Quality in Health Care 1994;3:53-57

Measuring handicap: motives, methods, and amodel

Rowan H Harwood, Sutthichai Jitapunkul, Edward Dickinson, Shah Ebrahim

Medicine is concerned with the manifestationsand consequences of disease, but patients'views on how ill health affects them may differfrom those of their doctors.' 2 The increasingimportance of chronic diseases, and the needto accommodate patients' views, haveprompted a reassessment of the way weconsider the effects of disease. Most patientsnow seen in primary care and hospital havechronic, irreversible diseases that do notalways lead to immediate or inevitable deathbut cause disability and disadvantage.'Modem medicine has developed excellenttechnologies to measure the immediateconsequences of disease, but our ability toappreciate and quantify the wider aspects isless advanced.4 Such measures are needed ifwe are to be able to assess the outcomes ofhealth care.

A framework for measuring healthThe International Classification of Impairments,Disabilities and Handicaps (ICIDH)5 hasattempted to clarify the consequences ofchronic disease by offering a new taxonomy.An impairment is any loss or abnormality ofpsychological, physiological, or anatomicalstructure or function (for example, shortnessof breath or weakness in a limb). A disability isany restriction or lack of ability to perform atask or activity (such as walking, dressing ormaintaining continence). Handicap is thedisadvantage suffered by an individual as aresult of ill health, due to the inability to fulfila role which is normal for someone of that age,sex, and culture.5 Doctors are used to definingimpairments as part of history taking andexamination, but disabilities and handicapsare often of greater concern to patients.

Department of PublicHealth and PrimaryCare, Royal FreeHospital MedicalSchool, LondonNW3 2PFRowan H Harwood,MRC health servicesresearch training fellowSutthichai Jitapunkul,British Council fellowEdward Dickinson,senior lecturerShah Ebrahim, professorCorrespondence to:Dr Harwood

Why measure handicap?The main motive for measuring handicap is touse it as an outcome measure. This is requiredfor trials of new therapies in chronic disease(drugs, operations, physiotherapy techniques,specialist rehabilitation units), for studies ofcost effectiveness, and for evaluating clinicalservices (for audit, for assessing serviceinnovations, or in monitoring contracts forhealth care provision, where a quantitativeoutcome assessment is required).

Defining impairments and disabilities isuseful in diagnosis and understanding how illhealth has its impact on someone's life. Anaccurate diagnosis may allow a specific therapyto be implemented. Measuring impairmentsallows the natural history of an illness orresponse to therapy to be assessed (such as

limb weakness after a stroke or peak expiratoryflow rate in obstructive airways disease). Whenthere is no specific therapy, assessing disabilitycan lead to rehabilitative interventions such asteaching someone with a hemiplegia to dressor manoeuvre a wheelchair. However, impair-

* . . impairments and disabilitiesare incomplete and inadequateoutcome measures in chronic

disease.

ments and disabilities are incomplete andinadequate outcome measures in chronicdisease. Some apparently trivial impairmentsmay result in major handicap (for example, afacial scar may lead to intolerable feelings ofinadequacy, depression, and social isolation).Despite considerable work on devising andvalidating disability scales,6'-0 their use inclinical trials has sometimes failed to showimprovement where clinical or other evidencesuggested that benefits should be found. '1-3Assessments of impairment, disability, andhandicap may be conflicting; in heart failurethere are poor correlations between measure-ments of cardiac output (an impairment),treadmill exercise tolerance and timed walkingtests (disabilities), and customary mobilitymeasured by pedometer (an aspect ofhandicap).'4 The lack of a direct relationbetween impairment, disability, and handicapalso applies in chronic airways disease. 156Finally, from the patient's point of view,impairments and even disabilities matter lessthan their effect on everyday life.Some outcome measures have tried to

embrace wider concerns. Measures ofperceived health or "quality of life" have beenused with some success 7- '; examples includethe Nottingham health profile, sickness impactprofile, quality of wellbeing scale, andMcMaster health index.20 Defining "quality oflife" and "health" precisely is difficult, butthese scales generally contain items aboutsymptoms, feelings, attitudes, impairments,disabilities, and handicaps. For example, thesickness impact profile has 136 items in 12categories including eating, ambulation,recreation, and social interaction. TheNottingham health profile has 38 items in sixdimensions (mobility, pain, emotion, socialisolation, sleep, and energy) in part 1, and asksabout problems in seven areas of daily life in

