Archives ofDisease in Childhood 1996; 74: 268-273

CURRENT PRACTICE

Evaluation of appropriateness of paediatricadmission

Ursula Werneke, Roderick MacFaul

Rising demand upon limited healthcareresources has led to questioning of the extentto which these are appropriately used. One ofthe methods used to judge effective and effi-cient use of services is utilisation review.Utilisation reviews apply defined explicitcriteria and/or expert opinion (implicit criteria)to the hospital episode taking account of theprocess of care, decision making, site, fre-quency and duration of care. Information isderived from the record or interviews withstaff. They have their origin in the US in the1960s as one attempt to contain rising costsof state funded (Medicare/Medicaid) pro-grammes,' and were later used by healthinsurers.2 Since 1972, the US government hasrequired monitoring of appropriateness of careand length of stay during admission (concur-rent review) by a professional standard revieworganisation (PSRO).3 Utilisation reviewsinclude use of protocols for assessment ofappropriateness of admission. The best knownof these are the adult appropriateness evalua-tion protocol (AEP) and its derivation for pae-diatric services: the paediatric appropriatenessevaluation protocol (PAEP).45 These proto-cols rely on criteria independent of diagnosisand are based on levels of care given. They areapplied by a trained rater to samples of casenotes concurrently or retrospectively. If anyone of the many criteria is met for theadmission day or day of care, that day is ratedappropriate. The delay tool6 and the intensity-severity-discharge (ISD) protocol for paedi-atrics7 are alternative instruments. The delaytool, developed for use in adults, has also beenused on paediatric patients.6 8 It can be usedconcurrently or retrospectively and attemptsto detect, quantify, and assign causes formedically unnecessary hospital days. ISD cri-teria are based on objective clinical indicatorsand use of diagnostic and therapeutic serviceswhich reflect the need for hospitalisation. Thecriteria are categorised into organ systemstaking account of illness severity or into use ofspecial facilities (paediatric intensive care,special care, neonatal intensive care, or reha-bilitation). Also included are measures ofpatient stability indicating readiness for dis-charge and consideration of alternative caresettings. Although these protocols use objec-tive explicit criteria, they are screening toolsrather than being definite arbiters of appropri-ateness.9

Despite different medical culture and prac-tice in Britain, some North American auditmethods are being imported into this country.It seems timely therefore to examine howappropriateness of paediatric admission mightbe assessed.

Why evaluate appropriateness ofapaediatric admission?In paediatric practice in the UK, admissionshave risen considerably while the number ofbeds has dropped slightly over the past 10-15years. Substantial reduction of length of stayhas increased bed availability augmentingcapacity for admissions. Increased bed avail-ability allows greater throughput implying eco-nomic gain for a hospital in a market drivensystem. In the UK, however, with a managedmarket and budget capping, cost disadvantagemay arise if 'overtrading' is not followed byfunding to support it. Cost issues are not theonly concern in paediatrics as a child shouldnot be admitted to hospital unless this is neces-sary. Thus the question arises whether the cur-rent increase in paediatric admissions isjustified. Also in view of constraints imposedon service delivery by limited availability ofstaff and new technologies, there is a need toreview service provision as objectively aspossible.

Appropriateness implies that for the child'sillness or symptoms, a hospital admission,however brief, was the most suitable way inwhich to manage the problem at the timetaking into account the medical disorderswhich could be present and needed to betreated or excluded.10 A hospital will be theplace where the necessary nursing andmedical staff, equipment, and investigationfacilities are concentrated to enable a speedyand flexible response to be made appropriateto a child's actual or potential illness. Suchresources are usually concentrated to makebest use of them for the population served.While some specialist care can be provided athome especially for chronic illness or recoveryfrom acute problems, it is unrealistic to expectthe full range of specialist facilities to beprovided there in the early phase of commonacute illness. A decision to admit to hospitalwill involve balancing disadvantages of admis-sion (cross infection, painful and possiblyunnecessary procedures, iatrogenic accidents,

London HealthEconomicsConsortium, LondonSchool ofHygiene andTropical MedicineU Werneke

Department ofPaediatrics,Pinderfields GeneralHospital, AberfordRoad, WakefieldWF1 4DGR MacFaul

Correspondence to:Dr MacFaul.

