in a county’s admin-istrative costs

11. Organizations that provide education about the organization,operation, and financing of the country’s health care system

12. Organizations that conduct management consulting in the healthcare sector

University and specialty-school training of managers, finance officers, and clerks

In-house training

Conferences and workshops

SOURCE Off Ice of Technology Assessment 1994 Based on W. A. Glaser. "Administration in Health Care A Plan for Cross-National Comparisons contractor paper prepared for the Office ofTechnology Assessment, revised edition, 1993

Chapter 2 Defining Administrative Costs 113

claims paid to providers, can be considered ad-ministrative.

■ Organizations that provide education, con-duct research, or consult on health caremanagement, organization, and financing(table 2-2, items 10-12). All such work in theseorganizations can be counted as administrative.

According to Glaser’s scheme, specific expen-ditures in some organizations must be prorated. Inthe case of government agencies and other orga-nizations that do some work outside the healthsector, the value of their administrative expendi-tures must be prorated by the proportion of theireffort devoted to health. For example, in theUnited States, the Department of Health and Hu-man Services (DHHS) has responsibility for So-cial Security and other programs that are not di-rectly part of the health care system. One wouldnot attribute the administration of such programsto health care. In the case of individuals who pro-vide direct health care services, one would want tocount on] y that portion of their time devoted to ad-ministrative functions, and not time spent on clin-ical activities.

This distinction between clinical and adminis-trative activities suggests at least one ambiguitynot addressed by Glaser. He identifies all work byhealth care providers related to medical record-keeping as administrative in nature, includingtime spent by clinicians in preparing these re-cords. However, since accurate medical records

are part of the way in which physicians and othersensure that they provide appropriate care for pa-tients, one could argue that the preparation ofthese records (at least the parts related (o patientcare) is actually a clinical, not administrative, ac-tivity.4

Glaser’s scheme also draws a distinction be-tween government payment and insurance pay-ment. A line agency of government makes pay-ments to providers from its general budget and taxrevenues collected for all purposes, thus makingthe administrative burden of paying providers aprorated share of all government financial admin-istration. Insurance payment, on the other hand, ismade by autonomous public agencies or corpora-tions, nonprofit carriers, for-profit insurance com-panies, or self-insuring third parties (e.g., employ-ers) from earmarked sources such as subscriberpremiums or social security taxes. Using this dis-tinction, Canadian health finance is governmentpayment, while the United States finances privatehealth insurance and Medicare through insurancepayment.

As described in chapter 3, Glaser has appliedhis definitions to the health care systems of fournations, making qualitative estimates of the ad-ministrative costs associated with each. However,as mentioned at the outset, the real purpose of hisenumeration is to serve as a protocol for a bottom-up measurement of administrative costs. No re-searcher has yet engaged in this endeavor.

4 In some instances it may lx diftlcult to distinguish between medical records kept for patient care and those used for truly administrati~ epurpo~es. For example, providers can record diagnostic information both to facilitate proper patient care and to allow insurance reimbursement.

MeasuringAdministration

0 nce one has defined the scope of administrative activities,one must also find data with which to measure the magni-tude of each activity identified. The data most often usedcome from accounting and present significant difficulties

for measuring the true economic costs of administrate ion. The eco-nomic costs of administration refer to the incremental value of re-sources used to produce an administrative function as measuredaccording to the next most valuable alternative use of those re-sources (38). The most common problem with accounting data isthat they do not always fully allocate fixed costs to appropriateadministrative activities, leading to an underestimate of adminis-trative costs. Thorpe offers several examples from the UnitedStates:

■ Medicare, a federal government program that provides healthinsurance to elderly and disabled individuals, has very low ad-ministrative costs relative to private insurance. However.Medicare contracts with private insurance firms to administerthe program. Because these private insurers already have theinfrastructure in place to process claims and perform other ser-vices, the additional cost of administering Medicare is mini-mal, and official estimates of Medicare administrative costsdo not include a prorated portion of the cost of acquiring theinsurer’s administrative infrastructure.

● A firm that sells insurance policies for health and other typesof insurance such as life and property may not include an ap-propriately prorated portion of its chief executive officer’s(CEO’s) salary as an administrative expense of its health insur-ance business.

● Hospitals may not necessarily prorate their data-processingcosts appropriately among billing and strategic planning/con-

3

L I

I 15


trol (administrative functions), and clinical re-search recordkeeping.

Another issue of particular importance in in-ternational comparisons is the accuracy and reli-ability of data collected for comparison. Differ-ences in accounting standards, data collectionmethods, and language can create differences bothwithin and across countries or over time. Thesedifferences, which must be understood to interpretthe data appropriately, may not be adequately doc-umented.

These limitations in using data gathered with-out close attention to the intended purpose ofcomparing administrative costs across countriesleads Glaser to advocate bottom-up, primary datacollection (15). Doing so would entail enormousexpense, time, and logistical difficulties. (An ob-vious question is whether it is worth doing). Al-most all work measuring and comparing adminis-trative expenses of health care within and acrossnations has used data already available for someother purpose.

ESTIMATES OF ADMINISTRATIVE COSTSAND INTERNATIONAL COMPARISONS

| Qualitative Estimates of AdministrativeCosts in Four Countries

Glaser describes the administrative structures offour countries: Canada, England. Germany, andthe United States (15). Although Glaser’s purposewas not to gather any data with which to measurethe magnitude of each administrative activity out-lined, these brief qualitative analyses:

help to illustrate the relationship between theoverall structure of a country’s health care sys-tem and the expected types and magnitude of itsadministrative costs,suggest reasonable hypotheses about howcountries compare in the relative magnitude ofdifferent administrative activities, andhelp to serve as a roadmap for future datacollection efforts.

United StatesThe U.S. health care system has multiple publicand private payers for health care, each with itsown rules, procedures, and administrative appara-tus. Public programs pay for health care for specif-ic segments of the population: elderly, disabled,and indigent citizens; some veterans; and activemilitary personnel and their families. A large por-tion of private insurance is administered throughthe workplace under contracts with private insur-ance firms or self-insured employers. Most pro-viders are autonomous and must interact withmultiple payers. However, a growing number ofpractitioners are employed by capitated healthinsurance plans or are part of one or more net-works of providers associated with a third-partypayer that establishes various cost containmentmeasures.

Glaser proposes that the United States signifi-cantly exceeds the other three countries examinedin administrative expenses. In general, his critiqueof the American system rests on its relative com-plexity (15). The existence of multiple, decentral-ized payers whose coverage guidelines and reim-bursement procedures must be understood byphysicians’ offices, hospitals, and other providerorganizations results in a substantial admnistra-tive burden. In addition, he emphasizes the re-sources required to study the health care system,the specialized training of individuals chargedwith administering it, and consultants employedby providers and other health care organizations tomaximize their revenue.

CanadaThe Canadian health care system is characterizedby full government funding of basic health caredecentralized to the provincial level. Hospitals,physicians, and other providers are autonomous,but they follow provincial standards for financialaccounting. Hospitals and physicians are repre-sented by provider associations. Hospitals operateunder prospective budgets, while physicians billprovincial public corporations for fee-for-servicereimbursement. Private health insurance is

Chapter 3 Measuring Administration 117

Hospitals in Canada, such as Montreal General Hospitalpictured above operate under prospective budgeting whichhelps minimize their admlnistrative expenses

minimal and limited to services and amenities notcovered by the provincial health plans.

Glaser suggests that administrative expensesfall mainly to the provincial agencies in charge ofimplementing the health system, the providers,and their associations. Some administrative acti-vities found in the United States do not occur orare found in relatively small amounts in the Cana-dian system. For example, the costs associatedwith marketing and underwriting insurance arelimited to the small market for private insurance.Employer costs associated with finding an insur-ance firm to provide primary coverage for em-ployees do not exist. Glaser also proposes thatmanagement consulting is largely limited to theuse of computer methodology because of the rela-tive simplicity of the health care system(compared with the United States) and the avail-ability of hospital management manuals devel-oped directly by the hospital associations.

Hospital billing of patients is limited to ameni-ties not covered by the provincial system. Physi-cians’ offices bill provincial public corporationsfor reimbursement, but standardized reimburse-ment rules and electronic claims-filing may helpto minimize these administrative expenses. Gov-ernment incurs the administrative costs of setting

standards, budgeting, revenue collection, dis-bursement of funds, capital planning, negoti-ations with provider organizations, and oversight.Provider associations have the administrative ex-penses associated with representing the interestsof their members at the provincial and nationallevels and in the courts. including the preparationof data and analyses to support their efforts.

England1

The National Health Service (NHS) owns andmanages most hospitals, employs specialist phy-sicians, and contracts with general practitioners.The NHS allocates its budget to 200 DistrictHealth Authorities (DHA). Family Practice Com-mittees (FPCs) contract with physicians and den-tists; they reimburse physicians mainly on a capi-tated basis and dentists by fee-for-service.

Glaser suggests that of the four countries he de-scribes, England has traditionally been adminis-tratively simplest. Under this system, the bulk ofadministrative expenses fall to the NHS and its lo-cal components. These activities include budget-ing, provider payment, preparation of expenditurereports, tracking patients. labor relations. and re-imbursement. The traditional reliance on capi-tated payments to reimburse for a large portionalso contributes to relative administrative sim-plicity. Unions and other associations of providershave a significant role in negotiating on behalf oftheir members, thus requiring their own adminis-trative staffs.

Recent innovations, however, may increasesomewhat the resources needed to administersome parts of the English health system. Somehospitals have become autonomous. leading togrowing local variation in administrative proce-dures. These hospitals also face the cost of mar-keting to patients, developing clinical emphases,setting prices, and balancing a budget. Because asmall number of hospitals and all nursing homesare private, they face these same administrativeexpenses. Some general practitioners have be-

) Among the other countrie~ of the Uruted Kingdom. Wales has a sy stern alrnmt identicd to that of England. Scotland and Nmthmm Irelandalw htivc similar hciilth care i) $terns although with greater wmxmmjf,


come “fund-holders” for their patients; they re-ceive increased cavitation payments to cover pa-tients’ tests, pharmaceuticals, and specialist andhospital care and must track patients’ utilizationand pay other providers. General practitioners arealso now receiving some reimbursements on a fee-for-service basis, thus requiring them to bill theirFPCs. Dentists require the office staff to seek ap-proval from the FPC for extensive procedures andto seek fee-for-service reimbursement for all ser-vices.

Glaser notes that although England performs asubstantial amount of health services and healtheconomics research in government, universities,and other organizations (particularly concerningpotential or enacted reforms), the country hastraditionally relied only minimally on indepen-dent management consultants or specially trainedhealth care administrators. However, he suggeststhat the use of such specialists is on the increasewith the increase in autonomy afforded providersand local jurisdictions.

GermanyLargely administered on a provincial level, theGerman health care system is characterized bymultiple payers called sickness funds, financedthrough payroll deductions. Hospitals can be for-profit, nonprofit, or public. The main role of gov-ernment (at both the national and provincial lev-els) is to enact overall guidelines for the system,monitor its operation, provide some financing.and settle disputes. All providers belong to re-gional associations that negotiate payment levelswith associations of sickness funds. The providerassociations also reimburse their members withthe money given by the funds for the care they pro-vide.

According to Glaser’s analysis, most adminis-trative costs in Germany are found within the sick-ness funds, provider associations, and physicians’offices. Hospitals are autonomous but operate on aprospective budget and, according to Glaser,maintain relatively few administrative staff. Thegovernment’s role is also limited. It makes, over-sees, and reforms the rules of the system, operates

teaching and municipal hospitals and local publichealth services, licenses hospitals, and providesgrants for capital improvements to hospitals.

Sickness funds, like insurance companies inthe United States, must have the administrativeapparatus to calculate and collect premiums. Theyalso collect employee contributions for the nation-al social security pension system. Employees payboth contributions by payroll deduction. In addi-tion, the funds bear the administrative costsassociated with provider negotiations and com-pliance with provincial and national oversight.Recent innovations to allow patients greater free-dom in the sickness fund they join will likelycreate marketing costs for the funds. In addition,the funds have had to undertake the provision ofcoverage in the former German Democratic Re-public.

The physician associations (known as the Kas-senartzliche Vereinigung or KV) also must sup-port reimbursement negotiations, as well as track,process, and pay claims made by their membersand reduce physicians’ fees if necessary to balancetheir budgets. Physicians and dentists must main-tain office staffs to track services provided to pa-tients and submit claims to the KV for reimburse-ment. Because German physicians perform manyprocedures in their offices that in other countriestake place in hospitals or clinics, some requireadditional administrative effort to acquire neces-sary equipment and supplies.

In summary, Glaser’s analysis suggests a fewgeneralizations:

■ Any organization with health care responsibili-ties will incur some administrative costs for itspersonnel functions, internal financing, budg-eting and accounting, and facility overhead.

■ Some health functions occur in similar fashionin all countries and are unlikely to change ordisappear through reform of health care financ-ing or organization. Prime among these func-tions is the collection, analysis, and dissemina-tion of vital statistics and, to a lesser extent,morbidity data. The comparability of these dataacross countries may vary significantly (46),but one would expect the relative magnitude of


the administrative activities associated withtheir collection to be roughly similar. However,true comparisons of this form of administrativeexpense would require actual measurement.

● The relative magnitude of administrative ex-pense associated with any organization withhealth care responsibilities appears to approxi-mate the organization’s role in the health caresystem. Larger responsibilities usually requirelarger organizations, which usually requiremore administration.

■ A number of countries have adopted variouspromarket reforms during recent years in theirhealth care systems in which providers, payers,and consumers have greater autonomy in carry-ing out their obligations. These tend to leadto greater decentralization of the health caresystem and for the most part would be expectedto increase administrative burdens at themargin.

| Quantitative Estimates from theOrganisation for EconomicCo-operation and Development

The Organisation for Economic Co-operation andDevelopment (OECD) has undertaken the only at-tempt to collect data on health administrationfrom many countries over time. However, the use-fulness of these data for comparing the adminis-trative burden associated with different healthcare systems is limited.

The OECD, comprising the most industrializedcountries of the world, publishes data on healthexpenditures and outcomes gathered from itsmember nations (27,28). Health expenditure datarequested from each country are based on the sys-tem of national health accounts (NHA) main-tained for the United States by the Health Care Fi-nancing Administration (HCFA).

The U.S. NHA definition of health administra-tion employed by the OECD is significantly nar-rower than any of those definitions of administra-tion presented above. It refers only to theadministration of public and private insurance,

leaving out the administrative costs of hospitalsand other health care providers and the costs intime or other resources borne by consumers in ob-taining insurance, health care services. or reim-bursement. It also does not include the cost of pub-lic and private health services research or the shareof administrative costs for general governmentaloperations or tax collection devoted to health.

This limited definition may be more importantin some countries than in others. For example,countries like the United States, with a large pri-vate health insurance system and multiple payers.would be expected to realize higher administra-tive expenses for consumers and providers thanwould countries with single payers, relativelycomprehensive benefits, and little out-of-pocketexpenses for consumers. Health service providersin the United States would 1ikely require moretime and resources to understand the system andits benefits and to receive reimbursement thantheir counterparts in countries with a single payer.Hence, the OECD’s underestimation of costs inthe United States may be greater than in countrieswith a small private insurance market and a small-er number of payers.

In addition to starting with a narrow definitionof administration, not every OECD country hasprovided data on health administration, and thecomparability y of data from those countries that doreport varies. Although the OECD and its membercountries have attempted to refine the comparabil-ity of international health accounting data. to datethey have worked with categories of health expen-ditures larger than administration. Administrate ionhas received less attention, in large part. because itrepresents a relatively small portion of most coun-tries’ reported expenditures (31). Figures 3-1 and3-2 present estimates of health administrationoutlays for recent years standardized as a percent-age of total recorded health expenditures in eachcountry.

Poullier’s 1992 analysis of the OECD data onadministrative costs does not provide a compre-hensive explanation of each data point in the


GermanyAustralia Canada France

Country

Netherlands Sweden b United States

a Based on 1989 datab Estimates by J P Poullier, OECD

SOURCE Off Ice of Technology Assessment, 1994 Based on data from J. P. Poullier, “Admistrative Costs on Selected Industrialized Countries, ’Health Care Financing Review 13 (summer 1992) 4167-172

Australia Canada France Germany Italy Netherlands Sweden United United

Country Kingdom States

a Based on 1989 datab Based on 1987 data

SOURCE Off Ice of Technology Assessment, 1994 Based on data from J P Po~llller, “Admmlstratwe Costs on Selected Industrlallzed CountriesHealfh Care Financing Review 13summer 1992)4 167-172

Chapter 3 Measuring Administration | s21

OECD series on administrative costs.2 However,Poullier is able to point out some of the importantlimitations in interpreting the data. The major is-sues concern public sector expenditures.

Even with the limited definition of administra-tion employed by the OECD and countries’ vary-ing (and in some cases, unknown) ability to pro-vide data according to the OECD’S guidelines,Poullier does make some generalizations:

● OECD countries appear to devote between 1and 7 percent of their health expenditures to ad-ministration. Poullier concludes that this rangeis too large to be attributable only to the vaga-

ries of data described above (although the ra-tionale for this conclusion is not made explicit).

■ Those countries that have multiple, segmentedsources of health insurance tend to spend ahigher percentage of their health monies on ad-ministration. These countries include theUnited States, Germany, and the Netherlands(see figures 3-1 and 3-2).

. Time trends in administrative costs tend to re-flect changes in a nation’s health care system.Poullier contends that, all else being equal, therelative share of health expenditures devoted toadministration will tend to decrease over time;as the number and value of health services goup, the per-unit transaction costs decrease dueto economies of scale. Technological changesincluding standardization of claim forms andprocedures and computerization of existing ad-ministrative activities can further reduce per-unit administrative costs. Sweden and Austra-lia appear to have followed this decreasingtrend during the 1980s for both public and pri-vate expenditures, as have Canada, Sweden,the United Kingdom, and the United States fortheir public sector programs. Poullier indicatesthat France would have also demonstrated thistrend if its data were more representative of itsentire health care system. The increase in rela-tive resources devoted to administration inFrance is the result of added insurance benefits,increases in patient coinsurance payments, andthe imposition of cost containment measures,all of which work against the general tendencyfor administrative burdens to lessen overt time.s

Because expressing administrative costs as apercentage of total health expenditures can masksignificant differences between countries in theirspending on health, Poullier also presents per cap-ita estimates of administrative health cxpendi-

2 OEC’[) has not J et had the rcwiirce~ to in~ e~tigate in detail the extent to w hich each country’s admini~tra[iy e cki(a matchci or dI\ crgm fromthe CIetin]tlt)n OECD has :i\hcd thcm to emplo> (~()).

] In tact, Poiillier ~ugge$ts that, all else being equal, added new benefits, increased patient cost-sharing, and adoption of other c~l~t c(mtain -mcnt nw:i\urc\ R i II rcsul t in lnc’rciiscd p:iperwork and mon itoring—i .e., new administrative costs.

22 | International Comparisons of Administrative Costs in Health Care

150

120

90

60

30

0Australia Canada France Germany Italy e

Country

Netherlands Sweden Uni tede Uni tedKingdom States

a All figures in GDP purchasing power parity U S dollarsb Based on 1989 datac Based on 1987 datad Estimates by J P Poullier OECDe Estimates of health expenditures per capita for Itality and the United Kingdom mlsslng from the OECD database for 1990

SOURCE Off Ice of Technology Assessment, 1994 Based on data from J P Pouliler, Administrative Costs on Selected Industrlailzed Countries ‘Health Care Financing Review 13(summer 1992)4 167-172

turcs in Gross Domestic Product (GDP) purchas-ing power parity (PPP) U.S. dollars (figure 3-3aand 3-3b, above).4 This comparison reinforces thefinding that the United States, Germany, and theNetherlands spend more on administration thanmost of the other countries. In addition, the UnitedStates shows a major discrepancy between publicand private expenditures for administration.There are at least two potential, nonmutually ex-clusive reasons for this discrepancy:

The cost of administering public sector pro-grams is actually less than the cost of adminis-tering private insurance programs.The data do not capture all costs of public sectorprograms. In particular, the federal government

contracts with private insurance companies toadminister Medicare. Because these firms al-ready have much of the infrastructure in placeto carry out their Medicare functions, they onlyreport the added cost of administering Medi-care claims, not the fully allocated cost of thatinfrastructure.The OECD data thus provide some very gener-

al insights into resources devoted to administeringsome countries’ health care systems and somechanges in administrative costs over time. How-ever, use of these data are limited and reflect anal”-row definition of administrative costs whencompared with fuller enumerations of administra-tive costs such as that of Glaser, summarized

~ GDP pur~htisin~ ~Wer Pari[les compare tie Cost of purchasing a precise set of goods across countries; strict currency conversion rates can

obwur-c d] ffcrences in the relative prices of different items between two countries.


Australia Canada France Germany Italy Netherlands

Country

Sweden UnitedKingdom

UnitedStates

a All figures in GDP purchasing power parity U S dollars

SOURCE Office of Technology Assessment 1994 Based on data from J P Poullier Admmistrative Costs on Selected Industriailzed Countries

to the United States, and estimate the cost of in-creased coverage and utilization that a Canadian-style system would bring about. Such researchis driven largely by the availability of data gath-ered for other purposes, rather than beginningwith a detailed typology like that of Glaser andthen attempting to gather new data to fit the idealcategories.

This section focuses on the major attempts tocompare administrative costs in the current U.S.health care system with the Canadian system orwith a hypothetical Canadian system implement-ed in the United States. Some of the studies re-viewed attempt to predict health care costs under areformed, Canadian-style health plan for theUnited States, including estimates of the costsassociated with extending coverage to the unin-sured, expanding insurance benfits, and in-creased utilization of services due to the elimina-


(ion of patient deductibles and copayments.s

Assumptions about the effects of a change in theU.S. system are not critiqued in this paper, whichfocuses only on the assumptions and methodsused to derive administrative costs.

Methods and ResultsThe most thorough comparisons of U.S. and Ca-nadian health care administration are contained inwork by Himmelstein and Woolhandler, by theU.S. General Accounting Office (GAO), and bySheils and his colleagues at Lewin/VHI, a healthpolicy consulting firm. Several other authors haveeither critiqued these approaches or commentedon the role of administration in explaining differ-ences in health care spending between the twocountries. Table 3-1 summarizes the methods andestimates of each of the major cormparisons.

Estimates by Himmelstein, Wool handler, andColleaguesHimmelstein and Woolhandler entered this area ofinquiry with a 1986 comparison of administrativecosts in the current U.S. system and under a Cana-dian-style system (20). Their approach, which hasserved as the basis for subsequent comparisons bythese and other authors, proceeds according thislogic:

m

●

■

Divide the health care system among compo-nent sectors: health insurance organizations,physicians, hospitals, and nursing homes.For each, estimate the percentage of total ex-penditures attributable to administration in theUnited States and in Canada using variousavailable data.Estimate potential gross administrative savingsof adopting a Canadian-style system in theUnited States by assuming that the reformedAmerican system would devote the same per-centages of spending to administration as doesthe Canadian system.

Himmelstein and Woolhandler chose 1983 asthe year for their comparison and then estimatedadministrative costs in each of the four major sec-tors of health care. For private health insurance inthe United States, they measured administrativecosts as the difference between premiums col-lected and benefits paid, using the national healthexpenditure accounting data collected by HCFA.Hence, their implicit definition of administrationincludes items such as taxes paid by insurancefirms and profits. However, this definition ex-cludes insurers’ return on the investment of thepremiums they collect. They used the same HCFAdata for estimates of the administrative costs ofrunning Medicare, Medicaid, and other publicprograms.

For physicians, Himmelstein and Woolhandlerrelied on data collected annually by the AmericanMedical Association (AMA) on the socioeco-nomic characteristics of a random sample of allnonfederal, patient care physicians practicingin the United States (excluding trainees). Theydefined administration for physicians as all oftheir professional expenses—a broad categorythat includes items such as malpractice insurancepremiums.

For hospitals and nursing homes, no nationaldatabase routinely estimates administrative costs.Because some individual states do make such esti-mates, Himmelstein and Wool handler drew on re-ports from the California Health Facilities Com-mission (CHFC), which stated that in 1983 forhospitals and nursing homes, 18.3 and 14.4 per-cent of total costs, respectively, went for adminis-tration. As evidence of the national representa-tiveness of the California data, the authors notethat Florida and Texas report similar percentagesand assume that the same proportions applied tothe rest of the country.

In the case of Canada, the authors drew on datacollected by Health and Welfare Canada and Sta-tistics Canada for estimates of the percentage of

5 Ano~er ~eccnt OTA ~epo~ examines the cost imp] lca[ions of major approaches to health care reform considered by the I ~sd Congress.

This analysis includes an examination of the estimated costs of ex panded coverage and utilization under $ingle-pa)er tind other types of systems(47).

StudyHimmels-teln andWoolhan-dler1986a

Year of Categoryestimates of costs

1983 Insurance

Physicians

Hospitals

Nursinghomes

31 1 36% of grossincome

269 8% Of hospitalexpenditures

41 10 % Ofnursing homespending

2 4 9

11 7

30

6 2

152

1.1

26 I International Com

parisons of Adm

inistrative Costs in H

ealth Care

r=

II

N.nI

I

TABLE 3-1 continued: Comparisons of Administrative Costs in the United States and Canada

Estimates

Canada/Canadian System Implemented United States in the United States

$ per capita Total $ per capita Total Year of Category Methodsl Data Percent for administration Percent for administration Difference

Study estimates of costs source administrative administration is billions} administrative administration (!. billions\ ($ billions)

Hlmmels- 983 Insurance National health 15.6 2.5% of pro- 11.1 6.7 teln and expenditure gram costs. Woolhan- data for both dler, United States 19868 and Canada.

Physi- Total self- 45% of gross 31.1 36% of gross 24.9 6.2 clans employed phy- Income. Income.

siclan profes-sional ex-penses. U.S. data from AMA survey Cana-dian data from Health and Welfare Cana-da.

HU::;I-.Jltdi::; CdnaJiar I Jata 18.3% of hos- 26.9 8% of hospital .7 b2 from hospital pital expendi- expenditures. cost reports tures. collected by Health and Welfare Cana-da; American data extrapo-lated from California hos-pital cost re-ports.

Nursing Canadian data 14.4% of nurs- 4.1 10.5% of nurs- 3.0 homes from annual ing home ex- ing home

survey of resi- penditures. spending. dential care fa-cilities by Sta-tistics Canada, American data extrapolated from California hospital cost reports

Study

Woolhan-dler andHimmels-tein,1991 c

Year of Category Methods/ Dataestimates of costs source

1987 Insurance National healthexpendituredata for bothUnited Statesand Canada,all Canadiandollars con-verted to U Sdollars at ex-change of$133 (Cana-dian) = $100(us )

Physi- Method 1cians Physician of-

fice expensesplus physi-clans’ owntime on admmin-istration U Sdata from AMAsurvey, Cana-dian databased on ad-justed tax re-turnsMethod 2Cost of physi-clan office per-sonnel de-voted to ad-ministrationPlUS physi-cians owntime U Sdata from CPS,Canadian ex-trapolatedfrom OntarioMedicalAssociation

Percentadministrative

51 % of totalhealth care ex-penditures

United States

$ per capitafor

administration

106

106-203

Estimates

Canada/Canadian System Implementedin the United States

Total $ per capita Totaladministration Percent for administration Difference

($ billions) administrative administration ($ billions) ($ billions)

1. 2% of total 17 ---

health carespending

41-80


I

I I

I I

I

I

I


I

I

I

30 I International Com

parisons of Adm

inistrative Costs in H

ealth Care

II

Study

Sheils and Young. 1992g

Year of estimates

99-

Category of costs

Insurance

Physicians

Methods! Data source

U.S National health expenditure data; average administrative overhead for public and private insurance. 1983-1989. Canadian system in US.: U.S. Medicare administration adjusted for lower utilization by nonelderly and elimination of p3 tient cost -sharing.

U.S.: All nonpatient care costs reported in survey of multispecialty physician offices by Medical Group Management Assoc Canadian system in U.S Interviews with industry experts about how current administrative cost categories would change.

Estimates

United States Canada/Canadian System Implemented

in the United States ----

$ per capita Total $ per capita Total Percent for administration Percent for administration Difference

administrative administration ~lIion!iL administrative administration ($ billions) ($ billions)

13.7% of 38.2 $80 per 15.7 claims for pri- over-55 benefi-vate insur- ciary; $48 per ance. 3.5% of under-55 bene-claims for pub- ficiary. lic insurance.

43.3 32.2 11

(continued)

.—

Difference($ billions)

134

I


hospital, nursing home, and overall insurance pro-gram spending devoted to administration (8, 10.5,and 2.5 percent respectively). For self-employedphysicians, they use an estimate that professionalexpenses in the province of Ontario average 36percent of gross income. Applying these percent-ages to 1983 spending in the United States, theyconclude that a Canadian-style system could havesaved $29.2 billion in administrative costs, anamount equal to 8.2 percent of actual spending.

Himmelstein and Woolhandler also preparedsimilar estimates for Great Britain using data frompublished sources. According to the NHS, centraladministration of the system costs 2.6 percent oftotal expenditures, while hospital administrationwas 5.7 percent of total hospital spending. Be-cause long-term care is more integrated into theNHS system, the authors assumed that the admin-istrative rate for hospitals also applied to nursinghomes. For physicians, they used a published esti-mate of an average of 29 percent of gross incomefor professional expenses. Applying these per-centages to 1983 U.S. health expenditures, the au-thors conclude that a British-style system wouldhave saved $39.3 billion.

Himmelstein and Woolhandler concede thatthey may have underestimated the administrativesavings possible had the United States imple-mented the Canadian or British system prior to1983. In particular, they cite the lower wages paid

to physicians in those two countries as leading to a$25 billion to $30 billion underestimate in poten-tial savings.6

A Second ComparisonIn 1991 Woolhandler and Himmelstein revisitedthe topic of U.S. and Canadian administrative ex-penditures, this time for 1987 (54). In addition tousing more recent data, the authors also refinedtheir methods, especially for estimating the ad-ministrative costs associated with physicians inprivate practice. The units used to compare theUnited States and Canada also differ from those inthe first study. Instead of estimating the savingsthat could be realized if the United States faced thesame percentages of expenditures devoted to ad-ministrative costs that Canada faces, they esti-mated administrative costs in both countries in1987 U.S. dollars per capita (see table 3-1 ).

The authors estimate the cost of providing in-surance in the same manner as before, drawing onHCFA’s national accounting expenditure data forprivate and public insurance and unpublished datafrom Health and Welfare Canada and StatisticsCanada. For hospitals and nursing homes in theUnited States, they again extrapolate from datacollected by the CHFC. However, this time theyprovide details of the specific cost categoriescounted as administrative.7 For Canadian hospi-tals and nursing homes, the administrative esti-

6 Himmel~teln ~d Woolhmd]er d. not provide tie me~ods underlying his estimate. ~ey also suggest hat some nonadministrative SaV-

ings would result, as the imposition of a national health system would decrease ~inancial incendves to provide “excessive medical intervention.superfluous medical services and products, and the duplication of health institutions. . .“ (20), although they provide no quantitatifc c$timatesof these behavioral changes.

7 Included in heir e5timate of admini5tra[ion tie genera] accounting, patient accounting, credit and COlleCtkXh admitting, Other fiscal $er-

vices, hospital administration, public relations, persomel department, auxiliary groups, data processing, communications, purchasing, medicallibrary, medical records, medical staff administration, nursing administration, in-service education, and other administrative services. Excludedare research administration, administration of educational programs, printing and duplicating, depreciation, amortization, leases and rentals,insurance, licenses, taxes, central services and supply, other ancillary services, and unassigned costs.

.-


mates came from unpublished federal data drawnfrom provincial reports, which were verified byexamining data directly from the provinces.8

Rather than relying solely on physicians’ re-ports of their entire practice expenses as a proxyfor their administrative costs, Himmelstein andWoolhandler also attempted to estimate costsbased on the number of personnel devoted to ad-ministration in physician’s offices. They suggestthat the expense method overestimates the differ-ence between Canadian and U.S. administrativecosts, while the personnel method underesti-mates, thus providing a reasonable range aroundthe likely truth.

Professional expense data for the United Statescame from the AMA’s socioeconomic survey ofphysicians practicing in the United States, whileCanadian data came from a sample of physicians’tax returns corrected for distortions in groupspractice reporting. Data on physician office per-sonnel in the United States came from the CurrentPopulation Survey, a representative survey doneannually by the Census Bureau.

Canadian estimates of physicians’ administra-tive expenses were based on a study of physicianoffice staffing patterns in Ontario done in 1977.9They valued each full-time employee at $35,000(U. S.) in both countries and then added the valueof outside billing services in the United States ac-cording to an AMA survey. For both methods and

countries, the authors added in estimates of thevalue of physicians’ own time spent on billing.

When the authors recalculated 1987 adminis-trative costs in a manner exactly comparable totheir 1983 estimates, the numbers show that dur-ing this four-year period administrative costs inthe United States rose from 21.9 to 23.9 percent oftotal health expenditures. while in Canada theydeclined from 13.7 to 11.0 percent.

National Estimates of U.S. Hospital CostsOne of the criticisms leveled against both studiesby Woolhandler and Himmelstein is that they gen-eralize from the experience of California to makenational estimates of hospital administration(2,25). Although they found the California esti-mates to be comparable to seven other states, theauthors did re-estimate hospital administrativecosts for 1990 using national Medicare cost re-ports drawn from 6,400 hospitals that participatedin Medicare that year, close to the universe of allhospitals in the United States (55). ‘() They allo-cated each reported hospital expense category aseither administrative, clinical, both, or neither.The “both” category comprises the cost of thephysical plant and employee benetits. 11

This analysis showed that administration was24.8 percent of national hospital expenditures in1990, with a range of 20.5 to 30.6 percent amongthe states. This estimate is higher than those used

X ~ey, ~~tlmate total hospl[al adnllni~[ra[l~ e costs bv adding together the categories of “other” hospital adminiwtitmn. ad~ crtl~lll~. ~I~wJL’ 1~-

tion-member~hip fees, busine~~ machines, collection fees, postage, auditing and accounting, other nonmedicai profe~~ional fcei, \cr\ ice-bu-reau fees, telephone and telegraph. board members’ indemnity, travel and convention expenses, medical records, ho<pi[al library, and nuriingadministration. Excluded are educational and refearch administration, insurance, interest, printing, stationery and office supplies, materialmanagement, and central supply. For nursing homei, administration constituted only a single category.

In August 1994 Woolhandler and Himrnelstein issued a correction to their 1991 study indicating that an error in their raw data had c;iu~edthem to undere~timatc the cost of hospital nur~ing administration in Canada. The correct data would have raised hospital Canadian pcr c{Ipif~/administration from $50-S58 (Canadian) and the range of total per capita administration from $ I I 7- S 156 (Canadian), to $ 125-S 164 (C”;ina-dian) (56). Because of the late date of this correction, this background paper’s discussion of their work and the associated tables do not incorpo-rate this change.

9 Woolhandler tind Himmelstein report that staffing in the 1977 sun ey appeared to be somewhat higher than informal 1991 mtirnatc~ pro-vided b} the Ontm-io Medical Association.

lo According t. [he American Ho\pital Association, there were 6,720 hospitals in the United Statef In 19X9 ( I ).

1 I me Prownion of Phy \ica] P]ant a[[rlbu[able t. administration was assumed to be the same as the proportion Of all Other co~t~ ~ttribut~tblc

to admini~tration in the ho~pital. For employee benefits, all salaries of employees who administer the benefit~ were a~~umcd to be admmi\tra-tive. All remaining co$ts were allocated between administrative and c1 inical in the same manner as phyfical plant costs.


in the 1983 and 1987 U.S.-Canadian comparisonsbased on data from California hospitals alone(18.3 and 20.2 percent, respectively). The 1990estimate for California only was even higher: 27.7percent.

The authors do not attempt to explain the differ-ence between this and their earlier estimates, stat-ing only that their method of allocating expensesfor physical plant and related capital and interestmay somewhat overestimate administrative costs.If one assumes that no part of these expenses is at-tributable to administration, the overall estimateis reduced to 20.8 percent. Schwartz and Mendel-son (34) have suggested other ways in whichWoolhandler and colleagues’ Medicare estimatesmay overstate the cost of hospital administrationin the United States:

In their Medicare cost reports, hospitals tend toshift expenses from clinical to administrativecategories to increase reimbursement.The authors do not exclude the portion of gen-eral administration attributable to research andeducation in the hospital; they exclude only thedirectly itemized costs for these programs.l2

U.S. General Accounting Office (GAO)EstimatesIn 1991 GAO issued its own analysis of the eco-nomic costs and benefits of implementing a Cana-dian-style system in the United States, includinga comparison of administrative costs in the two

countries (43).1 3 Using data from various sourcesfrom the late 1980s, GAO projected administra-tive cost estimates to 1991 for both countries (seetable 3-1 ). Although GAO followed the same gen-eral procedure of breaking administrative costsdown among its component parts and even usedsome of the same data sources as Woolhandler andHimmelstein, there are significant differences inmethods and results. GAO did not include esti-mates of nursing home administrative costs for ei-ther country. For the United States, GAO:■

■

■

broke physician administrative expenses intothree components using data from the AMA’s1988 socioeconomic survey: proportion ofphysicians’ time spent on insurance (4.4 per-cent), nonphysician payroll ($42,500 per phy-sician), *4 and the cost of contract billing ser-vices (14 percent at a cost of $8 per claim, or$3,224 per physician); 15estimated hospital administrative costs usingdata from the American Hospital Association1988 Monitrend, prepared under contract to theU.S. Prospective Payment Assessment Com-mission (15.4 percent of total hospital ex-penses) (23,48); 16 andused the 1988 HCFA national accounting datafor health expenditures to calculate the propor-tion of insurance expenditures devoted to over-head defined as “administration and the net costof private health insurance” or the differencebetween premiums and benefits paid (5.8 per-

IZ Schwtiz and Mende]s~n also point out tia[ the category of general administration contains expenses such as utilization review, which

might not be able to be eliminated under a Canadian-style system without some decrease in quality or increase in overall costs and, as discussedlater in this background paper, that Himrnelstein and Woolhandler’s approach to comparing costs in the United States and Canada may underes-timate administrative costs inherent in the Canadian system (34). Furthermore, utilization review may be diff]cult to categorize as either anadministrative or clinical expense since it affects both.

13 GAO detailed tie me~~s used in tiis analysis in a separate publication published in 1992 (44).

1A Implicit in GAO’S me~ods is tie assurnp[ion that the whole difference in the nonphysician wage bill between Canada and the United

States is attributable to administration and not other factors such as differentials in wages and intensity of clinical services, This latter factorcould be especially important since nonphysician personnel include nurses and technicians.

15 Data on total num~r of physicians and physician expenditures include physicians employed by HMOS. However, GAO suggests thatthis could not distort their estimates in any significant way since physicians empioyed by HMOS represented only 2 percent of all practicingphysicians (44).

16 Uslng data provided t. propAC, GAO calculated administrative expenses as a proportion of the cost per hospital discharge. 1n this data-

base, administration comprises the categories of general accounting, patient accounts and admitting, medical records, purchasing and stores,and data processing (23,44).

cent), which is the same definition used byWoolhandler and Himmelstein. 17

For Canada, GAO:■

■

■

used unpublished data from the Ontario Medi-cal Association to estimate the nonphysicianwage bill for that province (an average of$28,033 per physician). Because the same dataindicated that physicians spend little time onbilling and insurance, GAO assumed that theyspent 1 percent of their time on these matters.It was also assumed that there are no contractbilling services in Canada and that the experi-ence of Ontario is representative of the entirecountry;used unpublished data from Health and WelfareCanada that administrative costs were 9 percentof total hospital expenditures in 1987; 18 andused a 1987 Canadian national health account-ing data category called “prepayment adminis-tration” as the measure of the administrativecost of providing public and private insurance(1.2 percent of total health expenditures).

GAO concludes from its estimates that a Cana-dian-style system implemented in the UnitedStates in 1991 would lead to $67 billion less in ad-ministrative costs than were spent under the cur-rent system. This difference breaks down to $34billion in insurance overhead, $15 billion in phy-sicians’ administrative costs, and $18 billion inhospital administration.

Comparison by Sheils and YoungIn January 1992 Sheils and Young, analysts at theprivate consulting firm Lewin/ICF,19 releasedtheir own comparison of U.S. and Canadian ad-ministrative costs (36,37). In proposing their anal-


ysis, they offered several critiques of the work byHimmelstein and Woolhandler (36). mostly con-cerning the suggestion that implementation of aCanadian-style system in the United States wouldlower administrative costs. A specific criticismconcerned the accuracy of Himmelstein andWoolhandler’s measurement of administrativecosts in either of the two countries. In particular,Sheils and Young suggest that many indirect costsof running the Canadian provincial health pro-grams, including those associated with facilitiesand equipment, were left out.20

Their other critiques focus on the nature of orpotential explanation for the differences they find.They observe that a significant portion of provid-ers’ administrative costs in the United Stateswould not necessarily change with a new reim-bursement system. These include costs associatedwith malpractice, supplies, security, grounds, andwage differentials. These authors also suggest thathigher administrative costs in the United States re-flect, in part, higher capitalization (i.e., more med-ical equipment and facilities ) and higher Constitu-tional standards for legal due process. whichraises the costs of claims adjudication. Highercapitalization can change only in the longer run,while there is no reason to believe that standardsfor due process would necessarily change at all(37).

Like GAO, Sheils and Young summed the ad-ministrative costs for insurance, physicians, andhospitals to arrive at an overall figure. However,their methods and some of their data (see table3-1 ) vary from those used by either GAO or Wool-handler and Himmelstein. Most significantly.their analysis is not actually a comparison of U.S.and Canadian administrative costs. To correct for

11 ~1~ Categov comprises tie accounting categories of administrative costs, net additions to reserves, rate credits and di~ idends, premiumtaxe~, and profits or losses. Both GAO and Himmelstein et al. calculated the administrative costs of insurance using HCFA data estimtites of thenet co~t of pri~ate health insurance as a percentage of total expenditures on health services and supplies (44,51 ).

1~ GAO attempted [. inClu& exPnSe categories Comparable to those measured for the United Sta[ej: genera) adnllnlstratlOn (minus liabil-

ity infurance, interest payments. and utilities), material management, central supply, medical record~, and hospital library (44).

19 ~1~ f i rm is now known as Lewin-VHI.

20 However, they provide no reference or detail for this, only alternative methods of mea$uring admini~tratlve costs.


the problems they find in the work by Woolhan-dler and Himmelstein, they base their estimatesfor a hypothetical Canadian-style system imple-mented in the United States on assumptions abouthow current U.S. costs would change under a re-formed system. They do not use any data from ac-tual Canadian experience to make their estimates.

For the United States, Sheils and Young calcu-lated insurance overhead using HCFA’s nationalaccounting health expenditure data. They basedtheir extrapolation on the average administrativeoverhead rate for the period 1983 to 1989 to avoidyear-to-year fluctuations, and calculated adminis-trative overhead as a percentage of claims paidseparately for private health insurance (13.7 per-cent) and public programs (3.6 percent).

To estimate administrative costs under a Cana-dian-style system in the United States, Sheils andYoung extrapolated from Medicare administra-tive costs (with some adjustments). They arguethat this approach compensates for characteristicsof the U.S. health care system not found in Canadathat influence administrative costs and are notnecessarily subject to change under a single-payersystem. This approach also corrects for the factthat data on Canadian insurance administrationdoes not include overhead for buildings, equip-ment, fringe benefits, and personnel services(37).21 The authors estimate that total insuranceadministration would be $10.5 billion for the non-elderly population and $2.5 billion for the elderly.To this, they add an estimated $1.6 billion in theadministration of private health insurance and$1.1 billion for public programs that cover ser-vices not included under the national program, for

an estimated total of $15.7 billion in insurance ad-ministration under a Canadian-style system.

To estimate physician costs not directly relatedto patient care, Sheils and Young used data from a1990 survey of multispeciality medical groups bythe Medical Group Management Association thatincluded data on expenditures for different typesof nonclinical activities. To this, they added an es-timate of the value of physicians’ own time spenton insurance issues based on the AMA’s 1988 so-cioeconomic survey data. These methods yield es-timates of $17.4 billion in nonphysician salaries,$6.64 billion in physician time spent on adminis-tration, and $19.54 billion in other administrativecosts for a total of $43.58 billion in 1991.

To estimate hospital administrative costs underthe current U.S. system, which they define as ev-erything except direct patient care, Sheils andYoung drew on the same detailed cost accountingdata collected for California used by Woolhandlerand Himmelstein.22 Summing all nonclinical costcategories and extrapolating to the country as awhole, they estimate hospital administrative costsin 1991 to be $93.9 billion (or 33.3 percent of totalhospital spending), which includes $9.4 billion innet hospital revenues extrapolated from the na-tional net revenue rate reported in 1989 Medicarecost reports.

For hospital and physician administrative costsof a Canadian-style system implemented in theUnited States, Sheils and Young examined eachcategory of administrative costs under the currentsystem, On the basis of interviews with unidenti-fied industry experts, they made assumptionsabout how each category of costs would change

z I l’hcv es[ima[e that while Medicare has administrative costs of $85 peremollee per year, a Canadian-style system would have costs of $80per elderly enrollee and S48 per nonelderly enrollee. These projected differences between the current Medicare program and a Canadian pro-gram would be the net result of the elimination of individual hospital claims, increased utilization due to the lack of copayments, and the fact thatnoneldedy beneficiaries would have lower utilization than do the elderly and disabled beneficiaries of Medicare. They assume utilization re.view programs would remain.

22 shei]~ and Young note that extrapolation from California to the rest of the country maybe problematic because California has a 14-per-cent lower average length-of-stay, a 50-percent higher average cost per day, a 5.5-percent higher staff-to-bed ratio than the nation as a wholeand recent legislation that may have increased administrative costs associated with contracting for negotiated discounts. However, they do no{comment on or attempt to replicate Wool handler and Himmelstein’s analysis that shows hospital administrative costs in California to be compa-rable to those in other states.


under a Canadian-style system. Summing thesecomponents, they estimate that hospital adminis-tration would cost $80.65 billion and physicianadministration $32.23 bill ion.

Summary of Estimated AdministrativeSavingsFor the four major analyses summarized above,table 3-2 presents the estimated impact on admin-istrative costs of implementing a Canadian-stylesystem in the United States. All estimates arein 1991 U.S. dollars. OTA has converted the percapita results from the 1991 Himmelstein andWoolhandler study (54) to total expenditures.

Leaving out the earlier of the two Himmelsteinand Woolhandler studies, the range of potentialsavings is $47 billion to $98 billion. Although thisrange is large, the findings do suggest that. all elsebeing equal, imposition of a Canadian systemcould lead to a reduction in administrative costs.

Other ApproachesOther authors have discussed differences in ad-ministrative costs in the course of comparing theU.S. and Canadian health care systems, but nonehas attempted any quantitative estimates indepen-dent of those discussed above. In their proposalfor health care reform in the United States, thePhysicians for a National Health Program rely onestimates by Himmelstein and Woolhandler (20)as evidence of administrative savings that couldbe realized under a single-payer system (16).Another reform proposal by the Economic andSocial Research Institute with support from theRobert Wood Johnson Foundation uses Himmels-tein and Woolhandler’s 1991 study as the basis forestimating administrative savings from adoptinga Canadian-style system.

Fuchs and his colleagues discuss differences inadministration as part of two studies comparinghealth care costs in the United States and Canada( 10,11). However, they do not attempt to measure

Himmelstein andWoolhandler, Woolhandler and Sheik and Young,

1986 b Himmelstein, 1991C GAO, 1991-92d 1992e—

Year of estimates 1983 1987 1991 1991.

Administrative savings inInsurance 9 26 34 23Physicians 8 19-35f 15 11Hospitals 20 32 18 13Nursing homes 1 5 —9 —9

Total estimated 39 81-98 67 47administrative savings—

a Data from Himmelsteln and Wool han dl er 1986 and Woolhand Ier and H I mmelsfeln 1991 inflated to 1991 U S dollars using the Gross DomesticProduct (GDP) Impllclt Price Deflator

b D U Hlmmelsteln and S Woolhandler ‘Cost Wlfhout Befieflf Admlmstratlve Waste n U S Health Care “ NEJM 311 (7), 441-445 Feb 13, 1986c S Wool handler and D U Hlmmelsteln The Deter orating Admlmstratlve Efflclency of the U S Health Care System, ” NEJM 324( 18)

1253-1258 May 2 1991d u s GAO Canadian Hea/th /nsurance Es?/rnaf/ng Costs arm Savings for the LMed .SYates, U S GAO, #HRD-92-83 April 1992, U S GAO

Canad/an Health hsurance Lessons for the Um?ed States U S GAO, ~HRD-91 -90 J.ne 1991‘J F Shel Is, and G J Young Nat lona Hea t h Spend ng U rider A Single Payer System The Can adlan Approach, ” staff working paper for Lewl n)

ICF Jan 8 1992I The range represents Wool handler ard H rnrnelstelns two met~,ods of estlmatlrg physicians admmlstratlve eXPense5 The text summarizesthese methods n greater detail

9 These studies d d not estimate nursing home adrmnstratve costs

SOURCE Off Ice of Technology Assessment 1994


administrative activities directly, In an analysis ofphysician services, Fuchs and Hahn speculate thathigher administrative costs are a prime source ofthe higher physician fees that they observe in theUnited States. As evidence of higher administra-tive costs, they cite Himmelstein and Woolhan-dler’s 1986 study as well as some of these authors’data sources (10,1 1). Similarly, they suggest high-er administrative costs and intensity of service inthe United States as “the most likely explana-tions” for the higher overall hospital costs but theyoffer no independent evidence to support this ex-planation (32).

Evans and his colleagues also have examinedand commented on differences in health care ex-penditures in the United States and Canada, sug-gesting administration as one of the sources of thehigher expenditures observed in the United States(3,8,9). However, they too do not try to measureadministration directly.

A Debate Over U.S.-Canadian Comparisons

Danzon’s Critique of U.S.-CanadianComparisonsDanzon (6) has offered an economic critique of theentire approach of using existing data to compareadministrative expenditures in different healthcare systems. Her analysis, which has provedcontroversial, goes to the heart of the definitionalissues considered in the first section of this paper.She first suggests that the national accounting datameasuring insurance overhead in the UnitedStates is not comparable to the estimated overheadof Canada’s provincial insurance program. Shesuggests that premium taxes, investors’ return oncapital, and investment income should be re-moved from the American estimates. 23 By her cal-culations, this adjustment would reduce Woolhan -dler and Himmelstein’s estimate of insurance

overhead for 1987 (54) from 11.7 percent of bene-fits to 7.6 percent.24

The more significant part of Danzon’s critiqueis that analyses using accounting data (like thoseof Himmelstein and Woolhandler, GAO, andSheils and Young) ignore important hidden or in-direct costs of administering publicly based healthcare systems like that of Canada. She includesamong the hidden costs of the Canadian system:

■ excessive patient time resulting from physi-cians’ tendencies to compensate for fixed feesby scheduling multiple, short office visits:

● diminished productivity, lost income, and low-er quality of life due to waits caused by ration-ing of hospital services; and

■ “dead-weight loss” in productivity and con-sumption as employers and consumers changetheir behavior to avoid activities that are taxedby the state to finance the health care system inlieu of private insurance premiums.

In addition to unmeasured overhead costs in theCanadian system, Danzon argues, there are un-measured benefits in the administrative apparatusof the U.S. system. She views claims processing, alarge component of administrative expendituresin the United States, as a check against “moralhazard,” or the tendency of consumers to overusehealth care services because they are insuredagainst all or much of their costs. In addition, shesees the diversity of insurance plans as a means ofaccommodating the variety of consumer prefer-ences, although she concedes that employer taxsubsidies for health insurance and the structure ofinsurance regulation in the United States may leadto more options in the current system than is effi-cient.

Although they are not directly related to over-head or administration, Danzon also cites the sub-stantial amount of health-related research and the

23 Danzon ~gues that ~ese components shou]d be removed because premium taxes are a transfer from employers and consumers to stategovernments, not an actual cost; because investment income is a return to insured individuals and groups for the use of the premiums that the)pay in advance; and because it is not clear what cost in a public insurance program would be comparable to the return on capital found in pritatcinsurance.

z~ ~is figure is compared with 0.9 percent for Canada.


diversity of nonphysician medical personnel asadditional benefits of the U.S. system of financinghealth care.

Critiques of Danzon’s AnalysisOther analysts have taken issue with several ofDanzon major points. Schlesinger (35) believesthat Danzon subjects Canada to a double standardby counting patient time from multiple or lengthymedical visits as a cost in Canada, but ignoring pa-tient time lost attempting to understand the detailsof insurance benefits, copayment requirements,and claims forms in the United States, Her argu-ment that Canadian rationing through patientwaiting leads to a lower quality of life is notweighed against the fear many Americans mayhave that they might lose their health insurance.And the “dead-weight loss” associated with tax-based financing in Canada is not balanced againstthe “dead-weight loss” of workers who cannot notmove to optimal jobs for fear of losing health in-surance on a temporary or permanent basis.

Schlesinger also criticizes Danzon for ignor-iin certain costs in the United States:

1.

2.

3. .

4.

the cost of evaluating and deciding among in-surance plans and provider systems,the costs to firms of trying to avoid hiring em-ployees believed likely to use substantial healthcare services,the cost of employee benefits personnel infirms, andthe cost of capital for private insurance over andabove the comparable cost for public programssince private firms must compensate investorsfor risk of bankruptcy.

On the subject of Canadian queues for services,Barer and Evans (3) argue that both the U.S. andCanadian systems ration, and that the Canadian

means of rationing through queues is preferablesince it is based on information (physicians’ judg-ments of medical necessity) rather than on abilityto pay. Woolhandler questions whether there aremedically significant waiting times in Canada atall, noting that there has been little empirical re-search on the subject(53 ). One recent study of ran-domly chosen breast cancer patients in British Co-lumbia (Canada) and Washington State (UnitedStates) actually found 13.4 percent of women inWashington experienced a delay of three monthsor more25 from time of first symptom to diagno-sis, while only 4.6 percent experienced such adelay in British Columbia (a statistically signifi-cant difference).2b

PERSONNEL AS A MEASURE OFADMINISTRATIONA significant component of a country’s health careexpenditures are personnel costs, including indi-viduals charged with carrying out administrativeduties. Through censuses and other population-based surveys, countries gather information ontheir labor forces on a regular basis. Analysis ofthe health care labor force may serve as a usefulproxy for expenditures devoted to administrationand patient care, especially when trying to assessthe relative investment in administration acrosscountries or to assess trends over time.

To investigate the usefulness of this approachand to understand better the health care laborforces of the United States and Canada, OTA com-missioned an analysis of national occupationaldata for these two countries by David Himmels-tein, Steffie Woolhandler, James Lewontin, andDonna Pound at the Center for National HealthProgram Studies, Harvard Medical School (2 1 ).27

~f~ NICCI1lirl [lnle~ frorll \}mp[oITl t. diagn~~i~ for [he over-a]] sample were relatit ely short and similar between the t~~’o regions (*4).

“ Himmclitein and colleaguc~ also m~ estigated occupational trends in the German health cares) stem. HOW ever. because of serious discre-piincim betw ecn Germany and the other tw o countries in defining Y arious occupatiomil categories ( 22), OTA omits the results of their prelimi-nary> :in:ily~ef of German) in thif document.


Summary of Methods28

For each country, Himmelstein and colleaguesgrouped into one of 17 occupational categories allindividuals whose principal place of employment,whether part time or full time, was the office of aphysician or other health practitioner, a hospital,a nursing or personal care facility, or other healthservice facility.

29 Using data on numbers of

employed individuals and hours worked, the au-thors calculated “full-time equivalents” (FTEs)for each job category in total and per capita for the

30 With these data theywhole U.S. population. analyzed trends in the size and composition of thehealth care workforces in each country andcompared the workforces of 1971 and 1986.3 Inaddition to focusing on relative numbers of ad-ministrative personnel in each country, the analy-sis also examines each country’s reliance on tech-nicians and technologists as a possible proxy forthe intensity of services and use of technology inCanada and the United States.

Employment information for the United Statescame from the U.S. Census Bureau Current Pop-ulation Survey (CPS) from 1968 to 1992, an annu-al survey of 60,000 households representative ofthe civilian noninstitutionalized population. Thesurvey records information on occupation andplace of employment and includes about 6,000 in-dividuals working in the health care sector. Dataon health care workers in Canada comes from the1971 and 1986 Canadian censuses; the first ofthese censuses just preceded the full implementa-tion of single-payer health insurance in Canada.

Although Himmelstein and colleagues wereable to identify clearly individuals with health-re-lated occupations (e.g., physicians, nurses, thera-

pists) in nonhealth care workplaces, a major limi-tation of their analysis is that the CPS data do notallow identification of administrative and clericalpersonnel who perform health care-related func-tions in such workplaces. Hence, their data do notinclude personnel in private firms who administerhealth insurance benefits for their employees,leading to underestimates of administrative per-sonnel in the United States, or health care manage-ment consultants who do not work in health careworkplaces.

Results

Health Care Personnel in the United StatesBetween 1968 and 1991, the number of FTEs forall U.S. health care occupations grew from 3.98million to 9.79 million (146 percent), as shown infigure 3-4. However, the number of administrativepersonnel grew much more than the average: man-agers and related personnel from 128,000 to907,000 (608 percent); administrative supportpersonnel except financial from 520,000 to 1.42million ( 183 percent); administrative support, fi-nancial from 70,000 to 269,000 (285 percent); so-cial service from 32,000 to 293,000(818 percent);therapists from 33,000 to 239,000 (606 percent);and technologists and technicians from 230,000 to802,000 (249 percent). The number of FTE clini-cal personnel (physicians and nurses) grew slight-ly less than the average increase, while there waslittle change in food service, laundry, cleaning,and maintenance personnel.

The change over time is also striking whencomparing the composition of the health careworkforce in 1968 and 1991 (figure 3-5). Man-

28 Appendix C gives a comp]e[e, detai]cd description of the methods used by Himmclstcin and ~oilcilgue~.

29 For yews 1968.71, the Cumcnt p~pulatlon Survcv (U.S. Census Bureau I only allows classification into two health care workplaces: hos-

pitals and “other.”

N Himme]stein and colleagues a]so adjusted for t}lc pos~iblc lack of comparability in certain job catcgoriej between the t~ o countries ~d

tested the sensitivities of their results to changes in the Census Bureau’s job classification schemes over time in the United States.

31 Because tie Cument population Suney is a samp]e survey, estimates made for the entire U.S. population using CPS data carry potentit~l

sampling error. These standard errors are taken into account in the 90 percent confidence intervals presented for the U.S. estimates in figure 3-4,and figures 3-6 through 3-11. Because the Canadian census is a 20-percent sample, the random standard errors of estimates from its data arenegligible (5 I ).


SOURCE

Thousands of FTEs

melsteln, S Woolhandler, J P Lewontin, D J Pound “Health Care LaborForce U S , Canada and West Germany, ” contractor paper preparedfor the Office of Technology Assessment Cambridge, MA Center forNational Health Program Studies, Harvard Medical School/The Cam-bridge Hospital Mar 19 1993

agement and administrative support personnelgrew from 18.1 percent of all FTEs in 1968 to 27.1percent in 1991. Nursing personne132 declinedfrom 40.6 percent of FTEs in 1968 to 35.6 percentin 1991. Other declines occurred among physi-cians (10.8 to 7.5 percent of FTEs) and food ser-vice, cleaning, laundry, and maintenance person-nel (14.9 to 8.2 percent). All other clinicalpersonnel combined increased from 10.7 to 14.8percent of all FE health workers.

Comparisons With CanadaIn 1971 the United States employed 22,000 personnel per million population; Canadaemployed 26,565 (see figure 3-6). In terms of thenumber of administrative personnel per capita, thetwo countries were almost identical (see figure

SOURCE D U Himmelstein, S Woolhandler, J P Lewontin D J Pound“Health Care Labor Force U S Canada and West Germany contractpaper prepared for the Off Ice of Technology Assessment CambridgeMA Center for National Health Program Studies, Harvard MedicalSchool/The Cambridge Hospital Mar 19, 1993

3-7). However, between 1971 and 1986 the healthworkforce of the two countries diverged. U.S.health FTEs per million rose 53 percent, whileCanada’s rose 19 percent. resulting in 7 percentmore FTEs per million in the United States thanin Canada (33,666 vs. 31,529) (figure 3-6).

All the U.S. excess in health personnel ascompared to Canada in 1986 is attributable to thegreater numbers of managers and support person-nel in the United States (figure 3-7). In 1986 theUnited States employed 85 percent more healthmanagers per million population than did Canada(2,634 vs. 1,425), 22 percent more nonfinancialadministrative support (4.593 vs. 3,778), and 65percent more financial administrative support(999 VS. 604).

Excluding administrative personnel, the twocountries employed roughly the same number ofFTEs per million in 1986 (25,440 in the United

32 NurS1ng ~rSonne] include ~egiStered ~urSeS (RNs), licensed practical nur~es (LpN\), and nursing lhea]th care aich.


‘ /

SOURCE Officeof Technology Assessment, 1994 Based on Himmels-tein, D U Woolhandler, S Lewontin, J P, Pound, D J , “Health Care La-bor Force U S , Canada, and West Germany” contract paper pre-pared for the Off Ice of Technology Assessment Cambridge, MA Cen-ter for National Health Program Studies Harvard Medical School/TheCambridge Hospital Mar 19, 1993

States vs. 25,722 in Canada). The United Stateshad fewer registered nurses (5,419 vs. 6,948),more licensed practical nurses (1,333 vs. 1,002),and more technologists and technicians (2,423 vs.1,988) (see figures 3-8 and 3-9).33

The divergence in the number of FTE techni-cians and technologists is particularly interesting.While this group grew 37 percent in Canada be-tween 1971 and 1986, the comparable increase inthe United States was 80 percent.

In 1986 Canada employed 18 percent fewerFTE technicians and technologists than did theUnited States. This finding supports other ob-servations that Canada uses less technology in

SOURCE Office of Technology Assessment, 1994 Based on D U Hirn-melstein, S WoolhandIer, J.P. Lewontin, D J Pound, “Health Care LaborForce U S , Canada, and West Germany,” contract paper prepared forthe Office of Technology Assessment Cambridge, MA Center for Na-tional Health Program Studies, Harvard Medical School/The Carn-ridge Hospital Mar 19, 1993

medical care than does the United States (33). Al-ternatively, this finding could bean indication thatCanada regionalizes its technology to a greater ex-tent than the United States—that is, it offers ex-pensive, high-technology services in a limitednumber of regional centers that specialize in theservice or procedure rather than diffusing thembroadly throughout the country (52).

Comparisons of the Labor Force inPractitioners’ OfficesHimmelstein and colleagues also examined thecomposition of the labor force specificallyemployed in practitioners’ offices. Practitioners’

~~ Whl]e the United States had more workers per mi]lion classified as “aides or other health service persomel,” it had fewer in the category“not elsewhere classified” (n.e.c. ), probably reflecting a difference in occupational coding procedures in the two nations. Classifications such as“aides” and “orderlies” appear to be more narrowly defined in Canada than in the United States. In addition, a single Canadian occupationalcode comprises therapists and nursing aides n.e.c. and was assigned to the “therapists” group for the purposes of this analysis (21).

8,000I A ,

I/

6,000

o

U.S. FTE/Million Population

Us. 90%Confidence Interval

A CANADAFTE/Million Population

SOURCE Off Ice of Technology Assessment, 1994 Based on D U Him-melstein, S Woolhandler, J. P. Lewontin, D J Pound, “Health Care LaborForce U S , Canada, and West Germany, ” contract paper prepared forthe Office of Technology Assessment Cambridge, MA Center for Na-tional Health Program Studies, Harvard Medical School/The Cam-bridge Hospital Mar 19, 1993

Over the past 20 years the number of technicians required tooperate high technology diagnostic equipment like the CTscanner pictured above have increased much more in theUnited States than in Canada

1 1 1 1 1 1 1 [

a Confidence Intervals are not calculated before 1978 because theCensus Bureau, which gathers CPS, does not consider the CPS esti-mates of less than a certain magnitude to be precise enough to war-rant calculation of standard errors

SOURCE Office of Technology Assessment, 1994 Based on D U Him-melstein, S Woolhandler, J P Lewontin D J Pound, ‘ Health Care LaborForce U S , Canada, and West Germany, ’ contract paper for the Off Iceof Technology Assessment Cambridge, MA Center for National HealthProgram Studies, Harvard Medical School/The Cambridge HospitalMar 19, 1993

offices in the United States employed about twiceas many FTEs per million population as did thosein Canada in both 1971 (4,325 vs. 2,219) and 1986(6,716 vs. 2,718). However, the value of suchcomparisons is not clear as some employees ofdentists’ offices in Canada are classified under“health services, n.e.c.,” but as working in practi-tioners’ offices in the United States. Disaggregat-ing the 1986 data as reported, striking differencesappear in the composition of office staffs betweenCanada and the United States. In particular, theUnited States has more managers (646 vs. 29),nonfinancial administrativc support workers (1148vs. 816), financial administrative support workers(282 vs. 89). social service personnel (138 vs. 4),other diagnosing professions (954 vs. 32), techni-cians (506 vs. 51 ), and aides (963 vs. 5).


20,000 ,

18,000

12,000

SOURCE Officeof Technology Assessment, 1994 Based on D U Him-melsteln, S Woolhandler, J P Lewontin, D J Pound, “Health Care LatinForce U S , Canada, and West Germany, ’ contract paper prepared forthe Off Ice of Technology Assessment Cambridge, MA Center for Na-tlonal Health Program Studies, Harvard Medical School/The Cam-bridge Hospital Mar 19, 1993

Comparisons of the Hospital Labor ForceThe U.S. hospital labor force went from smallerper capita than Canada’s in 1971 (13,405 vs.18,446) to slightly larger in 1986 (17,690 vs.16,034) (figure 3-10). While the two countries hadcomparable numbers of managers and administra-tive personnel in hospitals in 1971, by 1986 theUnited States had substantially more of all threecategories of administrative workers (managers:1,191 vs. 607; administrative support personnel:3,035 vs. 2,108) (figure 3-11 ). In 1986 U.S. hospi-tals also employed more social service personnel,technologists and technicians, and aides, whileengaging fewer registered nurses, food serviceworkers, and “other” personnel.

Comparisons of the Nursing Home LaborForceIn contrast to other health care workplaces, theUnited States had many fewer workers per capita

Us. 90%0

1,000 Confidence Interval

SOURCE Officeof Technology Assessment, 1994 Based on D U Him-melstein, S Woolhandler, J P Lewontin, D J Pound, “Health Care Labo r

Force U S Canada, and West Germany, ” contract paper prepared forthe Office of Technology Assessment Cambridge, MA Center for Na -tlonal Health Program Studies, Harvard Medical School/The Cambridge Hospital March 19, 1993

in nursing homes than did Canada in 1971 (2,720vs. 4,113), a difference that widened even furtherby 1986 (5,236 vs. 8,850). The difference in 1986is explained by fewer managers and administra-tors (506 vs. 1,1 81), nonhealth professional andltechnical workers (16 vs. 1,477), social servicepersonnel (168 vs. 953), registered nurses (408 vs.904), therapists (47 vs. 387), food service workers(313 vs. 617), and other workers (101 vs. 1,398),Although the United States had more aides (2,609vs. 1, 121) and cleaning personnel (566 vs. 467)per capita, this discrepancy may in part reflect dif-ferences in classifying workers; many people clas-sified as aides in the United States probably ap-pear as “other” in the Canadian data (21).

lmplications of Labor Force AnalysesWhat do these results say about the relativeamount of health care administration in Canadaand the United States? What do they tell policy-


makers about the two countries’ overall healthcare systems and the usefulness and limitations ofhealth care labor force analyses more generally?

The results of Himmelstein and colleagues’analysis is consistent with studies finding that theUnited States spends more on measurable healthcare administration than does Canada. In addition,their analysis shows that the growth in administra-tive personnel is the largest contribution to the in-creasing divergence in the per capita sizes of theAmerican and Canadian health care labor forcesduring the 1970s and 1980s.

As a proxy for total spending on administra-tion, labor data are limited as they provide no in-sights into the relative wages in Canada and theUnited States that could explain at least part of anydifference in spending, although recent analysesindicate that the two countries have similar wagesin the health care sector (7,17,49). Another limita-tion is that personnel data do not offer a solution tothe problem of potentially unmeasured costs in apublicly financed system.

Although Himmelstein and colleagues’ workdemonstrates that analysis of census data and pop-ulation-based surveys are particularly useful inunderstanding trends in the use of labor resourceswithin given countries, there are limitations in us-ing the data to make international comparisons.As suggested earlier, one major limitation in thisanalysis is the inability to identify nonmedicalpersonnel in the United States who perform healthcare duties in nonhealth care settings, particularly,administrative personnel in private firms who ad-minister their employees’ health insurance bene-fits. Insurance companies in the United Stateswrite policies for more than just health care ex-penses, and it is not possible to determine from the

CPS data what proportion of all these administra-tive personnel is devoted to health insurance.Even though it was not possible to count theseworkers, the United States had more administra-tive personnel than Canada in 1986. The effect onthis U.S./Canadian comparison of including allpersonnel who administer insurance outside ofhospitals or providers’ offices is unclear, sincedata from neither country separately identify gov-ernment workers at the national or state/provinciallevels who administer insurance programs. Inclu-sion of insurance company administrators wouldonly broaden the gap between the two countries.

In examining the United States and Canada,Himmelstein and colleagues appear to have cho-sen two countries that employ largely comparableoccupational classifications. Where discrepanciesexist, they occur either in relatively small occupa-tional categories (e.g., the n.e.c. categories) or areknown and taken into account by the authors intheir analysis and interpretation (e.g., exclusion ofdentists from the practitioners’ offices categoriesin Canada). However, extension of this analysis toother countries can prove problematic. Himmels-tein and colleagues’ attempts to explore the healthcare workforce of the former federal Republic ofGermany using census data foundered on difficul-ties in interpreting some German occupationalcategories and differences in classification con-ventions. Their experience suggests that while in-ternational labor force comparisons may offer im-portant insights into structural differences in thehealth care systems of different countries andsome of the implications of potential changes inour own country, the analysis becomes more diffi-cult to interpret and requires greater resources asthe culture and language become more foreign.

To SimplifyAdministration 4

The international comparisons examined thus far rest onthe premise that aspects of other countries’ systemsmight be less administration-intensive than the U.S. sys-tem. Adoption of some other system, or aspects of it,

might then be a way to reduce administrative Costs here. How-ever, aspects of health care administration that are essentially in-dependent of the reimbursement system and changes in them alsohold the potential for savings. Some of these are:

■ standardization of insurance claims forms,. electronic submission and payment of insurance claims (which

would require standardizing claim forms), and■ the use of card and other technology to keep administrative

and/or medical information in electronic format.

Although some health care reform proposals in the UnitedStates contain some or all of these changes, ] consideration ofsuch technological changes predates proposals currently beforeCongress to change the U.S. health care system.2 Few of these ef-forts have relied on analyses of similar uses of technology tostreamline administration in other countries. In large part (but notentirely), this is because there are few examples on which to draw.

I A1lothcr recent OTA repofl examines the assumptions and methods underlying esti-rnatc~ of nationul hetilth ckpenditure~ under major health care reform proposals in the~lnr[cd Sta[e\, inc]uding e~[ima[es of adminiswative costs (47). This report briefly reviewsai~unlplions made about administrative ~a~ ingi expected from standardization and au-tomation. but point~ out that \uch projcctcd sa~ ing~ are relatively minor compared withother categoric~ of health e~penditures.

2Hear]ng\ held before the Houw Subcommittee on Health of the Committee on Waysand ~lean~ rev Icw cd such efforts through Apri 1 of 1992 (45).

| 47


STANDARDIZATION AND AUTOMATIONOF INSURANCE CLAIMSThe multiplicity of payers in the U.S. health caresystem results in no standard form or set of proce-dures through which providers or patients can bereimbursed for services. By definition, suchmechanisms exist in countries that have single-payer systems. To the extent that these countriesreimburse on a fee-for-service basis, this includesa standardized claim form and, in some countries,electronic claims filing and payment. Analystssuggest that a standard form in the United Stateswould save money by reducing the amount of t i meproviders and patients spend trying to understandand complete them (37). They claim that electron-ic submission and payment would reduce person-nel and paperwork costs involved in preparing,processing, and paying claims. Estimates of themagnitude of these savings vary considerably,however (58).

In November 1991, then Department of Healthand Human Services Secretary Louis Sullivanformed the public-private Workgroup for Elec-tronic Data Interchange (WEDI) to standardizeelectronic communications in the health care in-dustry. Through a steering committee and adviso-ry groups, WEDI has issued two reports to theSecretary with recommendations and cost projec-tions (57,58). The 1993 report suggests that theuse of electronic communications to administerthe current U.S. health care system could save $13billion to $26 billion annually, not counting theinitial implementation costs of $5 billion to $17billion.3

Among the international comparisons re-viewed earlier in this paper, only Sheils andYoung specifically address the impact of automa-

tion on administrative costs (37).4 They estimatemore modest savings from these changes thandoes WEDI. They also suggest that standardiza-tion of claims forms in and of itself is likely to re-sult in only very small savings because most pub-lic and private insurers already accept HCFA’sclaim form in lieu of their own, and for those whodo not, software exists for the easy creation ofclaim forms according to insurance companies’standards. Finally, Sheils and Young state that us-ing a standardized format to process claims elec-tronically would save about $0.50 per claim (ac-cording to unspecified industry data), resulting in$400 million in total annual savings.5

HEALTH CARDSThe use of card systems represents another poten-tial change in the administration of health care inthe United States. Health card systems compriseseveral underlying technologies and multiple ap-plications designed to reduce costs, improve qual-ity of care, or both (26). Card systems usually con-sist of the card itself and “readers’ ’-computerterminals or other devices that can read, translate,and in some cases, record and update data on thecards. The cards themselves can be of the follow-ing types (29):

■ Simple paper or plastic cards. Most health in-surance programs already use these to identifythe card-holder and the type of insurance he orshe carries. The issuer of the card prints or eml-bosses the information directly on the surfaceof the card so that it can be read directly byanother person. Some hospitals also use thistype of card system to identify their patients.This is the least expensive of the card technolo-

3WED1 breaks ~ese estimates down into tieir component parts and indicates that they were prepared by a technical advisory .grOup (.58).4Among he o~er major quantitative attempts t. compare administrative costs in the United States and Canada (Z0,QS,44.SA), Standardi~a-

tion and automation mayor may not be subsumed among the bundle of changes assumed to take place if a Canadian-st~le single-payer system isimplemented in the United States.

Sshei]s ~d ~ol]eagues also assume hat a~opti~n of a Canadian-style system would reduce physician Oflice administrative expenses for

claims tiling and patient billing by 50 percent, but they do not imiicate how much (if any) of this reduction is attributable to standardization andautomation as opposed to the simplified reimbursement rules of a single payer (37).

Chapter 4 Technology To Simplify Administration 149

●

■

■

■

gies and holds no more than the visible in-format ion.Magnetic strip cards. This technology is mostfamiliar to Americans in the form of automatedbank teller (ATM) and many credit cards. Themagnetic strip on the back of the card can holda limited amount of information such as thecard-holder’s identity and that person’s insur-ance coverage. Information on these cards canbe changed. The manufacturing costs of thecards range from $0.20 to S1.00. Readers costbetween $300 and $800” (U.S.). Newer, moresophisticated magnetic strip cards can hold sig-nificantly more data. These cards cost two tofour times more than conventional cards, andthe readers are up to three times more expen-sive.Smart cards. This term refers (o a family of’ re-lated technologies in which a silicon microchipis embedded within a plastic, wallet-sized card.Some cards are made only for storing data, buttrue smart cards are able to process data as acomputer would. The microprocessor’s centralprocessing unit (CPU) controls access to thecard’s memory (i.e., data storage) as well ascommunications with the smart card reader viametal contacts on the face of the card. Cardsvary in the size of their memory and their abili-ty to update data stored in their memory. Thecards’ manufacturing cost ranges from $1 to$50, depending on their capabilities, manufac-turer, and quanitity produced. Readers for smartcards are cheaper than those for magnetic stripcards, ranging from $50 to $250. (Combinedmagnetic strip and smart card readers run be-tween $700 and $800. )Optical cards. Like compact disks, these cardscan record large amounts of in format ion in dig-ital format, making them potentially useful forextended medical records. However, once in-formation is recorded on the card, it cannot bechanged. This technology is also expensive.with cards costing between $5 and $20 andreaders from $3,000 to $4,000.Holographic cards. This technology. in whichdata is recorded in a hologram embossed on thesurface of a plastic card, has been used mainly

as payment for public telephone calls. Its rela-tively large potential for fraud, its lack of flexi-bility, and the cost of its readers ($1 ,000) havelimited interest in this technology for healthcare applications.

■ PCMCIA/JEIDA cards. This technology re-fers to a standardized format defined by the Per-sonal Computer Memory Card InternationalAssociation (PCMCIA) and the Japan Elec-tronics Industry Development Association(JEIDA). Such cards can store large amounts ofinformation and are designed to fit into slots onthe back of personal computers, terminals thatare part of a larger computer network, or otherelectronic devices. Two manufacturers havedeveloped smart cards that can be read in aPCMCIA, allowing any computer with such aslot and the necessary software to become asmart card reader. Although precise cost dataon these cards are not available, they are moreexpensive than conventional smart cards, mak-ing this technology most cost-effective for ap-plications involving the storage of largeamounts of information.

Uses of card systems in health care to date canbe divided into four categories that describe theirfunctions. Some specific card systems currently inuse have more than one function (See box 4-1.):■

■

Health insurance card systems. Designed toreduce administrative costs by simplifying in-surance claims and reimbursement proceduresand facilitating admission to hospitals or othermedical institutions, these cards can contain in-formation identifying the card-holder, his orher insurance policy, and information aboutcovered services and the extent of payment.Such cards can be components of electronicdata interchange systems that electronically re-imburse providers without the use of paperclaim forms.Medical card systems. These systems usecards to store patient medical information or toserve as a key to a larger computer database thatcontains such information. Their purposes areto 1 ) improve the quality of care by reducing theduplication of medical tests, preventing the use


Smart card technology IS largely a French Innovation, and France has begun to use smart cards in

many sectors of its economy, including health care The French experience offers inslghts into the po-

tential contributions and limitations of health card systems for other countries

The Uses of Smart Cards in the French Health Care System

French experiments with health cards Include examples of all four types of systems discussed in the

text Insurance cards, medical cards, emergency cards, and health professional cards. These Include

projects sponsored by the national government and the primary insurers in France as well as by com-

mercial isurers and mutual aid societies that offer complementary private insurance, and projects de-

signed for limited populations

Projects Sponsored by the National Government and Primary InsurersVitale/SESAM Card. Begun in 1989 by CNAM-TS (the National Health Insurance Administration,

which administers the primary health insurance for 80 percent of the French population as part of the

country’s social security system), this experiment seeks to replace paper insurance claims forms with

smart cards The experiment currently includes about 140,000 residents of Boulongne sur Mer (a city in

northern France) who are insured by the social security system. Three-quarters of the city’s medical

professionals participate Encoded on the smart card is the card-holder’s name, social security number,

birth date, and information about the extent of coverage and payment under the beneficiary’s insur-

ance. To protect the security of Information contained on the card, it also contains a confidential code

that the card-holder must enter into the reader at each medical visit The second stage of this experi-

ment WiII expand the cards to additional cities with hopes of including the entire nation by the year

2000 The major criticisms of Vitale/SESAM have come from physicians who complain that they are usu-

ally the ones to update Information on the cards, requiring time and resources. They also have com-

plained that inclusion of a diagnostic code on the claim form, a novel concept in France, could jeopar-

dize doctors’ professional autonomy

Santal Card. This card, first used in 1987, holds both admministrative Insurance and medical informa-

tion for patients treated in any one of eight hospitals in Saint-Nazaire, a region of western France In

addition, 300 medical professionals outside the hospital including 11 medical laboratories accept the

card In addition to reducing administrative costs within the hospital and simplifying admission proce-

dures, the designers of this card hope it WiII improve the flow of information among hospitals, laborato-

ries, and other medical providers The medical information contained on the card is limited to recent

tests and treatment and basic information needed in an emergency, although the administrative identifi-

ers on the card could be used as a key to more complete data files By October 1992 about 35,000

cards and 160 card readers were in use In addition to expanding the number of card holders, adminis-

trators of this card system plan to use more sophisticated smart card technology as it IS made avail-

able Cards with greater storage capabilities will allow for additonal Information, including drug pre-

scriptions and nursing records

The Health Professional Card. Already in existence for some local projects like the Santal card

described above, the Ministry of Social Affairs and Integration IS working with all parties in France using

health card systems to develop a standard format for Health Professional cards As described in the

text, physicians and other health professionals will use these cards to gain access to information on

patients’ cards or in other computerized databases, they serve as a means of preventing unauthorized

access to confidential patient records.

(continued)

Chapter 4 Technology To Simplify Administration |51

I

I

Projects Sponsored by Complementary Insurers

Carte Sante. This project uses smart cards to create portable adminlitrative and medical files for

patients and to initate payment to medical professionals With this card, patients do not have to pay

physicians out-of-pocket and then seek reimbursement from their Insurer Since 1989 the Federation of

Mutual Insurance Companies of France (FMF) has issued 250,000 cards and 1,000 card readers to

beneficiaries and providers in the regions of Provence-Alpes-Cote d’Azur, Rhone Alpes, Languedoc

Roussillon and Burgundy FMF provides complementary Insurance coverage for services and copay-

ments not reimbursed by social security or other primary health Insurers The administrative file con-

tains patient identifying data and information about the patient’s “reimbursement rights” under his or her

Insurance policy and other Information needed to pay the provider The medical file contains emergen-

cy medical lnformation and records of preventive health services received

Sante-Pharma Card. This card eliminates the need for patients to pay pharmacists in advance for

their prescription drugs Launched in 1986 it iS the result of an agreement among insurers (both prima-

ry and complementary) and the national pharmaceutical syndicate The card, which contains informa-

ton about the patients complementary insurer and pharmaceutical coverage, iS used along with the

paper social security card indicating the patents primary health Insurer and an optically read paper

claim form Two milllon cards are in use in 76 administrative zones (called departments) representing

77 percent of French pharmacies. Pharmacies file about 800,000 Insurance claims each month

Projects Designed for Specific Populations

French Army Health Card. This smart

card contains administrative Information

on patients treated in French army hospi-

tals Since 1988 the Army has implement-

ed this project on an experimental basis

in two hospitals with the potential to ex-

pand to 20 others The card which holds

no medical Information and iS not used

as a means of paying providers, has two

forms The “personal” card iS provided to

patients who are treated at Army hospi-

tals on a long-term or recurring basis and

gives them direct access to all hospital

services A “shuttle” card iS provided to

patents who are expected to have a The Robert Debre' Hospital in Paris is part of the Frenchshort one-time hospital stay The cards health care system, which iS characterrzed by universa/

are designed to eliminate paper records coverage, mu/t/p/e insurance schemes financed through

by recordng pat ient ident l fy ing in- payroll taxes, and public and private providers

formation data on Insurance coverage and the number of previous hospital stays. As of November

1992 60000 personal cards and 30000 shuttle cards were in use A total of 270 hospital employees

were authorized to use the system on 55 card readers

Paris Sante Card. This iS one of several card systems developed by local health authorities to im-

prove access to health services for poor jobless, or homeless indviduals Available since 1989, it IS the

result of an agreement among the city of Paris and 6,800 health providers The local health authority

(continued)—


administers health Insurance through the national social security system for unemployed individuals

and their familles. The card IS made of embossed plastic This system could use smart card technology

in the future, although there currently are no specific plans to do so The card allows beneficiaries free

choice of any participating provoder and providers file paper claims for which they receive payment

within 10 days (a process that took as long as six months)

Dialybre Card. This smart card contains both administrative and medical information for kidney dial-

ysls patients receiving care at any of three French hospitals Begun in 1989, it is designed to provoie

dilaysis patients greater freedom to receive treatment at a location other than where they usually go It

avoids duplication of medical records, reduces the time necessary for admissions, and offers greater

communicatoin among facilities providing care to an individual patient In addition to patient identifiers

and Insurance Information, the card contains emergency medical data and the patient’s dialysis history

As of 1992 about 1,100 of France’s 15,000 dialysis patients had cards, Financed by Insurance compa-

nies, private foundations, and drug firms, the system IS currently expanding to at least 50 dialysis cen-

ters with the long-term goal of revolving all 600 such facilities.

Issues Raised by the Use of Smart Cards in France

The experiments with smart cards in France have given rise to a number of general or cross-cutting

issues that must be considered in their expansion to involve larger numbers of people and institutions

or to their transfer to other countries Among the most significant are 1) standardization of technology

and format, 2) patient confidentiality, 3) professional autonomy, and 4) costs

Standardization. Gwen the large number of different health card experiments under way on a rela-

tively small scale m France, standardization of the technology and design of the system IS Iikely to be

necessary if any of these projects are to be Integrated into one or two cards that uses a single type of

reader Such Integration may be a means of achieving economies of scale in establishing and running

card systems, although they could run counter to the concerns over confidentiality and professional

autonomy outlined below.1 Standardization of card systems iS not just a concern in France, but through-

out the European Community, which has established standards for data to be included on emergency

medical cards Furthermore, Germany has already begun to provide smart cards with administrative

health Insurance Information to its citizens Other European nations are conducting their own smart card

experiments The problem of standardization of technologies iS complicated by the multiple choices

available to policy makers and the rapidly growing capabilities of smart cards and other new technolo-

gies One strategy for standardization in France would be the full Implementation of a card system in a

program that Involves all or most French citizens. The natural candidate would be the Vitale/SESAM

card being developed by the CNAM-TS that covers 80 percent of the French population. The final de-

sign of that card could take the needs of smaller systems into account Once Vitale/SESAM is in place,

smaller systems might feel an economic Incentive to adapt their design to the larger system To date,

the government has not begun to provide the Vitale/SESAM card to all social security beneficiaries

Patient Confidentiality. As in the United States, confidentiality of patient medical records is a major

public concern. To develop appropriate poilcies for the use and protection of all prviate records in

France, the Parliament established a commission (Commssion Nationale de I’lnformatique et des Liber-

tes, or CNIL) that enforces a 1978 law governing Information systems and Individual rights, CNIL must

approve all government programs that establish information systems on French cittzens, including

smart card projects The health professional card and security codes that patients must enter to gain

1 Slandardlzallon could Increase the amount of patient Information to which an individual could potentially gain unauthorized ac-cess, although It does not affect the probability of overall unauthorlzec access

(continued)—


access to these records are two measures designed to protect computerized medical records. How-

ever even with these safeguards, there IS not yet a consensus or even a proposal to establish a full

medical record in any electronic form in France

Professional Autonomy. An issue in France that has not been a major concern to date in the United

States concerns the autonomy of medical professionals In particular, they worry that the inclusion of

detailed medical records on health cards or other computerized systems wiII make them vulnerable to

questioning of their medical Judgment by other physicians, insurers, or the government This concern

has contributed to the Iimited amount of medical records included in computerized systems and has

even kept diagnostic information off Insurance claim forms

Costs. Setting up a card system involves slgnificant costs in choosing the appropriate technology

decidnig what information IS to be placed on the cards, having the cards manufactured and distributed

and educating patients, providers, and administrators in their use Although standardization of card

systems would offer opportunities for economies of scale, some organization must bear these initial

start-up costs The ongoing costs and risks of using a card system must also be weighed against its

benefits

The French Health Care System

The French health care system iS characterized by universal coverage of the population through one

of several programs financed through payroll taxes (comprising contributions from both employers and

employees), a mixture of public and private hospitals, ambulatory care offered mainly through private-

practice physicians, patient choice of providers, and professional autonomy for physicians

Patients usually pay their physicians directly on a fee-for-service basis and are reimbursed by insur-

ers. Physician fees are set through negotiations among the government, insurers, and providers, al-

though physicians are free to charge patients more than these fees. Public and most private nonprofit

hospitals receive fixed budgets A small number of private, for-profit hospitals handle most surgical and

obstetric cases, receiving revenues on per-diem or fee-for-service basis Eighty-four percent of the pop-

ulation has private health insurance to cover services not paid for by their primary Insurance

In 1990 France spent 91 percent of its gross domestic product on health care Payroll taxes cover

74 percent of personal health expenditures, with another 16 percent being paid out-of-pocket by pa-

tients and their families The remainder iS financed through public subsidies and complementary pri-

vate health Insurance

SOURCES : VG Rodwin, S Sandier, “Health Care Under French National Health Insurance, ” Health Affairs fall 1993 pp 110-131,

E M Monod, A Tour d Hor[zon of Health Cards In Europe, Srnarf Card Techno/ogy/nfernaf/ona/ (January 1994), E M Monod Minis-try of Social Affairs and Health, International Relations Republic of France Personal communications Mar 30, 1994 June 13 1994N Paquel C Frizzole S Glaziou Smart Cards in the French Health Care System Final Report Unpublished OTA Contract Paper

Paris France 1993

of therapies or procedures incompatible withthe patient’s overall medical condition, andhelping to ensure that patients with chronic orspecial medical conditions receive needed ser-vices; 2) facilitate communication betweeninstitutions, such as hospitals and patients’ per- ■

sonal health professionals; 3) simplify hospitaladmissions; and (4) help in the collection of

health statistics. Technological limitations andconcern over the privacy of medical recordshave limited the extent of card systems de-signed to hold extensive amounts of informa-tion.Emergency card systems. These systems con-tain only essential information identifying thecard-holder and medical information—such as


Smart card systems, which have played an increasing/y signifi-cant role in health care systems of France and other Europeancountries, consist of smart cards (left) and readers (right) usedto read and update information contained on the cards

■

chronic illnesses, blood type, and allergies—important in case of medical emergency. Ifavailable over wide geographic regions, suchsystems could make travel safer, especially forthose with existing medical conditions.Health professional card systems. These sys-tems are designed to help protect the security ofpatient medical information and are used inconjunction with other card systems or largercomputerized databases. Issued to individualhealth professionals, they serve as access keysto patient information. They can be designed tolimit the health professional’s access to onlythose data needed to perform his or her job.

Understanding the potential for card systems inthis country comes, in large part, from experienceswith them in other countries. While experience inother countries may be instructive when consider-ing potential applications and problems of cardsystems, analysis of their cost implications offerminimal lessons for the United States for severalreasons:

■

■

The underlying technologies and their costs arechanging rapidly;The level of costs associated with card systemsin many countries depends heavily on thosecountries’ reimbursement systems, which maydiffer fundamentally from that of the UnitedStates; and

■ Most experience with card systems in othercountries so far has been limited to demonstra-tion projects among very specific populationsor geographic areas; applications among largergroups for extended periods may realize econo-mies or diseconomies of scale not found in ini-tial experiments.

In an attempt to understand more about anothercountry’s experience with cards, OTA commis-sioned an analysis of France efforts to use so-called smart cards in their health care system.Smart cards, which are usually the size of creditcards, have an embedded silicon microprocessingchip that can store and process information. Usu-ally issued to patients or health providers, they canstore administrative or medical information orserve as a key to gain access to a larger medicalcomputer system. In addition to describing thevarious applications of this technology in France,the OTA-commissioned analysis also examinessome of the difficulties experienced in imple-menting smart card projects. (See box 4-1 for asummary of this analysis.)


Health cards are just one piece of an overall sys-tem for administering health care and maintainingmedical records. The decision to use cards or tochoose a specific type of card technology is de-pendent on the intended application, the system’susers, and the cost.

In France implementation of card systems washindered by concerns over the confidentiality ofcard systems and difficulties in getting physi-cians, administrators, and patients to keep in-formation on cards or other computerized medicalrecords. These issues are likely to arise in theUnited States should a card system be implement-ed. However. concerns arising from French physi-cians’ tradition of not sharing diagnostic or thera-peutic information with other health professionalsor payers should not cause problems in the UnitedStates.

The Clinton Administration’s proposed HealthSecurity Act (S. 1757) would issue every Ameri-can citizen and legal resident a Health Security

Card. Some Administration documents have indi-cated that this card would employ a magnetic striprather than smart card technology. reflecting an at-tempt to reassure patients that these cards will pro-tect their privacy by containing only basic identi-fication information similar to that contained on abank automated teller machine card rather thanencoding any sensitive medical records (50).6

In reality, the experience from France, wherepatient privacy also has been a major issue, sug-gests that protection of such privacy has less to dowith the choice of magnetic strip or smart cardtechnology than with the privacy safeguards builtinto the overall computer system. Any kind of sys-tem has the potential to limit the amount of in-formation in the system and access to it (29).

The Administration has given no assurance thatthe adoption of Health Security Cards will resultin administrative savings apart from the adoptionof standardized claim forms (50).

hAno[her ~cccnt OTA ~[udy examlne~ Privac} is~ues in computerized medical records in greater det~il (~)

OTA wishcs to thank the individuls and organizations listed below for their assistance with this background pa-per. These individuals and organizations do not necessarily) approve, disapprove, or endorse this report. OTA as-sumes full responsibility for the report and the accuracy of its content.

Gerard AndersonCenter for Hospital Finance and ManagementJohns Hopkins UniversityBaltimore, Maryland

Eivind HoffmanBureau of StatisticsInternational Labor OfficeGeneva, Switzerland

Rosamund KatzHealth Finance and Policy IssuesU.S. General Accounting OfficeWashington. District of Columbia

Hagen KuhnWissenschaftzentrum Berlin fur SozialforschungBerlin, Germany

Daniel L. MaloneyU.S. Department of Veterans AffairsWashington, District of Columbia

Elsbeth MonodMinistere des Affaires Sociales,

la Sante et de la VineDivision des Relations InternationalesRepublique FrancaisParis, France

Walter PeisslTechnology Assessment UnitAustrian Academy of SciencesVienna, Austria

Thomas S. ScoppLabor Force Statistics BranchBureau of the CensusU.S. Department of CommerceWashington, District of Columbia

Stephen SeidmanSmart Card MonthlyMontara, California

Kenneth E. ThorpeU.S. Department of Health and Human ServicesWashington, District of Columbia

I 57


C. Peter Waegemann J. Terry WilliamsMedical Records Institute EBT/Smartcard Manager

Newton, Massachusetts WIC ProgramDepartment of Health

Claudia WildTechnology Assessment Unit

State of WyomingCheyenne, Wyoming

Austrian Academy of Sciences Barbara H. WootenVienna, Austria Bureau of Labor Statistics

U.S. Department of LaborWashington, District of Columbia

Appendix B:Comparison of

Health Care AdministrationFound In

Four Countries BBused on W.A. Glaser, “Administration in Health Care: A Plan for Cross-National Comparison s,” contractpaper prepared for the Office of Technology Assessment, revised edition, 1993.

I 59

—

Health care system

Multiple pubilc andprivate payers. Publicprograms pay forhealth care for elderly,disabled, and indi-gen! citizens; someveterans, active mili-tary personnel andtheir families, Mostproviders are autono-mOUS, with a growingnumber of practitio-ners employed bycapitated health in-insurance plans or partof one or more net-works of providersassociated with athird-party payer thatestablishes variouscost-containmentmeasures (managedcare).

Informationand

publication

Federal governmentcollects vital statisticsand morbidity data fromstate and local govern-ments and publishesthem, collects and dis-seminates data onMedicare program forelderly citizens andMedicaid program forindigent citizens. Otherfederal agencies andprivate organizationscollect and disseminatedata on health care fa-cilities, personnel,practice, organization,financing, and the ef-fectiveness or cost-ef-fectiveness of particularinterventions.

ImplementingAgencies of

Policymaking Government

Multifaceted and occurs atall levels of governmentthrough the executive, legis-Iative, and judicial brancheswith support from their staffsand agencies, commissions,private-sector foundations,and interest groups. Federalgovernment makes policy forprograms in funds and drugand device regulation. Stategovernments with primary re-sponsibility for insuranceregulation and licensing ofhealth facilities and person-nel, admmistration of Medic-aid program within the state,and shared responsibiltywith local governments forpublic health programs,

Government develop-ment and updating ofregulations to imple-ment legislation andprograms (especiallyat federal level) Ad-ministration of publicclinics and hospitalsat all levels of govern-ment.

ProviderOrganizations

In addition to usual internaladministration, hospitals,nursing homes, and homehealth agencies requiresignificant administrativepersonnel and infrastruc-ture to understand reim-bursement rules and pro-cedures for multiple payers(including managed careorganizations) and billthose payers and/or pa-tients. Hospital administra-tion also includes imageand marketing, Iitlgation,regulation and accredita-tion, and management ofadmittlng privileges. Someprivate proivder organiza-tions are part of chains thatcentralize marketing, sup-plies, and financial man-

aement activities.

IndividualPractitioners—

Move from billing of patientsto direct billing of insurershas increased administrativecosts for individual practitio-ners because of varying re-imbursement rules and man-aged-care procedures. A feeschedule exists for onlyMedicare, hence, practitio-ners or their staff often checkwith Insurers on acceptablecharges before doing proce-dure. Fear of Iiability mayadd to administrative costsby increasing volume of re-cords kept and need to shopamong Iiability insurers.Growth in group practicesand group and staff modelHMOS alleviates some ad-ministrative burdens forassociated physicians.

(continued)

o

Providerassociations

Numerous providerassociations at na-tional, state, and locallevels requiring signif-icant admmistrativesupport. They lobbyfor their members in-terests, interact withthe mass media, pub-lish professional jour-nals, operate profes-sional committees,conferences, andworkshops, providemembers for govern-mental and other ad-visory commissions,and collect and pub-lish statistics abouttheir membership.

Suppliers

Drug and device sup-pliers face administra-tive costs related tomarketing to physiciansand other customers,patenting and relatedactivities, Iicensing bythe Food and Drug Ad-mmistration, and lobby-ing Growing adminis-trative effort devoted tointeraction with third-party payers about cov-erage and reimburse-ment levels. Medicaiddrug reimbursementsindirectly regulatedthrough rebate schemerequiring administrativeactivity by manufactur-ers.

Insurers

Government: States reim-burse providers for servicesprovided under Medicaidwith state-by-state variationin rules and benefits andshared Federal and Statecosts, nonstandardizationmay raise administrativecosts Medicare contractswith private insurers to proc-ess claims and reimburseproviders within defined geo-graphic areas Existing infra-structure within these privatecontractors helps minimizeMedicare’s administrativecosts Federal governmentbears Medicare administra-tive costs of developing reg-ulations, resolving disputes,and contracting.Private: Prviate insurershave slgnificant administr-ative costs associated withmarketing in a highly com-petitive environment, under-writing and rate negotiationwith employers, benefit de-sign, application processing,determination of provider eli-gibility, claims processingand reimbursement, reservesmanagement, and financialreports Self-insured employ-ers face all of these costs ex-cept marketing and applica-tion processing. Managedcare procedures introducedto contain costs and insurequality raise administrativecosts.

Education

Very large number ofspecialized educationprograms (degreeand continuingeducation) for hospi-tal and health careadmministration

Research

Siginifcant volume ofhealth services and relatedresearch done in acade-mia, government, and pri-vate sector, all resulting inits own administrative ex-penses

Managementconsulting

Significant amount of man-agement consulting andsupplementary conferenceswithin health care organiza-tions covering finance, gov-ernment standards regula-tions, reimbursement rules,and labor standards

Information implementingand Agencies of Provider Individual

Health care system publication Policymaking Government Organizations Practitioners

Full government fund-ing of health care de-centralized to provin-cial level. Autono-mous providers thatfollow provincial stan-dards for financial ac-counting. Providerassociations repre-sent interests of doc-tors and hospitals.Little private health in-surance


Usual vital statistics.Provincial collection ofdata from hospitals andother provider orga-nizations about ser-vices, utilization, per-sonnel, and spending,aggregated by nationalhealth ministry. Providerassociations collectand aggregate dataabout their membersfor reimbursement ne-gotiations.

Decisions about changesin system made by pro-vincial government (min-istries, cabinet, legisla-ture, and ad hoc com-missions). National re-sponsibilities for drug li-censing and pricing, vitalstatistic reporting guide-lines

Suppliers Insurers

Incur large portion ofCanada’s administrativecosts. Provincial minis-tries (or delegated dis-trict councils) scrutinizehospital reports, negoti-ate total budget with trea-sury, allocate annual in-creases among hospi-tals, distribute grants forconstruction, inspecthospitals for compliancewith safety, personnel,and quality regulations.Some provinces also re-imburse for nursinghomes and home healthcare agencies usingsame procedures as forhospitals. Provincial pub-lic corporations negotiatewithl physician associa-tions for fee scheduleand process claims andarbitrate disputes.

Education

Usual organizational man-agement (personnel,physical plant, supplies,inventory, medical records,patient communication,and marketing), Hospitals’prospective budgets, retro-spective cost reports, andspecial requests for grantsfrom provincial ministriesfor capital Improvementsconstitute relatively simpleform of administration, indi-vidual patient billing foramenities. Limited numberof teaching hospitals mini-mize administrative costsassociated with residentsand research.

Research

Usual expenses of running amedical or dental office withsome sharing of offices, es-pecially in urban and ruralareas. Practitioners completefee-for-service forms by mailor Computer and send topublic corporation; paid byelectronic transfer or periodiclump sums, Billing of patientsor Private Insurers for dentist-ry, extra services, and treat-ment of foreign patients.


Provincial associa- Drug and device Limited portion of total Admminstration of one or Health services research Minimal. Limited to manage-tions with staff to col- manufacturers with ad- national administrative more university health Iimited to university teams ment information system de-Iect and analyze clini- ministrative work to sup- expenditures because of care administration pro- supported by provincial velopment, computer train-cal and economic port pateninng, licens- small size of private in- grams in each province, governments to perform ing, and consulting, Hospitalstrends, publish pro- ing, and pricing regula- surance market. Adminis- minimal compared tofessional journals,

policy-oriented research use management manualstion by national govern- tration includes under- United States, where on health economics, ser- developed by their provincial

communicate with/ ment wrlting, marketing, ap- many Canadian health vices, and technologies. and national associations.lobby ministries, leg- plication processing, care managers receiveislature, media, mem- general overhead, claims their education.bers, and provide processing, and reim-data to national bursements Employersassociations. National that offer private insur-associations publish ance to employees maynational data and are have some administrativeparty to Iawsuits over expenses.issues affecting pro-fessions —

a Off Ice of Technology Assessment, 1994 Based on Glaser, W A , “Admmlstratton m Health Care A Plan for Cross-National Comparisons, ” contractor paper prepared for the Off Ice of Technology Assessment, revised edltlon, 1993

o

Health care system

National Health Ser-vice (NHS) owns andmanages hospitalsemploys specialistphysicians and con-tracts with generalpractitioners Minimal,growing local varia-tior in administrativeprocedures as somehospitals become au-tonomous Reim-bursement systemprovides little admin-istrative informationPhysician associa-tions play a role in ne-gotiating work rulesand other policy Lim-ited private hospitalsand private insur-ance


Unions and associa-tions with strong rolein negotiating forhealth professionalsIncluding NHS andhospital administra-tors, thus requirngtheir own administra-tive staffs

Information andpublication

Government producesvital statistics and dataon NHS services utiliza.tion, personnel, andspending No data onpatient or other privatehealth care spending

Policymaking

Health ministry assim-iIates analyses andrecommendationsfrom NHS, public, oth-er interest groups,and mass media toproduce staff reportson budget, Iegislation,potential reformsSupplemented bywork of Royal Com-missions and WorkingParties Fourteen re-gional boards sup-ported by staff makerecommendations tonational governmentParliament, Cabinet,and Prime Ministerand their staffs alsoInvolved m budgetand reforms

Suppliers Insurers

Drug and equipment Private health insur-companies require admin- ance Iimlted to acci-istrative staff to apply pat- dent, private hospital-ents and licenses to sell ization, specialist andtheirr products Drug com- other appointmentspanies also have adminis- without a wait, andtrative costs associated amenities Carrierswith price regulation and negotiate rates withNHS formulary approval prviate hospitals and

reimburse patients afixed rate for each pri-vate physician ser-vice performed

Implementing agenciesof government

Health ministry with staff sup-port competes withln Cabi-net for health budget NHSallocates to 200 DistrictHealth Authorities (DHAs) forreimbursement of servicesNewly autonomous hospitalswith administrative functionsof marketing, pricing, andbilling patients and DHAsFamily Practice Committees(FPCS), Independent ofDHAs, contract with generalpractitioners and dentistsFPCS track fee-for-service fordentistry and Increasingnumber of medical proce-dures, capitation payment forall other general practiceservices Ministry negotiateswith unions and professionalorganzations over employeepay NHS prepares periodicexpenditure reports fromDHAs and other organiza-tional units

Education

Litle specialized educationin health care admmistrationdue to relative simplicity andausterity of system Healthcare administrators tendedto be gifted amateurs andaccountants Specializedcontinuing education andworkshops have becomemore common since the1980s

Provider organizations

Increasing number of au-tonomous hospitals leadsto increasing administra-tive expenditures (market-ing to patients and generalpractitioners, developmentof clinlial emphases set-ting prices, budget balanc-ing) All nursing homes areprivate and face thesesame administrative ex-penses There are a smallnumber of private hospi-tals Chains own some pr-ivate hospitals and nursinghomes and perform someof their administration

Research

Slgnificant tradition of research in uiversties,government, and inde-pendent institutes abouthealth care and healtheconomics with particularemphasis on analyses ofpotential NHS reforms andevaluations after imple-mentation Specialized re-search has been neces-sary to learn about usuallyoverlooked pvivate sector

Individual practitioners

General practitioners (GPs)and dentists with usual ad-ministrative expenses of run-ning an office GPs musttrack patient enrollment andsend fee- for-service bills toFPCs for some services1980s Innovation of GP“fund-holding for patientsprovides Increased capita-tion payments to cover pa-tients’ tests, pharmaceuti-cals, specialist referrals andhospital cares results in in-creased administrative bur-den Dentists bill FPCs for allservices and must seek ap-proval for all extensive treat-ments


NHS has traditonally reliedon own staff and researchersfrom universities and inde-pendent Institutes Rise in au-tonomous hospitals andDHAs may give new opportu-nities to Private managementconsultants in the future

Health care system

Many insurers (sick-ness funds) in eachprovince allassociated with na-tional organizationHospitals are for-prof-it, nonprofit, and pub-IiC Government (atboth the national andprovincial levels) en-acts rules for the sys-tem, provides some fi-nancing, monitors,and settles disputesProvider associationsperform significantfunctions in negotiat-ing for and payingmembers

Information andpublication

National and provincialministries collect andpublish vital statisticsand data on some healthfacilities and personnelRelevant provincial pro-vider organizations col-lect data on hospital op-erations, spending, phy-sicians’ and dentists’work, and revenue onannual or quarterly basisindvidual provider datacome from claim forms.Provider data are aggre-gated and published byresearch centersassociated with nationalprovider associations.Provincial sickness fundassociations collect andpublish data about theirmembers National Minis-triles of Health and Laboraudit summaries of thesedata and publish theirown reports

Policymaking

Government role inadministration ofhealth system rela-tively small. Reformsof system crafted atnational level amongpolitical parties andInterest groups withinParliament, Cabinet,and ministries Recentreforms aimed at costcontainment andsome expansion ofbenefits. Public healthfunctions adminis-tered by provinceswithin national guide-lines developed inMinistry of Health andits secretariat.

Implementing agenciesof government

Government role in adminis-tering and paying for healthcare limited to provincialteaching hospitals, municipalhospitals, and local publichealth services Provincialhealth ministries license andInspect private hospitals andprovide grants to hospitalsfor capital improvementsMinistry staff evaluate needfor such grants. Public healthservices supported fromgeneral revenue

Provider organizations

Hospitals are mainly pri-vate nonprofit and for-profit, but public, munici-pal hospitals also oper-ate autonomously, Ger-man hospitals have rela-tively few staff, includingfor admministrative pur-poses Administrative ac-tivtis include usual iinternal administration,preparation of annualprospective budget, andbudget negotiations withcommittee of local sick-ness funds Negotiationshave been tradditionallyquick and simple, buthave become more strin-gent in the 1990s

Individual practitioners

German physicians use theiroffices to perform many am-bulatory procedures per-formed m hospital and outpa-tient clinics in other countries,thus requiring additional ad-minirstration to acquire equip-ment and supplies Physi-clans and dentists send outfee-for-service bills Physi-clans who work in privateclnics have hospital privi-leges and rely on the clinic to “

bill payers for them.

(continued)

Providerassociations Suppliers Insurers

All office physiciansbelong to provincialKassenartzliche Vereinigung (KV) that ne-gotiates with provin-cial committee ofsickness funds for alump sum and thenpays all claims Physi-clans do not bill pa-tients for any addition-al payments Provin-cial KVS with signifi-cant administrativeapparatus to negoti-ate with funds, trackmembers’ utilization,process and payclaims, and reducefees if necessary toavoid deficits. Nation-al association of KVSnegotiates with na-tional associations ofsickness funds overwork rules, reimburs-able procedures, feeschedules, andapproximate paymentlevels Provincial ar-bitration committeessettle disputes anddeadlocked negoti-ations. Provincial hos-pital associations per-form parallel functionsfor their members

Administrative work for Sickness funds enrollmanufacturers for pat- members, calculateents, marketing Iicenses, and collect premiumsand recently Introduced and social securitydrug price regulation pension contributions,

negotiate with hospitalsand KVS, scrutinize KVstatistical reports, com-municate with and payproincial associationof KVS and hospitals,cooperate with nationaland provincial financialaudits Marketing WiIIlikely increase due torecent reform increas-ing citizens’ choices infund enrollment Na-tional associations ofsickness funds haverelatively large adminis-trative burden strategicplanning, Iobbying forreforms, negotiating atthe national level, orga-nization of health insur-ance in former EastGermany, preparing re-ports, and publishingjournals for membersand the public Privatehealth Insurance pro-vides primary coveragefor 10 percent of popu-lation and has adminis-trative functions parallelto sickness funds. Pri-vate insurers also haveadministrative costsassociated with policiesfor long-term care andother extra benefitsEmployers’ admminstra-tvee work limited to pay-roll deductions andpayments to sicknessfunds

ManagementEducation Research consulting

Educational programs forhealth care managers tradi-tionally limited to generalbusiness and financial man-agement courses Somenew curricula in medicalschools and in-house train-ing by some sicknessfunds

Significant tradition of Significant number of man-health services research agement conferences andin universities and private workshops Some consultingInstitutes in Germany, iS done for new cost ac-often commissioned by counting methods orgovernment ministries Introduction of computing

technologies in hospitals, butit is Iimlted since all playerswork within a single set ofnational accounting stan-dards’ and necessary train-ing IS usually done by nation-al ministries or the contractconsultants

a Ofhce of Technology Assessment, 1994 Based on Glaser, W A “AdmmlstratJon m Health Care A Plan for Cross-National Compar~sons, ” contractor paper prepared for the Off Ice of Technology Assessment, revised edltlon, 1993

o

—0)al

Appendix C:Methods Used in Himmelsteinand Colleagues? Analysis ofU.S. and Canadianc Health Care Labor Forces

U.S. DATAHimmelstein and colleagues’ principal source of data for theUnited States is the Current Population Survey (CPS) AnnualDemographic File collected annually by the U.S. Bureau of theCensus and available in machine readable form since 1968. Him-melstein and colleagues analyzed the CPS file for each year from1968 to 1992. For several years they analyzed two different ver-sions of the CPS data, one prepared according to revised codingand/or weighting procedures and the other reflecting the proce-dures used in the prior year, in order to establish reliable timeseries.

The CPS is a Census Bureau survey of approximately 60,000households representative of the civilian noninstitutionalizedpopulation. About 6,000 individuals employed in the health caresector fall into the CPS sample each year. The part of the surveyconducted in March of each year collects demographic informa-tion and data on employment and income for the previous weekand for the previous calendar year. Himmelstein and colleagueschose to use the CPS rather than the Bureau of Labor Statistics’establishment survey (whose larger sample size allows estimateswith narrower confidence intervals) because the CPS spans alonger time period and the data are more closely comparable toavailable Canadian data. All estimates of numbers of health per-sonnel in the United States as a whole were derived from the CPSsample using the March CPS Final Weight, a multiplier assignedby the Census Bureau to each individual in the sample to allowaccurate extrapolation to the U.S. population as a whole, adjust-ing for thi sample design and the failure to obtain interviews with

66 I

Appendix C Methods Used in Himmelstein and Colleagues’ Analysis 167

management and related;administrative support, except financial;administrative support, financial;professional and technical except health;social service;other health diagnosing;therapists;other health assessment and treating;health technologists and technicians;aides and other health service;food preparation and food service;cleaning, building service and laundry;building construction and maintenance; andall occupations not elsewhere classified (n.e.c.).

In the Census Bureau’s classification, physi-cians, registered nurses, and licensed practicalnurses are each identified by a single code. Each ofHimmelstein and colleagues’ 14 other groups in-cluded several individual occupations.

Equivalents (FTEs)Himmelstein and colleagues defined one FTE as2,000 hours of work per year (40 hours/week x 50weeks/year). For years since 1976 the authorsconstructed this variable from responses to theCPS questions about place and occupation of em-ployment, and hours and weeks worked during theprevious calendar year. They calculated FTEs bymultiplying each respondent self-reported usualhours of work by weeks of work and dividing by2,000. However, prior to 1976 the CPS did notcollect comprehensive data on hours of employ-ment during the previous calendar year. For theseearlier years Himmelstein and colleagues ana-lyzed employment and hours of work based ondata for the week preceding the survey (which al-ways takes place in March), on the assumptionthat this single week’s data were representative ofemployment for the full, concurrent calendar year.Each respondent’s “actual hours of employment”in the reference week was multiplied by 52 and di-vided by 2,000 to arrive at an FTE figure.

Himmelstein and colleagues assessed the ef-fects of this methodologic change by calculating


health employment for 1975 using both the “lastweek” data from the 1975 CPS and the “last year”data from the 1976 CPS. Both 1975 estimates aregiven in each of the tables derived from the CPSdata. As expected, the number of people indicat-ing that they had worked in health care at any time“last year” exceeded the number saying that theyhad worked in health care “last week.” However,this discrepancy vanished after extrapolation toFTES during 1975. Thus, continuity of time seriesdata is somewhat better for FTEs than for numbersof persons employed.

Himmelstein and colleagues inspected graphsof time trend data on the number of persons andFI’Es employed in each occupation group. A dis-continuity was evident in the data by number ofpeople in 1976, while the FTE curve showed nosuch discontinuity. The gap between the lines fornumber of persons and FTEs was an indicator ofthe average work schedule for members of the oc-cupational group; for groups whose work year ex-ceeds 2,000 hours (i.e., physicians), FTEs exceedpersons. Conversely, part-time employment iscommon in many predominantly female occupa-tions in which the number of persons employedexceeds the number of FTEs.

Himmelstein and colleagues calculated FTEsper million population by dividing the number ofFTE health workers by the U.S. resident popula-tion as reported in the Statistical Abstract of theUnited States.

Occupation CodesBetween 1968 and 1991 the Census Bureau un-dertook two major reclassifications of occupa-tions following the 1970 and 1980 censuses(40,41,42), as well as several minor reclassifica-tions. The second of the major revisions involveda change in the philosophy of occupation classifi-cation, relying less on job titles and more on thecontent of work.

Himmelstein and colleagues dealt with theseclassification changes by preparing a comprehen-sive list of every occupational code represented inthe health sector between 1968 and 1991. For eachjob title Himmelstein and colleagues reconciledthe three systems of classification by comparingoccupation titles (and, when necessary, the oc-cupational definitions) in each of the classifica-I ion schemes. Where there was not a clear identitybetween occupational titles or descriptions in thedifferent systems, they allowed the codes to standas distinct occupations.

Sample DesignThe Bureau of the Census updated the recoding,imputation procedures for dealing with missingdata, and/or the weights used to extrapolate theCPS to the population in 1975, 1983, and 1987.For each of these three years, Himmelstein andcolleagues analyzed CPS data processed usingboth the old and new procedures, and report bothsets of values.

CANADIAN DATADetailed data on health care workers in Canadacome from the 1971 and 1986 Canadian censuses.Statistics Canada provided Himmelstein and col-leagues with data tapes including all individualsemployed in health sector industries, based on in-dustry classifications similar to those used by theU.S. Census Bureau since 1971. However, inspec-tion of the data revealed that more nonphysicianpractitioners’ offices appear to be classified under“Health and Medical Services, n.e.c.” rather thanunder the rubric “Offices of Practitioners,”compared with the U.S. data. This means (hatcomparisons of the labor force employed in practi-tioners’ offices in the United States and Canadaare subject to error.

The occupational classification of Canadianhealth care employees was based on StatisticsCanada’s 1971 Standard Occupational Codes(S. O. C.) codes. In most cases these codes closelycorrespond to the U.S. occupational coding sys-tem. Where discrepancies or uncertainties arose,Himmelstein and colleagues consulted with offi-

Appendix C Methods Used in Himmelstein and Colleagues’ Analysis 169

cials at Statistics Canada as well as the Interna-tional Labor Organization’s International Stan-dard Classification of Occupations. Canadianhealth occupations were grouped into the same 17categories as those used for the United States. In afew cases the Canadian classification conventionsappear to differ from those used in the UnitedStates. This is most evident in the assignment ofpersonnel to the occupational group “aides andother health service.” The Canadian census ap-pears to define these occupations more narrowlythan does the United States. Hence, many individ-uals classified under the rubric “all occupations,n.e.c. ” in the Canadian data would probably beclassified as “aides and other health service” un-der U.S. conventions.

Statistics Canada’s data classified the numberof hours worked as a range (e.g., 20-30 hours). Tocalculate FTEs, Himmelstein and colleagues as-signed each employee to the midpoint of the spe-cified range of hours (for the category >50 hours/week Himmelstein and colleagues assigned theemployee to 52.5 hours), multiplied by the num-ber of weeks worked during the year, and dividedby 2,000.

Himmelstein and colleagues calculated em-ployees and FTEs per million population usingthe Canadian resident population for each year asthe denominator.

andD Abbreviations

AMACEOCHFC

CNAM-TS

CNIL

COBRA

CPS

CPU

DHA

DHHS

FMF

FPC

GAO

GDP

701

American Medical AssociationChief executive officerCalifornia Health FacilitiesCommissionCaisse Nationale d’AssuranceMaladie des Travailleurs Salaries(National Insurance Association ForSalaried Employees, France)Commission Nationale deL’Informatique et des Libertes(France)Consolidated Omnibus BudgetReconciliation Act of 1985Current Population Survey (CensusBureau)Central processing unit(microprocessor)District Health Authorities(England)U.S. Department of Health andHuman ServicesFederation of Mutual InsuranceCompanies of FranceFamily Practice Committees(England)Full–time Equivalent

General Accounting Office (U.S.Congress)Gross domestic product

GPHCFA

HMOJAMA

JEIDA

KV

LPNNEC (n.e.c.)NEJMNHANHSOECD

OTA

PCMCIA

PPPProPAC

R&DRNSOCWEDI

General practitionerHealth Care FinancingAdministration (DHHS)Health maintenance organizationJournal of the American MedicalAssociationJapan Electronics IndustryDevelopment AssociationKassenartzliche Vereinigung(Germany)Licensed practical nurseNot elsewhere classifiedNew England Journal of MedicineNational health accountsNational Health Service (England)Organization for EconomicCooperation and DevelopmentOffice of Technology Assessment(U.S. Congress)Personal Computer Memory CardInternational AssociationPurchasing power parityProspective Payment AssessmentCommission (U.S. Congress)Research and developmentRegistered nurseStandard Occupational CodeWorkgroup for Electronic DataInterchange

References

1.

2.

3. .

4.

5.

6.

7.

8.

American Hospital Association, HospitalStatistics: 1990-91 Edition (Chicago, IL:199 1).Archer, B., “Administrative Efficiency of theU.S. Health Care System” (letter), New Eng-land Journal of Medicine 325(18):1316-1317, Oct. 31, 1991.Barer, M.L. and Evans, R. G., “InterpretingCanada: Models Mind-Sets and Myths,”Health Affairs 11(1 ):44-61, spring 1992.Blumenthal, D., “Administrative Costs andHealth Care Reform” (letter in reply), NewEngland Journal of Medicine 329(22): 1656-1657. Nov. 25, 1993.Blumenthal, D., “Administrative Issues inHealth Care Reform” (editorial), New Eng-land Journal of Medicine 329(6):428-429,Aug. 5, 1993.Danzon, P., “Hidden Overhead Costs: Is Can-ada’s System Really Less Expensive?,”Health Affairs 11(1 ): 21-43, spring 1992.Detsky, A. S., Stacey, S. R., and Bombardier,C., “The Effectiveness of a Regulatory Strate-gy in Containing Hospital Costs: the OntarioExperience, 1967-81 ,“ New England Journalof Medicine 309(3):151 -159, July 21, 1983.Evans, R. G., Barer, M. L., and Hertzman, C.,“The 20 Year Experiment: Accounting for,Explaining and Evaluating Health Care CostContainment in Canada,” Annual Review ofPublic Health 12:481-5 18, 1991.

9. Evans. R. G., et al., “Controlling Health Ex-penditures—The Canadian Reality,” NewEngland Journal of Medicine 320(9):571-577, Mar. 2, 1989.

10. Fuchs, V. R., and Hahn, J. S., “How Does Can-ada Do It? A Comparison of Expenditures forPhysicians’ Services in the United States andCanada,” New England Journal of Medicine323:884-890. Sept. 27, 1990.

11. Fuchs, V. R.. and Hahn, J. S., “Technical Ap-pendix to ‘How Does Canada Do It?.’” un-published manuscript, Stanford. CA, Nation-al Bureau of Economic Research, September1990.

12. Gauthier, A. K., et al., “Administrative Costsin the U.S. Health Care System: The Problemor the Solution,” Inquiry 29:208-320, fall 1992.

13. Glaser, W. A., Paying the Hospital (San Fran-cisco, CA: Jossey-Boss, 1987).

14. Glaser, W. A., Health Insurance in Practice(San Francisco, CA: Jossey-Boss, 1991).

15, Glaser, W. A., Professor, Graduate School ofManagement and Urban Policy, New SchoolFor Social Research, New York. NY. “Admin-istration in Health Care: A Plan for Cross-Na-tional Comparisons,” unpublished contractorreport prepared for the Office of TechnologyAssessment. U.S. Congress. Washington,DC, December 1993.

16. Grumbach, K.. et al., “Liberal Benefits, Con-servative Spending: The Physicians for a Na-tional Health Program Proposal.” Journal ofthe American Medical Association 265(19):2549-2554, May 15.1991.

17. Haber, S. G., et al., “Hospital expenditures inthe United States and Canada: Do HospitalWorker Wages Explain the Difference?.”Journal of Health Economics 11:453-465,1992.

| 71


18. Hahn, J. S., Assistant Professor, School ofPublic Health, University of North Carolinaat Chapel Hill, Chapel Hill, NC, “Administra-tive Costs in the Health Care Systems of theUnited States, Canada, Germany, and theUnited Kingdom: A Framework for Compar-ative Analysis,” unpublished contractor re-port prepared for the Office of TechnologyAssessment, U.S. Congress, Washington,DC, June 1993.

19. Hellander, I., et al., Administrative Waste inthe U.S. Health Care System in 1991: TheCosts to the Nation, the States, and the Dis-trict of Columbia (Washington, DC: PublicCitizen Health Research Group, 1991).

20. Himmelstein, D. U., and Woolhandler, S.,“Cost Without Benefit: Administrative wastein U.S. Health Care,” New England Journalof Medicine 3 11(7):441-445, Feb. 13, 1986.

21. Himmelstein, D. U., et al., Center for nationalHealth Program Studies, Harvard MedicalSchool/The Cambridge Hospital, Cambridge,MA, “The Health Care Labor Force: U. S.,Canada, and Western Germany,” unpublishedcontractor report prepared for the Office ofTechnology Assessment, U.S. Congress,Washington, DC, Mar. 19, 1993.

22. Hoffmann, E., Chief, Statistics of Employ-ment and Unemployment Section, Interna-tional Labor Office, Geneva, Switzerland,personal communication, Apr. 21, 1993.

23. Katz, R., Assistant Director, Human Re-sources Division, U.S. General AccountingOffice, personal communication, Apr. 21,1994.

24. Katz, S.J., Hislop, T. G., Thomas, D. B., Lar-son, E. B., “Delay From Symptom to Diagno-sis and Treatment of Breast Cancer in Wash-ington State and British Columbia,” MedicalCare 3 1:3 (March 1993) 264-268.

25. Lauve, R., and Long, H. W., “AdministrativeEfficiency of the U.S. Health Care System”

(letter), New England Journal of Medicine325(18):1317, Oct. 31, 1991.

26. Monod, E. M., “A Tour d’Horizon of HealthCare Cards in Europe,” Smart Card Technolo-gy International (London: Global ProjectsGroup, 1994).

27. Organisation for Economic Co-operation andDevelopment, OECD Health Systems—Vol-ume 1: Facts and Trends, 1960-1991, HealthPolicy Studies No. 3. (Paris, France, OECD,1993).

28. Organisation for Economic Co-operation andDevelopment, OECD Health Systems—Vol-ume 2: The Socio-economic Environment,Statistical References, Health Policy StudiesNo. 3. (Paris, France, OECD, 1993).

29. Paquel, N., Frizzole, C., and Glaziou, S., CA-NOPE Consulting, Paris, France, “SmartCards in the French Health Care System,” un-published contractor report prepared for theOffice of Technology Assessment, U.S. Con-gress, Washington, DC, 1993.

30. Poullier, J. P., Director, Education, Employ-ment and Social Affairs, Organisation forEconomic Co-operation and Development,Paris, France, personal communication, Aug.31, 1992.

31. Poullier, J. P., “Administrative Costs in Se-lected Industrialized Countries,” Health CareFinancing Review 13(4): 167-172, summer1992.

32. Redelmeier, D. A., and Fuchs, V. R., “HospitalExpenditures in the United States and Cana-da,” New England Journal of Medicine328:772-778, Mar. 18, 1993.

33. Rublee, D. A., “Medical Technology in Cana-da, Germany, and the United States,” HealthAffairs 8(3):178-181, fall 1989.

34. Schwartz, W. B., Mendelson, D. N., “Elimi-nating Waste and Inefficiency Can Do Littleto Contain Costs,” Health Affairs 13(1 ):224-238, spring 1994.

References 173

35. Schlesinger, M., Associate Professor, YaleSchool of Medicine, New Haven, CT, person-al communication, Aug. 15, 1994.

36. Sheils, J. F., Young, G. J., and Rubin, R. J.,”0Canada: Do we expect Too Much From itsHealth System?,” Health Affairs 11(1):7-20,spring 1992.

37. Sheils, J. F., and Young, G.J., ‘“NationalHealth Spending Under A Single Payer Sys-tem: The Canadian Approach—staff paper forLewin/ICF, Jan. 8, 1992.

38. Thorpe, K. E., “Inside the Black Box of Ad-ministrative Costs,” Health Affairs 11 (2):41 -55, summer 1992.

39. U.S. Department of Commerce, Bureau of theCensus, Data User Services Division, DataAccess and Use Staff, “Current PopulationSurvey, March 1992 Technical Documenta-tion,” unpublished report, Washington, DC,1992.

40. U.S. Department of Commerce, Bureau of theCensus, Demographic Surveys Division andPopulation Division, 1970-1980 OccupationClassification: CPS-Based 1970 Sort (Wash-ington, DC: 1984).

41. U.S. Department of Commerce, Bureau of theCensus, 1970 Occupation and Industry Clas-sification Systems in Terms of Their 1960 Oc-cupation and Industry Elements, TechnicalPaper No. 26, prepared by J.A. Priebe, J.Heinkel, and S. Greene (Washington, DC:1972).

42. U.S. Department of Commerce, Bureau of theCensus, The Relationship Between the 1970and 1980 Industry and Occupation Classifi-cation Systems, Technical Paper No. 59, pre-pared by J.A. Priebe, and P.L. Vines (Wash-ington, DC: 1989).

43. U.S. Congress, General Accounting Office,Canadian Health Insurance: Lessons for theUnited States, GAO/HRD-91 -90 (Washing-ton, DC: June 1991 ).

44. U.S. Congress, General Accounting Office,Canadian Health Insurance: Estimating

Costs and Savings for the United States,GAO/HRD-92-83 (Washington, DC: April1992).

45. U.S. Congress, House of Representatives,Committee on Ways and Means, Subcommit-tee on Health, Options for Health Insurance:Administrate\’e Simplification in Health Care,hearing. Apr. 2, 1992, Serial No. 102-90(Washington, DC: U.S. Government printingOffice, 1992).

46. U.S. Congress, Office of Technology Assess-ment. Protecting Privacy in ComputerizedMedical Records, OTA-TCT-576 (Washing-ton. DC: U.S. Government Printing Office,September 1993).

47. U.S. Congress, Office of Technology Assess-ment, Understanding Estimates of NationalHealth Expenditures Under Health Care Re-form, OTA-94-594 (Washington, DC: U.S.Government Printing Office, 1994).

48. U.S. Prospective Payment Assessment Com-mission, Medicare Prospective Payment andthe American Health Care System: Report tothe Congress (Washington, DC: U.S. Govern-ment Printing Office, June 1990).

49. Weil, T. P., and Miller, W. H., An Alternative inHealth Care Reform: The Facts on the Cana-dian Health System, manuscript (PhoenixAZ: Silver & Cherner, 1993).

50. White House Domestic Policy Council,Health Security: The President’s Report to theAmerican People (Washington, DC: U.S.Government Printing Office, October 1993).

51. Woolhandler, S., Director, Center for Nation-al Health Program Studies, Cambridge, MA,personal communication, June 8, 1994.

52. Woolhandler, S., Director, Center for Nation-al Health Program Studies, Cambridge, MA,personal communication, July 5, 1994.

53. Woolhandler, S., Director, Center for Nation-al Health Program Studies, Cambridge, MA,personal communication, Aug. 8, 1994.

54. Woolhandler, S., and Himmelstein, D. U.,“The Deteriorating Administrative Efficiency


of the U.S. Health Care System,” New Eng-land Journal of Medicine 324(18):1253-1258,May 2, 1991.

55. Woolhandler, S., Himmelstein, D. U., and Le-wontin, J. P., “Administrative Costs in U.S.Hospitals,” New England Journal of Medi-cine 329(6):400-403, Aug. 5, 1993.

56. Woolhandler, S., Himmelstein, D. U.,

“Correction: The Deteriorating Administra-tive Efficiency of the U.S. Health Care Sys-tem,” letter, New England Journal of Medi-cine 331(6):336, Aug. 4, 1994.

57. Workgroup for Electronic Data Interchange,WEDI 1992 Report (Washington, DC: 1992).

58. Workgroup for Electronic Data Interchange,WEDI 1993 Report (Washington, DC: 1993).

ti U.S. GOVNW2dll! NT PRINTING OFFICE; 19c)t4 - 301-90fI - 814/27415

international comparisons of administrative costs in ...studies are international comparisons of...

Documents