53

on April 4, 2021 by guest. P

rotected by copyright.http://qualitysafety.bm

j.com/

Qual H

ealth Care: first published as 10.1136/qshc.3.1.53 on 1 M

arch 1994. Dow

nloaded from

http://qualitysafety.bmj.com/

HJit1\ goods, -7itapunikul, Dickin.7on, bbral'il

part 2. The quality of wellbeing scale hasdimensions of mobility, physical activity, andsocial activity and a checklist of symptoms.Sensitivity to change as a result of inter-ventions will depend on the extent to whichthe items measured are relevant to the type ofintervention (and hence how many of theitems are relevant). This has been shown to bepoor in some cases.

Quality of life is influenced by many factorsother than health - for example, relationships,wealth, interests, and employment. A study ofthe impact of a health promotion and exercisepackage for older people failed to demonstratechanges measured by the Nottingham healthprofile, but mean scores were twice normativevalues, reflecting the inner city residence of thesample studied. Probably any quality of lifebenefits from the health promotionprogramme were trivial compared with theadversity associated with socioeconomicdeprivation.A further problem of some multi-

dimensional quality of life indicators is thatassociated with aggregation of the data into asingle index. When using the Nottinghamhealth profile results from each of sixdimensions and seven questions have to beconsidered separately, and no means ofpooling the information has been developed;thus the interpretation of inconsistent findingsbetween dimensions is impossible. Thesickness impact profile generates an overallscore simply by adding the scores of individualdimensions. The quality of wellbeing scale ismore advanced in weighting responses on thethree dimensions and the symptom list beforeadding them.Although alleviating some impairments

(pain, nausea, anxiety) or disabilities (inconti-nence) may be a valid objective of healthcareinterventions in its own right, impairment anddisabilities will often have handicaps associ-ated with them, such as limited mobility orphysical independence due to pain orrestriction of occupation and social integrationdue to anxiety or incontinence. The reversal ofimpairments and the reduction of disability areimportant strategies to explore in rehabili-tating people with chronic disease. However,many diseases (for example, stroke, dementia,and arthritis) are associated with irreversibleimpairments and disabilities. Measuringreduction in disability (strictly defined interms of the ability to perform tasks or activities)takes no account of what someone actuallydoes in their everyday life. If someone cannotwalk they may benefit from being taught howto use a wheelchair. However, the ability to usea wheelchair is irrelevant if it isn't actuallyused. This will depend on factors likemotivation (does the subject want to use it?),physical environment (unadapted steps orstairs?), relationships (is there someone topush?), and resources (an electric-poweredwheelchair, an adapted car or flat). Someabilities may be unimportant or irrelevant to agiven individual. The Barthel index" includesitems on climbing stairs and bathing, butclimbing stairs may be of no concern to

someone who lives in a house without stairsand who has no desire to go anywhere whichhas and the inability to use the bathindependently may be of little additionalconsequence to someone who also needswashing and dressing. Rehabilitation has notnecessarily failed where disabilities areirremediable. Reducing handicap covers awide range of rehabilitative interventions,including resettlement in appropriateaccommodation; providing services such ashome nursing, domestic help, and day centres;and psychological adjustment to illness.Reducing handicap must be the criterion bywhich intervention at any level is judged. Aswith quality of life scales, measures that fail toseparate out the different consequences of illhealth lose precision as some items will beredundant or inappropriate.

Current methods of measuring handicapAs described, the relation between impair-ment, disability, and handicap is not straight-forward.' 15 Similar degrees of impairment canlead to different disabilities and similardisability produces different levels ofhandicap. 2 I l i Thus we cannot infer levelsof handicap from information on impairmentsor disabilities.

.we cannot infer levels ofhandicap from information onimpairments or disabilities.

The ICIDH is a classification, not ameasurement scale. Its handicap sectiondefines six basic "survival roles," whichdescribe disadvantage in orientation (theability to perceive and understand theimmediate environment including sight,hearing, and cognition), physical indepen-dence (from human or mechanical assistance),mobility (the distance one can move fromone's bed), occupation (employment,domestic work, and recreation), socialintegration, and economic self sufficiency(including the ability to earn an income andthe possession of resources enabling problemsto be overcome). These survival roles may bethought of as dimensions into which morecomplex roles (such as professional, parent, orcitizen) can be resolved. Handicap is classifiedaccording to the disadvantage associated withdeficiencies in each dimension. An example ofdifferent levels of disadvantage is evident forthe dimension mobility: bedfast or chairfast,roombound, housebound, confined to im-mediate neighbourhood, unlimited mobility.The emphasis is not on how you get there buton how far you get in practice.We identified four putative handicap

measures: the modified Rankin scale,'2 theJefferys measure, the functional autonomymeasurement system (systeme de mesure del'autonome functionelle (SMAF),>" and theEdinburgh rehabilitation status scale." None

54

on April 4, 2021 by guest. P

rotected by copyright.http://qualitysafety.bm

j.com/

Qual H

ealth Care: first published as 10.1136/qshc.3.1.53 on 1 M

arch 1994. Dow

nloaded from

http://qualitysafety.bmj.com/

Measuing a handicap: motives, methods, and a model

of these instruments proved entirely suitablefor measuring handicap. The Jefferys scale,27used in the 1971 Office of Population,Censuses, and Surveys handicap survey30 andin community surveys in Canterbury,23measures disability. The other three useICIDH definitions and attempt to measureglobal handicap. The SMAF estimateshandicap by identifying disabilities which, arenot compensated for by aids or assistance.This is an attractive approach but remainslimited by some of the criticisms of disabilityscales. Moreover, compensating disabilities isonly one of the ways in which disadvantagecan be reduced. Provision of health serviceresources tends to be concentrated on thosepeople with moderate or severe physicaldisability, but the impact of these resources inalleviating disability (as opposed to handicap)may be very limited. Availability, acceptability,and accessibility of services are also important.Therefore, this instrument may not be suitablefor comparing populations which havedissimilar service systems, such as differingemphases on institutional and domiciliarycare. Although the "modified Rankin scale"26was derived from the original Rankin disabilityscale25 by introducing the phrases "interfereswith life style" or "restriction of life style" insome grades, other descriptions are still ofdisability rather than handicap. It covers onlythe physical independence dimension ofhandicap in the ICIDH classification, andwith only six points, while simple, is relativelycrude. The Edinburgh rehabilitation statusscale describes itself as a "measure of medico-social dysfunction", and has dimensionscovering dependency, activity, social inte-gration, and "the effects of symptoms onlifestyles".29 Activity and social integrationclosely reflect ICIDH roles, but the others aremore global. No attempt was made to weightitems on the scale; the scaling is thus arbitrary.Nevertheless, some evidence was providedthat it was sensitive to changes occurringduring rehabilitation.Another group of measures seek to assess

handicap due to specific impairments ordisabilities. Assessment of auditory handicapis most advanced, with several valid andreliable measures [for example reference 31].All include items that would be classified asdisability under the ICIDH. This criticism canalso be levelled at specific handicap measuresrelating to dental health32 and the impact ofdizziness.33 Some other measures allude tohandicap, such as the New York HeartAssociation classification34 and the Rosser andKind quality of life index disability subscale.35Both of these are short and rather limited inscope.

A new model for measuring handicapTo quantify handicap three questions must beaddressed, as follows36:* How are states of health to be described?* How are different health states to be valued

relative to each other?* How are individuals' values to be

transformed into a group value?

The system for describing different states ofhealth needs to be comprehensive enough tocover all possible states incurring disadvan-tage, sensitive enough to allow meaningfulchanges to be measured, yet clearly enoughworded to achieve acceptable reliability. Theuse of the "survival roles" described in theICIDH as different dimensions to describehealth states seems particularly appropriate.By categorising disadvantage on each of thesedimensions an individual subject's handicapcan be described.To determine how different health states

should be valued, and how results should beaggregated, we must examine the origins ofhandicap. An individual's role in life isinfluenced by many factors other than abilitiesor disablities, including social, cultural,economic, and psychological factors. Eachindividual presumably balances these factorsto reduce perceived disadvantage and achievepersonal satisfaction. A change in any factorshould prompt a new equilibrium point whichoptimises satisfaction. Those who are affectedby impairment or disability will adopt a rolethat maximises satisfaction with life withintheir limitations. When the adopted role isdisadvantageous compared with the norm forsomeone of the same age, sex, andbackground, the individual is classified ashandicapped. Pursuit of satisfaction is themotivation behind the choice of differentroles, and dissatisfaction is an expression ofperceived disadvantage and role preference,which are highly subjective features. Twopeople suffering from the same disability maydiffer in the disadvantage experienced becauseof differences in expectations and ambitionsand in the extent to which the disabilityinterferes with their normal activity. Forinstance, a woman with impaired hands due torheumatoid arthritis cannot dress her babywhich affects her maternal role. Anotherwoman with the same disability is moreconcerned about her inability to drive whichinterferes with her social life and workopportunities. A wealthy business woman inthe same position may have unrestricted work,travel and family capabilities by virtue of thenature of her work and the resources she canuse to overcome problems.

Measuring handicap directly . ..is difficult since the referencestandard is normality ....

Measuring handicap directly in a givenindividual is difficult since the referencestandard is "normality", and this covers a widerange of abilities, achievements, andexpectations. Moreover, physical environ-ment, resources and social situations, all ofwhich impinge on handicap, are very diverse.All we can do is describe qualitatively anindividual's unique predicament and assesstheir satisfaction with that state of affairs. Todevelop a quantitative measure from this we

55

on April 4, 2021 by guest. P

rotected by copyright.http://qualitysafety.bm

j.com/

Qual H

ealth Care: first published as 10.1136/qshc.3.1.53 on 1 M

arch 1994. Dow

nloaded from

http://qualitysafety.bmj.com/

H6i-aLool, _irotapuonkul, Dickinson, Ebrh'lliol

can use the econometric concept of "utilities"to measure the severity of handicapped states.The utility of a health state is the "value" or"worth" people put on it.3 By convention, afully healthy state is given a utility of + 1 0, anddeath a utility of 0 0 (states "worse thandeath" can have negative values). Utilitiesrepresent non-arbitrary values reflecting therelative desirability of various states, at aninterval level of measurement. Clearly, utilitieswill reflect relative preferences and values, andwill be closely related to satisfaction.We could measure utilities by asking

individuals to rate the desirability of their ownstate of health against an idealised normalstate. In practice this is very difficult, althoughit has been described as a measure of stress incarers." For many epidemiological and healthservices research purposes we need to measurea variable in a population or group rather thanknowing individual values per se. In thesecases we can describe each individual's health(by classifying each to the appropriate level oneach of the six handicap dimensions), thenapply a valuation of that description that hasbeen predetermined from a study of arepresentative group of "judges". The use ofutilities averaged from representative judgesmeans that an individual's handicap score willrepresent his or her handicap only to theextent that his or her views coincide with thoseof the judges. However, it does allowmeasurements to be made on importantgroups whose subjective opinions of their ownhealth would be difficult to obtain, such asthose with communication disabilities orcognitive impairment (since we can describetheir state of health objectively and then applythe valuations of those states made bv thejudges). The assumption has to be made thatvaluations made by judges able to take part ina valuation exercise would be the same aspatients' who are not. The same argumentsapply equally to other health indices with scaleweights derived from groups of judges,including the sickness impact profile,Nottingham health profile, quality of well-being scale, and Rosser-Kind index. Therehave been some interesting recent attempts toindividualism the items and scale weights inquality of life measures (D Ruta et al, thirdEuropean Health Services ResearchConference, London, 1991)," but these havehad to limit themselves to a few items andrequire considerable cooperation andcognitive ability from subjects.

A market research solutionA comprehensive, quantitative scale requires avaluation of the severity of all the possiblestates described by the health state descriptivesystem. In the ICIDH5 each of the sixdimensions was given nine levels of severity(from no disadvantage to extremedisadvantage). Any state of health can bedescribed by a combinations of one level fromeach of these six dimensions, giving 531 441(9 X 9 X 9 X 9 X 9 X 9) possible combinations(even though some of them are improbable;such as being severely disadvantaged in

orientation and mobility, yet normal inphysical independence). To measure theutility of each potential state one at a timewould clearly not be feasible. However, if thedimensions are relatively independent (or atleast can be imagined as being independent),various statistical techniques can be used toconstruct mathematical models for estimatingutilities for any state from measurements madeon a small sample of states. For example,Torrance et al used "multi-attribute utilitytheory" in developing a general health index.However, this method makes a number ofarguable assumptions and is computationallvdifficult. A simpler alternative called conjointanalysis has been developed for use in marketresearch."l'Market researchers commonly face a

problem exactly analogous to that describedhere. They need to know which characteristicsof a product or service are most important toconsumers. An ideal product will have all thebest characteristics, but in practice trade offsare required - for instance, between theperformance, reliability, fuel consumption,and price of a car. Each of these characteristicswill have several possible "levels", such as ahigh, medium, or low fuel consumption.Product designers need to know the effect thateach level of each different characteristic hason overall desirability. The problem is that ifthere are several different characteristics, andeach has several possibilities, the total numberof possible combinations soon becomesunmanageable (as with our descriptions ofhandicap). Conjoint analysis depends on theability of subjects to make judgements abouthypothetical products or states of health.Series of states are evaluated, each beingdescribed by a combination of different levelsof each characteristic (in this case differentseverities of disadvantage in mobility, physicalindependence, occupation, and so on).Respondents may be unable to explain whichcharacteristics they are using in making avaluation or how they are combining them toform overall judgements. However, conjointanalysis infers a value system from thesejudgements. The term "conjoint" refers to thisprocess of joint evaluation. In developing ahandicap measurement scale we wish to knowthe relative contributions of each of theseverities within the six dimensions ofhandicap.The technique for modelling responses is

computer-dependent and iterative." Given aset of judgements about health states, theprogram assigns arbitrary "part utilities" toeach level of each characteristic. These partutilities are used to calculate an estimatedoverall utility for each health state (usuallyusing an additive model and ignoringinteractions), the estimates are compared withthe measured values and a goodness of fitstatistic is calculated. The part utilities arethen systematically varied until the calculatedvalues correspond as closely as possible to thedata. Fortunately the procedure does notrequire data from all the possiblecombinations of different levels of each

56

on April 4, 2021 by guest. P

rotected by copyright.http://qualitysafety.bm

j.com/

Qual H

ealth Care: first published as 10.1136/qshc.3.1.53 on 1 M

arch 1994. Dow

nloaded from

http://qualitysafety.bmj.com/

Measuring a handicap: motives, methods, and a model

characteristic; estimates can be made from asmall sample if the states are varied in certainsystematic ways. Although interactions aretheoretically likely (in market research, mostimprovements to a product affect its price, andphysical independence is partly dependent onmobility), as a general rule the simplestpossible model is chosen, and its robustnessthen tested in practice (as part of validation).A method must be chosen for determining

the utility of health states to provide the rawdata for the modelling process. Utilities can bemeasured by a number of techniques,including the standard gamble, time trade off,or simpler magnitude estimation or visualanalogue scale. Some work suggests that theygive approximately similar results,44 althoughother evidence is contradictory. There is noclear advantage for any one method ontheoretical grounds. Often simple ranking ofseverity (or desirability) is used.With the approach described here, a scale

can be constructed comprising a handicapclassification questionnaire (enabling anindividual's health state to be described interms of different levels of disadvantage oneach of the six dimensions), and a matrix ofscale weights relating to each level in eachdimension (the "part-utilities" from theconjoint analysis). A formula is then used tocombine them into a final handicap score,representing an estimate of the utility that the"judges" would have given that state were theyto have rated it directly. An accompanyingpaper in this issue (p 1 1)45 describes such ascale we developed from these ideas.

SJ was supported by the British Council and ED by the RoyalCollege of Physicians of London; RHH is an MRC healthservices research training fellow.

1 Bury MR. The ICIDH: a review of research and prospects.International Disability Studies 1987;9: 118-28.

2 Williams GH. Disablement and the social context of dailyactivity. International Disability Studies 1987;9:97-102.

3 Royal College of Physicians. Physical disability in 1986 andbeyond. J R Coil Physicians Lond 1986;20: 160-94.

4 Seale C, Davies P. Outcome measurement in strokerehabilitation research. Int Disabil Studies 1987;9:155-60.

5 World Health Organisation. International Classification ofImpairments, Disabilities and Handicaps. Geneva: WHO,1980.

6 Mahoney FI, Barthel DW. Functional evaluation: theBarthel index. Maryland State Medical Journal 1965;14:61-5.

7 Nouri FM, Lincoln NB. An extended activities of dailyliving scale for stroke patients. Clinical Rehabilitation1987;1:301-5.

8 Martin J, Meltzer H, Elliot D. The prevalence of disabilityamong adudlts. London: HMSO, 1988. (OPCS SocialSurvey Division.)

9 Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW.Studies of illness in the aged. The Index of ADL: astandardized measure of biological and psychosocialfunction. JAMA 1963;185:914-9.

10 Ebrahim S, Nouri F, Barer D. Measuring disability after astroke. J Epidemiol Community Health 1985;39:86-9.

11 Bond J, Gregson BA, Atkinson A. Measurement ofoutcomes within a multi-centred randomised controlledtrial in the evaluation of the experimental NHS nursinghomes. Age Ageing 1989;18:292-302.

12 Gilchrist WJ, Newman RJ, Hamblen DL, Williams BO.Prospective randomised trial of an orthopaedic geriatricinpatient service. BMJ 1988;297: 1116-8.

13 Garraway WM, Akhtar AJ, Hockey L, Prescott RJ.Management of acute stroke in the elderly: follow up ofa controlled trial. BMJ 1980;281:827-9.

14 Cowley AJ, Fullwood LJ, Muller AF, Stainer K, Skene AM,Hampton JR. Exercise capability in heart failure: iscardiac output important after all? Lancet 1991 ;337:771-3.

15 Ebrahim S. Measurement of impairment, disability andhandicap. In: Hopkins A, Costain D, eds. Measuring theoutcomes of medical care. London: Royal College ofPhysicians of London, 1990:27-41.

16 Williams SJ, Bury MR. "Breathtaking": the consequencesof chronic respiratory disorder. International DisabilityStudies 1989;11:114-20.

17 Caine N, Harrison SCW, Sharples LD, Wallwork J.Prospective study of quality of life before and aftercoronary artery bypass grafting. BMJ 1991;302:511 -6.

18 Croog SH, Levine S, Testa MA, Brown B, Bulpitt CJ,Jenkins CD, et al. The effects of antihypertensive therapyon qualify of life. N Engl _t Med 1986;314:1657-64.

19 Bombardier C, Ware J, Russell IJ, Larson M, Chalmer A,Read JL. Auranofin therapy and quality of life in patientswith rheumatoid arthritis. Amn J Med 1986;81:565-78.

20 Walker SR, Rosser RM. Quality of life: assessment andapplication. Lancaster: MTP Press, 1988.

21 Fitzpatrick R, Ziebland S, Jenkinson C, Mowat A, MowatA. Importance of sensitivity to change as a criterion forselecting health status measures. Quality in Health Care1992;1:89-93.

22 Ebrahim S, Williams J. Assessing the effects of a healthpromotion programme for elderly people. J Public HealthMed 1992;14:199-205.

23 Warren MD. The Canterbury studies of disablement in thecommunity: prevalence, needs, and attitudes. Int 7Rehabil Res 1985;8:3-18.

24 Wood PHN. Maladies imaginaires: some commonmisconceptions about the ICIDH. International DisabilityStudies 1987;9:125-8.

25 Rankin J. Cerebral vascular accidents in patients over theage of 60. 2. Prognosis. Scot MedJ7 1957;2:200-15.

26 UK-TIA Study Group. The UK-TIA aspirin trial: interimresults. BMJ 1988;296:316-20.

27 Jefferys M, Hyman M, Millard JB. A set of tests formeasuring motor impairment in prevalence studies.Journal of Chronic Diseases 1969;22:303-19.

28 Hebert R, Carrier R, Bilodeau A. The functional autonomymeasurement system (SMAF): description and validationof an instrument for the measurement of handicaps. AgeAgeing 1988;17:293-302.

29 Affleck JW, Aitken RCB, Hunter JAA, McGuire RJ, RoyCW. Rehabilitation Status: a measure of medicosocialdysfunction. Lancet 1988;i:230-3.

30 Harris Al. Handicapped and impaired in Great Britain.London: HMSO, 1971.

31 Schow RL, Nerbonne MA. Hearing handicap and Denverscales: applications, categories, interpretation. JAMA1980;13:66-77.

32 Fiske J, Gelbier S, Watson RM. The benefit of dental careto an elderly population assessed using a sociodentalmeasure of oral handicap. Br Dent J 1990;168:153-6.

33 Jacobson GP, Newman CW. The development of theDizziness Handicap Inventory. Arch Otolaryngol HeadNeck Surg 1990;116:424-7.

34 O'Brien B. