QoC: Quality of Care

Robin B. Lake, Ph.D. lake@apk.net

QoC Seminar Objectives

• An appreciation of Quality• One way to approach a problem, given a set of tools, data, and

resources• The value of innovation and risk-taking• The value of UNIX, C, and structured design

• If you can wait long enough, your work may be appreciated …

• Notes and references should be at:– hal.case.edu/~rbl

Why Another Quality Measure?• When data is incomplete, inaccurate, or inconsistent, classic

statistical methods often are unable to cope properly• Classic statistical results are then misleading• QoC represents a ground-up re-engineering of Quality

Measurement– Developed for health care, where medical records are

incomplete, inaccurate, inconsistent and uncertain (but the method works more generally)

– Proven approach for assessing and managing risks, uncertainties and unexpected events

– “You Can’t Manage What You Can’t Measure”

QoC’s Components• A Method (Quality Algebra™)

Provides a quantitative measure to help assure Quality

• A System

Yields a valid measure of the Quality of Care for each case in a given DRG and models to assist in analysis and to provide accountability for quality variations.

• A Follow-on Process

Interpretation to identify specific problems in the delivery system, enabling management to control Quality which has proven to both reduce costs and improve Quality of Care

The QoC Process is Based on

Feedback of Information• Feedback Provides Quality Accountability and Assurance

– Control theory proves that each time a provider’s records pass through the QoC System the Quality of Care is more certain.

QoC Method Measured QualityCare Records

Identified ProvidersQoC Process

Focus on Specific Providers

QoC: The Origins

• The Context– Biometry Computer Lab

• The Opportunities– Geriatric Study

– Incomplete, Inaccurate, Illogical Data

– Ohio Area XI PSRO• Darran N. Huggins, M.D., M.S.

• The Context II– UNIX, C, Software Tools

CWRU Culture … Then

1970

The Biometry Computer Lab

Algebras

• X + 2y = z• Fuzzy Algebra• Boolean: [ a NAND b ]• Particle Physics:

W. B. Vasantha Kandasamy

2003

Fundamental Axioms of Algebra

Axiom of Choice• AC, the axiom of choice, because of its non-constructive character, is the

most controversial mathematical axiom, shunned by some, used indiscriminately by others. This treatise shows paradigmatically that: - Disasters happen without AC: Many fundamental mathematical results fail (being equivalent in ZF to AC or to some weak form of AC). - Disasters happen with AC: Many undesirable mathematical monsters are being created (e.g., non measurable sets and undeterminate games). - Some beautiful mathematical theorems hold only if AC is replaced by some alternative axiom, contradicting AC (e.g., by AD, the axiom of determinateness). Illuminating examples are drawn from diverse areas of mathematics, particularly from general topology, but also from algebra, order theory, elementary analysis, measure theory, game theory, and graph theory. [Horst Herrlich, “The Axiom of Choice”, Springer]

• (ZF = Zermelo-Fraenkel Set Theory)

Readable Reference

Quality Algebra™

• Just as computer logic has its own algebra (Boolean Algebra) and atomic physics has its own (Quantum Algebra), we have developed a new mathematics specifically for the assessment, management and control of Quality (Quality Algebra)

• Allows for uncertain data, yet provides a valid, certain measure of Quality

From Concept to Working System• We have the Concept

– We believe we have produced a measure for quality of medical care --- that is coherent, logical, quantitative, adaptive to peer standards, and can be normalized to allow inter-disease comparisons. We also believe that we have embedded this quality measure within a constructive, non-threatening system to help assure regional health care quality.

Implementing QoC

–Resources»UNIX - Pipes and Filters»The C Programming Language»“Software Tools” Software Toolkit

–Dataflow Design–Test: 93 Reports

in 10 days

Apply the Measure• The PPRO selected each review focus, based on patients whose

medical records indicated a specific surgical or medical procedure (Diagnostic Related Group). For each of nine Diagnostic Related Groups (DRGs), medical records were abstracted from as many as 500 patients from 25 hospitals. Private and third-party pay cases were collected along with the legally authorized Medicare/Medicaid cases by prior agreement with the hospitals and providers.

• Quality Score for each Case– Quality Score for each Provider

• Quality Score for each Institution

DRGs Studied• Inguinal Herniorraphy• Cholecystectomy• Non-operative Ulcer• Lens Extraction• Femoral Neck Fracture• Acute Myocardial Infarction• Essential Hypertension• Congestive Heart Failure• Transuretheral Prostatectomy• Also: Diabetes; Pneumonia

Data Elements• DRG-specific Elements (48 - 50 per study)• Surgery Time• Anesthesia Time• Surgeon Demographics (Age, Board Status)• Where Patient Discharged To• Other Diagnoses• Other Procedures Performed• Patient Demographics• Payment Source• 300+ Care Elements (Action, Date, Result)

Analyses

• Quality of Care for Each Patient• Quality of Care for Each Provider and

Institution• Models Using Stepwise Linear Regression

– Lengths of Stay: Pre-op; Intra-op; Post-op– Mortality– Disability upon Discharge

Inguinal HerniorraphyPost-Operative Length-of-Stay N=244

• Post-Op LOS = 5.72 days • + 0.0201 days per minute of anesthesia time (p = 0.000000)• - 43.3 days per Overall Quality Score value (p = 0.000000) • - 1.53 days if Private Pay Source (p = 0.000000)• + 0.0446 days per year of Surgeon Age• + 0.72 days if surgeon is Board Certified • + 0.2338 days per year of Patient Age

• This model says, for example, that every extra minute of anesthesia time would predict an additional 0.02 days of post-operative stay. It also says that increasing quality (as measured by QoC's Quality Algebra™) certainly significantly reduces the post-operative length of stay.

Provider and Case QualityInguinal Herniorraphy

Effective Results• Federal Medicare reimbursement law changed.• 13% pre-admission testing in north central Ohio rose

to 50%• If projected to all conditions: 8.7 million discharges

65 and over in the US in 1980.• Assume 50% elective admissions where pre-

operative testing could be done,and allow for 1 day reduction in in-patient care at $300 per day,then savings are 0.37 * 0.5 * $300 * 8.7 million = $482,000,000 per year.

Mortality ModelFemoral Neck Fracture N=351

• Death = -0.03• + 0.168 If Abnormal Blood Gas pH (p=0.000000)• + 0.00781 Times Pre-operative Length-of-stay in days (p = 0.000000)• - 0.00409 Times Post-operative Length-of-stay in days (p = 0.000000)• + 0.116 If Other Surgical Procedures Performed (p = 0.000000)• + 0.0994 If Operative Site Culture Performed (p = 0.000000)• + 0.0533 If Elevated Blood Pressure (p = 0.000000)• + 0.00151 Times Surgeon Age in Years (p = 0.000000)• + 0.0628 If Elevated Blood Urea Nitrogen (p = 0.000000)• - 0.00050 Times Length of Surgery in minutes (p = 0.000000)• - 0.09527 If Elevated Creatinine (p = 0.000000)• + 0.00163 Times Patient Age in years (p = 0.000000)• - 0.487 If Abnormal Prothrombin Time (p = 0.000000)• + 0.118 If Final Fever and No Antipyretics (p = 0.000010)• + 0.0880 If Sputum Culture Performed (p = 0.000167)• + 0.0326 If High Serum Sodium (p = 0.000935) In addition to the acute adult respiratory distress suggested by the Abnormal Blood Gas pH, there are clear

indications of the risks associated with post-operative wound infections, uncontrolled abnormal blood clotting, and perhaps untreated fever.

ResultsFemoral Neck Fracture

• The first model, that mortality, presented the unusual finding that the most significant predictor of death was Abnormal Blood Gas pH. Further investigation of the literature led the PSRO Medical Director and then the Peer Review Committee to conclude that this finding was an indicator of acute adult respiratory distress. This respiratory distress could be logically attributed to either the patient's reaction to the shock of the fracture or possibly to fat emboli entering the blood stream at the site of the fracture. In either case, the conclusion was that certain patients were metabolically compromised by the acute respiratory distress, yet no actions were taken (as indicated in their medical records) to stabilize these patients metabolically before subjecting them to anesthesia.

• With this unusual finding in hand from 25 non-teaching hospitals, we approached the Chairman of the Department of Orthopedic Surgery at a major university teaching hospital. Intrigued by the finding, he allowed us to collect the past ten years of death records from the Department's practice.

• Our model of the teaching hospital again clearly indicated that Abnormal Blood Gas pH was a significant predictor of mortality.

• Although there was a clear and easily remedied problem in the health care delivery structure --- one which extended beyond the north central Ohio area --- by the time the PSRO review activities had ended there were no noticeable changes in policy to encourage a respiratory consult prior to anesthesia for femoral neck fracture patients.

• Based on the death rates observed in Ohio Area XI hospitals and confirmed at the university teaching hospital, we project that 1880 lives could be saved each year through assuring that femoral neck fracture patients are stabilized with respect to acute adult respiratory distress before surgery.

The Potential of QoC

• Concurrent Review– While the Patient is in the Hospital

• Concurrent Concurrent Review– As the Order is Issued

• Cost• Risk• Quality

The Road Ahead

• Need for an effective system to implement findings

• A natural for EMR systems, but:–Software vendors don’t want to change–Institutions seem to be reluctant to go

beyond the vendors–May threaten existing power structures

Additional Materials

• http://hal.cwru.edu/~rbl• Three draft papers• References