268

Evaluation of appropriateness ofpaediatric admission

disruption of the family, or separation fromfamily and emotional upset to the child) withthe benefits (reduction of risk from the illness,easing of parental concern and anxiety, andan opportunity to enhance parental confi-dence in management of present and futureillness).

Appropriateness of paediatric admissionshould be assessed for a number of reasons: tojustify commitment and provision of resource(financial, facilities, or personnel); to identifyinefficient use; to identify the need for and typeof alternative configuration of service provi-sion; or, to identify those children who should,but do not, access the service. An 'audit tool'for appropriateness could be used in severalways:

(1) Aiding clinical judgment in the decisionabout whether a child should be admitted:an application not likely to be accepted byclinicians.(2) Identifying inappropriate use of hospitalfacilities on a day to day basis and ways ofpromoting earlier discharge by developingother services. Daily use, however, would belimited by the time required for review.Relevance to clinical practice may be con-strained by over simplicity.(3) Examining variation in appropriatenesslevels between units for comparison of dif-fering practice assisting analysis of need anddemand and planning of services. Areaswhere further services might be developed or

altered could be identified, such as changesin primary care or provision of facilities foremergency day case assessment, day case

surgery or elective investigation.

Methods of evaluating appropriatenessUtilisation reviews can use diagnosis basedcriteria taking account of the medical need foradmission and length of stay. In the UK healthcare resource groups based upon length of stayin regard to diagnosis allow such profile analy-sis but have major limitations for UK paedi-atric practice as length of stay differs littlebetween minor and complex cases. Tierney etal in the USA showed that the collection ofsimple clinical data may be as good a predictorof inpatient costs as discharge diagnosisrelated groups." Implicit criteria applying aclinician's own judgment to appropriatenessof care for individual cases, have been used byMacFaul et al 12 and in the British PaediatricAssociation (BPA) appropriateness of admis-sion study due to report this year. The validityof this technique depends entirely on knowl-edge, skills, and judgment of the reviewer andmay yield weak results in unstructuredapproaches.' Explicit criteria may be based on

diagnosis or on guidelines for specific cate-gories of patients. However, guidelines are

complex instruments, and their establishmentinto routine clinical practice represents a sub-stantial workload. Explicit criteria can also beindependent of diagnosis and based on typeand level of care criteria. All these approachesallow retrospective, concurrent, or prospectivereview.

Any assessment of inappropriateness of careshould take account of those children whorequire care but do not receive it. In the UK,there are opportunities to link hospital activityand utilisation rates with the socioeconomicprofile of the population to identify deprivedpopulations who do not appropriately accessthe service.

The PAEPThere is no current available 'gold standard'on which to establish the clinical validity of anaudit tool for judgment about whether or nota paediatric admission was appropriate. Notool exists for judging the appropriateness ofthe whole of a paediatric admission relevantfor the short length of stay in UK, and thetrue magnitude of inappropriateness is notknown. The PAEP was developed from theadult AEP in the US and has been appliedthere4 5 as well as in Australia,'3 Canada,'4-16and South Africa.8 The PAEP has 20 criteriafor assessment of admission day and 28 fordays of care. Meeting one of these criteriarates the day appropriate. The criteria wereprovided by consensus groups of clinicians.The PAEP was not designed to judge overalladmission, nor does it take account of theviews of the referring general practitioner orparents. Criteria in the current PAEPmanual'7 are based on those published byKreger and Restuccia,5 though modifiedslightly from the original publication.Admission day criteria are based on subsets ofclinical services representing treatment gener-ally available only in a hospital setting, and onpatient condition - either major physiologicalconditions or signs of acute illness which aresufficiently severe to justify admission to anacute hospital. For day ofcare criteria, the sub-sets are based on criteria for medical services,nursing or life support or patient condition.Some of the criteria are defined with ques-tionable relevance to clinical practice andothers based on pulse, respiratory rates, orlaboratory values appear overly rigid takinginsufficient account of the state of the childand management of health risks involved atthe time. Some of the criteria are shown intable 1 together with our comments.

Derivation of the paediatric criteria from theadult protocol may explain some of the weak-nesses and limitations of the PAEP. Some ofthe criteria also are defined only in terms rele-vant to American practice such as care in anemergency room: the UK functional equiva-lent probably being a standard or short staychildren's ward rather than an accident andemergency department.The role of the PAEP in UK practice is

uncertain and it has not yet been validated foruse in this country, although the current BPAappropriateness of admission study is doingthis. A modified form of the PAEP has, how-ever, been applied to 3000 paediatric admis-sions in one English region by Esmailwho found about 10% of admissions to beinappropriate (A Esmail, personal communi-cation).

269

Werneke, MacFaul

Table 1 Some examples of the admission and day of care criteria in the PAEP with our comments

Criterion Comment

(Admission day) criterion of acute confusional state,coma or unresponsiveness or (day of care) criterionof coma - unresponsiveness for at least an hour oracute confusional state.

Persistent fever greater than 37 8°C orally or 38 3°Crectally for more than 10 days.

Any of the following conditions not responding tooutpatient (including emergency room)management: seizures, cardiac arrhythmia, bronchialasthma or croup.

Vital sign monitoring every 2 hours or more often (mayinclude telemetry or bedside cardiac monitor).

IV medications and/or fluid replacement (does notinclude tube feedings) or in another section parenteraltherapy - intermittent or continuous IV fluid withany supplementation (electrolytes, protein,medications).

Chemotherapeutic agents that require continuousobservations for life threatening toxic reaction.

IM antibiotics at least every 8 hours.Intermittent or continuous respirator use at least every

8 hours. Or in another section respiratory care -intermittent or continuous respirator use and/orinhalation therapy (with chest physical therapy,intermittent positive pressure breathing) at least threetimes daily, isoetharine hydrochloride (Bronkosol)with oxygen, Oxyhoods, oxygen tents.

Close medical monitoring by a doctor at least threetimes daily (observations must be documented inrecord).

Continuous vital sign monitoring, at least every 30minutes for at least 4 hours.

These criteria are the only ones which appear to apply to a fit and it isnot clear whether this applies to all convulsions after which there hasbeen recovery. In our view, a young infant admitted after a firstfebrile convulsion would be an appropriate admission but admissionafter a convulsion in a child with epilepsy may not be.

This is an unusual problem in children in this country, yet many febrilechildren would be appropriately admitted for assessment muchearlier in the illness than this.

This could be interpreted to judge all admissions with asthma or croupas appropriate unless definition of 'not responding' is more precise.

In our view monitoring may be equally or more important whencarried out less often, and general continual observation by nurses isequally relevant.

Establishment of tube feeding, for example, in an infant with cerebralpalsy would in our view be an appropriate reason for admission.Equally a child with gastroenteritis admitted to hospital is morelikely to have IV fluids given than one equally unwell at home. Thus,an admission might be falsely rated appropriate by this criterion.

This criterion does not seem to apply to ingestion of substances with apotential for poisoning.

An unusual type of practice in UK.If this includes nebuliser use, then this could be self determining as a

criterion for appropriateness as hospitalised children with asthma aremore likely to receive nebulisers than at home and over estimation ofappropriateness result. This is also likely if nebulised or spacer plusmask therapy is included in this definition. The 1991 PAEP manualis not specific on this point and greater clarity is required.

No definition is given of 'close medical monitoring' other than that itmust be on three differing occasions. In our view, a child reviewedby an SHO/registrar in the afternoon after a consultant round in themorning would be appropriately admitted for that day but would notmeet the requirement in this criterion. Also the need forobservations to be documented in the record creates a problem.

It is not clear why this is so prescriptive. For instance, nursingobservation of feeding pattern or to determine whether child vomitsor has frequent diarrhoea or to see whether a child becomesdesaturated during feeding may make a day appropriate.

IM=intramuscular; IV=intravenous; SHO=senior house officer.

Previous studies on paediatricappropriateness(1) STUDIES USING THE PAEPThe earliest two published were those byKemper4 and by Kreger and Restuccia5 each ofwhom independently developed a PAEP fromthe adult protocol (the AEP) to take account ofpaediatric conditions and the protocols differslightly. Modified versions of these PAEP weresubsequently applied in Australia, Canada,and South Africa in the early 1990s and tables2 and 3 compare the main results. In each ofthese studies, the PAEP has been changedslightly and applied to differing groups ofpatients (mixtures of paediatric medical andsurgical, of acute and elective admissions, ofsecondary and tertiary cases and differing agegroups). The studies were mainly based in ter-tiary centres in which a significant proportionof admissions are elective, which have differentlevels of appropriateness from emergencycases. Application of the PAEP also differed inregard to day of care or admission and exclu-sion criteria. Kreger and Restuccia criticisedKemper for including the day of discharge,which is likely to be rated as appropriate on dayof care criteria due to doctor and nurse inputand in their PAEP the day of discharge isexcluded. Medical and surgical cases wereevaluated together, although their characteris-tics and management are so different that theyshould not be analysed as one sample.Recommendations for use vary: whereasKreger and Restuccia do not recommend the

use of the PAEP for patients under the age of 6months, three other studies include this agegroup.4 13 15 Gloor et al and Kemper foundfewer appropriate days with younger patientsbut Kreger and Restuccia did not report anysignificant difference. Direct comparison of thestudies for appropriateness by length of stay isimpossible as each uses different categories.The studies of the PAEP in North America

have reported good interobserver reliability butonly Smith et al have tested validity,16 Kemperet al used sensitivity and specificity and did notreport a validity exercise based on expertpanels.18 Audit tools should firstly have highreliability and, secondly, good levels of validity.Validity indicates whether the instrumentmeasures what it purports to measure, that itlooks reasonable, and samples the relevantcontent. For constructs which are not readilymeasurable such as appropriateness, validityhas usually been sought by use of expertpanels. Both reliability and validity can beexpressed in terms of correlation coefficients ofwhich the K coefficient measuring agreementbeyond chance is widely used. An agreement of0-75 or greater is regarded as excellent,between 04 and 075 as fair, and less than 04as poor. K Coefficients of between 046 and089 for reliability for the PAEP have beenreported (see table 3) and Smith et al reportfigures of validity of 0-68 for day of admissionand of 0-47 without override and 0-6 withoverride for day of care. Kreger andRestuccia's instrument was based on validity

270

Evaluation of appropriateness ofpaediatric admission

Table 2 Previous studies of the PAEP: compaison ofsamples

Agerange Length

Author Facility (years) Age detail ofstay Exclusion criteria Rater

Kemper (USA)4 Secondary and 0-18 Mean 7-8 years, 13% Median 7 Paediatric ICU, bum unit, Physician, nursetertiary care under 12 months days psychiatric and eating

disorders unit, adultservices

Kreger et al Secondary and 2-15 Mean 7-5 years Mean 4-6 All but Medicaid patients Nurses(USA)5 tertiary care days

Gloor et al Secondary and 0-19 25% Under 12 months Paediatrics and neonatal Physicians(Canada)15 tertiary care ICU, normal newborns,

psychiatric patientsFormby et al Secondary and 0-13 - - Rehabilitation care, mental Medical coder,

(Australia)'3 tertiary care disorders physiciansSmith et al Secondary and 0-5-18 Mean 6-8 years, median Mean 5-7; ICU, special care nursery, Physician, nurse

(Canada)'6 tertiary care 6 years, 9-4% under median care by parent unit,12 months 3 days psychiatric unit

Henley et al Secondary and - 46% Under 12 months, -

(South Africa)8 tertiary 36% between 1 and 5years

Kasian et al Secondary and 0-18 Mean 5-3 years, median - Special care, neonatal Medical student,(Canada)'4 tertiary 3 years, 23-8% under ICU, normal nursery, certified

12 months paediatric ICU, record analystpsychiatric admission

In Britain typically paediatric admissions have a mean age around 3 years, and median in the region of age 1-8 years, length ofstay has a mean around 2-5 days and a median of 1 day.ICU=intensive care unit.

values for the adult protocol. K Coefficientsdepend on and vary with the true prevalence ofthe factor studied. Assessment of medical andsurgical cases together will lead to false esti-mates of appropriateness and of K reliabilityand validity scores as both are likely to behigher when an operation has been performedfulfilling an easily identified criterion com-pared with assessment of the wide range ofproblems encountered in acute paediatricpractice. A special version of the adult AEP isavailable for elective surgical admissions.Most studies allowed an 'override' option to

the PAEP criteria, described by Smith et al asthe ability to 'override the assessment in eitherdirection if the rater considered that thecriteria based assessment did not accuratelycapture the clinical situation'. Utilisationreview instruments in general9 tend to over-

estimate inappropriateness and use of over-rides reduces the levels of inappropriateness byreverting to implicit criteria. They may be mis-used by inexperienced reviewers and the levelof training needed by reviewers is increased.

Apart from validity and reliability, sensi-tivity and specificity can be used to assess an

instrument. Kemper reported for her instru-ment a sensitivity of 093, that is 93% ofinappropriate hospital days were correctlyidentified as such, but specificity was only 078and thus 22% ofhospital days would be judgedas inappropriate though they would have beennecessary. Overrides used to decrease levels ofinappropriateness can increase specificity.Finally Strumwasser et al have shown thatvalidity scores depend on the setting andmedical culture to which the expert panelbelongs: scores being lower when a fee forservice instead of a prospective payment panelundertook the exercise.9 It is also possible thatthese scores will be higher in a research projectrather than routine audit.

Comparison of these studies reveals that nosingle PAEP standardised instrument has beenused, and it is not always clear which version ofthe PAEP has been used. Gloor et al 15 andKasian et al 14 each seem to have based theirreview on Kemper's version, Smith et al,'6Formby et al,13 and Henley et al 8 used Kregerand Restuccia's instrument. Thus the state-ment of Gloor et al that 'the PAEP and the AEPhave become a standard means of assessing the

Table 3 Previous studies of the PAEP: compaison of results

Inappropriate Inappropriate Inappropriate Inter-rateradmission Inappropriate admission days days of care Casemix (ad combine reliability

Author (days) days ofcare with override with override surgical+medical) (K coefficient)

Kemper (USA)4 - 21-4% of 1098 - - Medical 68% (for 0 74 for days oftertiary medical care26-9% IA)

Kreger and 10-5% of 793 13-3% of 648 5-8% 9-4% - 0-68 for admissionRestuccia day, 0-46 for(USA)5 days of care

Gloor et al - 23-9% of 852 - - General paediatrics Relies on Kemper(Canada)'5 29-5% of admissions values

of which 39% IAFormby et al 23-8% of 495 19-4% of 211 13-3% of 495 9 9% of 211 - 0 75 for admission

(Australia)'3 day, 0-33 fordays of care

Smith et al 22-5% of 477 22- 1% of 547 - - Emergency admission 0-89 for admission(Canada)'6 53% of which 14% day, 0 77 for

IA days of careHenley et al 2% of 171 20-5% of 365 Not used Medical includes 0 7 for admission

(South Africa)8 elective day, 0-68 forday of care

Kasian et al - 16-2% of 1327 - Medical 21-3% IA, -

(Canada)'4 surgical 977% IA

IA=inappropriate.

271

Werneke, MacFaul

need for acute-care hospitalisation in childrenand adults' remains debatable. Modificationsby the individual groups further complicatecomparison. The version used by Smith et al inCanada'6 has been applied in the BPA study ofappropriateness ofadmission in three Yorkshirehospitals and is being validated using expertpanels. The results will be available in 1996.Formby et al allowed greater medical overrideswhen a sudden deterioration in a patient waspossible. They suggested further modificationsof the PAEP would be necessary for use inAustralia.

Although differences in the versions of thePAEP are small, they can alter the meaning ofthe criteria. Those using the PAEP must beaware of the different versions used, as well asthe limitations and shortcomings of the PAEP,especially if any conclusions are to be based oncomparative data.

(2) STUDIES ON THE WHOLE OF AN ADMISSIONEarly attempts to evaluate the whole of anadmission instead of single days were made inthe 1970s in the USA based on subjectiveimplicit physicians' criteria and taking accountof the views of the parent (mothers wereunanimous that the admission was necess-ary),19 or also by defining objective criteria.20

In the USA in 1990, Soulen et al reportedan attempt to determine the proportion ofpotentially avoidable admissions takingaccount of social factors.2' This study alsotried to identify alternative services which ifavailable, might have led to saved hospitaldays. A closed ended questionnaire was used,developed by paediatric generalists andspecialists. The results were compared withdata from the monthly conducted hospitalutilisation review. The study identified 166 of600 medical admissions as potentially avoid-able (surgical cases and intensive care wereexcluded). The majority (138/166) of avoid-able admissions were judged suitable for careon a 24 hour short stay ward which impliesthat they would have been admitted in the con-text of current British practice where at least50% of admissions are for one day or less.Admission for eight of the 38 social admis-sions found in the 600 was judged unavoidablebut others could have been prevented by pro-vision of outpatient care, home nursing, ortransport. The utilisation review found onlytwo out of 600 admissions medically unneces-sary. The authors concluded that admissionswere necessary in 100% given availableresources. They felt, however, that physicianjudgment was influenced by so many complex,subjective, and discretionary factors that itcould not stand alone as a reliable and validmeasure of necessity, but on the other hand,suggested caution in using objective criteriaalone to categorise hospital use as medicallynecessary.

In an Australian children's hospital 87-7%of a sample of admissions of under 24 hourswere found to be appropriate using criteriapartly subjective but largely based on thePAEP.22

In the UK, auditing of paediatric admissionis a relatively new field. In 1991 Rajaratnamreported a study in Cardiff when two paedi-atric consultant raters using subjective(implicit) judgment on admission notes found15% and 20% respectively of 620 admissions(which included elective ones) not to berequired and 9% and 3% to be purely social.23However, the level of agreement between theraters was poor with a K coefficient of 0'37. Hesuggested the need for the assessment ofappropriateness to be a routine part of audit ofpractice. In 1993, MacFaul et al reportedassessment of appropriateness of paediatricmedical admission based on a combination ofexplicit and implicit criteria using consultantjudgment on the discharge of the child.'2Parent's views were sought, and comparisonsmade between the consultant's, admittingjunior doctor's, and the parent's views. Of 267consecutive admissions, 19 5% were judged bythe consultant on discharge not to be needed.

An appropriateness protocol for UKpaediatric practice?The PAEP has a number of drawbacks for usein acute general paediatrics in the UK. It is notclear whether this tool studies what it sets outto do and if it truly reflects inappropriate orappropriateness. Little account seems to havebeen taken of sensitivity and specificity and theeffect of misclassification.24 No account istaken of whether the admission was regardedby the clinical team as medically necessary.Gloor et al 15 and Smith et al 16 state that theunderlying assumption when applying thePAEP is that the decision made for medicalpatient care is correct, although that is whatthe PAEP tries to determine using servicerather than diagnostic criteria. In our view, itoffers an overly mechanistic approach that isconfined to the moment rather than taking intoaccount the potential consequences of thejudgment, and despite the use of overridesmaking some allowance for this drawback,most of the studies express reservations abouttheir use. Soulen et al give emphasis to poten-tial overtreatment: '... every admission that isavoided reduced potential morbidity, mortal-ity, and cost created by nosocomial infectionsand iatrogenic complications'.2' However, ifan admission, which is truly necessary isavoided, potential morbidity and mortalitycould result outweighing the risk of unneces-sary admission. The PAEP user manualacknowledges this,'7 and Smith et al statethat appropriateness of 100% would beunrealistic.'6 The PAEP and similar instru-ments may not give sufficient attention tothe dynamics of illness especially in acuteconditions.

In judging appropriateness, account shouldbe taken of the potential clinical disordersassociated with a presenting illness in a child.In acute childhood illness especially in infantsand younger children, distinction of a mildfrom severe illness is often difficult. In the earlyphase, many conditions present with the samesymptom complex, and underestimation of

272

Evaluation of appropriateness ofpaediatric admission 273

severity may lead to adverse outcome.Paediatric observation, investigation, andtreatment may often be appropriate at the timeof admission in what turn out to be minor con-ditions. The published studies of the PAEP inNorth America and Australia have mainly beenin mixed secondary and tertiary centres wherepatients were older and length of stay longerthan is usually found in UK paediatric practice(see table 1). Thus they have limited relevanceto UK practice where at least 80% of paedi-atric admissions are to secondary care unitswith over 90% being emergency and over 50%staying for less than 24 hours so that PAEP dayof care criteria would not be applicable. Forthe UK, it may be more relevant to have anaudit tool which will assess the appropriatenessof the whole of an admission rather than toexamine whether the child should be inhospital on any given day. Nevertheless thePAEP is likely to be of value for elective or ter-tiary paediatric work or for children's surgicaladmissions - albeit with some modifications.The paediatric ISD criteria7 appear moreclinically relevant than the PAEP: for example,severity of illness criteria include presentingproblems. However, application of the ISDappears complicated with, for instance, addi-tion of differing clusters of criteria for differentbody systems to determine appropriateness.Appropriateness may also be rated if there is anaggregate of marginal criteria. If the criteria,which are applied concurrently before, at, orafter admission, are not met additional infor-mation may be sought from the physiciancaring for the child which may then make theadmission appropriate. An appeal procedure toa third physician not involved in the care isadvised if the caring physician disagrees withthe application of the criteria.

Utilisation reviews were developed in theUSA on the basis of cost justification and in adifferent financial culture. However, vindica-tion of resource allocation is equally importantin a public funded NHS which needs to makethe most efficient and cost effective use ofresources. Account should also be taken of theneed to minimise disruption and risk tochildren and families from acute illness,including reduction of emotional and financialstress, and indirect costs such as travel and lossof time from work should be considered.However, whether ambulatory care options aremore cost effective than conventional servicesremains debatable and an audit tool to evaluatethis practice would be useful. A short admis-sion to a unit sensitive to the child's andfamily's needs, could well prove to be the mostbeneficial and least costly of the variousoptions.

Admissions that would be preventable maybe necessary because social circumstances couldnot be changed at the time. Social remedies arenot likely to be immediately achievable, may becostly or unacceptable to a family (for example,putting in a foster mother or nurse overnight),

and not permanent. Such changes also lie out-side the healthcare responsibility.The BPA has received health department

funding to develop an appropriateness ofadmission protocol for evaluation of the wholeof an admission. The protocol has been devel-oped using a new approach not modifying anexisting tool. It will take account of thepresenting problem, clinical state and diagnosisin the child, and also of the social and primarycare factors affecting the management of the ill-ness. It aims to be suitable for use in accidentand emergency departments or even in primarycare. Consensus panels of paediatricians, acci-dent and emergency specialists, and generalpractitioners have helped to form the criteriaand to validate the protocol which should beavailable for general use early in 1996.

1 Payne SMC. Identifying and managing inappropriatehospital utilisation. Health Serv Res 1987; 22: 709-69.

2 Wickizer TM, Wheeler JRC, Feldstein PJ. Does utilizationreview reduce unnecessary hospital care and containcosts? Med Care 1989; 27: 632-47.

3 Westphal M, Frazier E, Miller MC. Changes in averagelength of stay and average charges generated following insti-tution of PSRO review. Health Serv Res 1979; 14: 253-65.

4 Kemper KJ. Medically inappropriate hospital use in a pedi-atric population. NEnglJMed 1988; 318: 1033-7.

5 Kreger BE, Restuccia JD. Assessing the need to hospitalizechildren: pediatric appropriateness evaluation protocol.Pediatrics 1989; 84: 242-7.

6 Klein JD, Beshannsky JR, Selker HP. Using the delay tool toattribute causes for unnecessary pediatric hospital days.Med Care 1990; 28: 982-9.

7 Inter Qual. Pediatric ISD criteria and review system. 293Boston Post Road West, Suite 180, Marlborough, MA01752, USA: Inter Qual, 1993.

8 Henley I, Smit M, Roux P, Zwarenstein M. Bed use in themedical wards of Red Cross War Memorial Children'sHospital, Cape Town. SAfrMedJ 1991; 80: 487-90.

9 Strumwasser I, Paranjpe NV, Ronis DL, Share D, Sell IJ.Reliability and validity of utilization review criteria. MedCare 1990; 28: 95-111.

10 Wise PH, Eisenberg L. What do regional variations in therates of hospitalisation really mean? N Engl J Med 1989;320: 1209-11.

11 Tierney WM, Fitzgerald JF, Miller ME, James MK,McDonald CJ. Predicting inpatient care with admittingclinical data. Med Care 1995; 33: 1-14.

12 MacFaul R, Glass EJ, Jones S. Appropriateness of paedi-atric admission. Arch Dis Child 1993; 71: 50-8.

13 Formby DJ, McMullin ND, Danagher K, Oldham DRA.The appropriateness evaluation protocol: application inan Australian children's hospital. Aust Clin Rev 1991; 11:123-31.

14 Kasian GF, Zinkiew K, Senthilselvan A. Inappropriatehospital bed days at a Canadian pediatric tertiary carecentre. Annals of the Royal College of Physicians andSurgeons ofCanada 1992; 25: 275-8.

15 Gloor JE, Kissoon N, Joubert GI. Appropriateness ofhospi-talisation in a Canadian hospital. Pediatrics 1993; 91:70-4.

16 Smith HE, Sheps S, Matheson DS. Assessing the utilisationof inpatient facilities in a Canadian pediatric hospital.Pediatrics 1993; 92: 587-93.

17 Utilisation Management Associates Inc. Appropriatenessevaluation manual. Boston: Utilisation ManagementAssociated Inc, 1991.

18 Kemper KJ, Fink HD, McCarthy PL. The reliability andvalidity of the pediatric appropriateness evaluation proto-col. Qual Rev Bu1l 1989; 15: 77-80.

19 Lovejoy FH, Carper JM, Janeway CA, Kosa J. Unnecessaryand preventable hospitalizations: report on an internalaudit. J Pediatr 1971; 79: 868-72.

20 Duff RS, Cook CD, Wanerka GR, Rowe DS, Dolan TF.Use of utilization review to assess the quality of pediatricin-patient care. Pediatrics 1972; 49: 169-76.

21 Soulen JL, Duggan AK, DeAngelis CD. Identification ofpotentially avoidable pediatric hospital use: admittingphysician judgment as a complement to utilization review.Pediatrics 1994; 94: 421-4.

22 Numa A, Oberklaid F. Can short hospital admissions beavoided? Med J7Aust 1991; 155: 395-8.

23 Rajaratnam G. A study of admissions to paediatric beds.PostgradfMedJf 1991; 67: 50-4.

24 Phelps GE. The methodological foundations of studies ofthe appropriateness of medical care. N Engl Jf Med 1993;329: 1241-5.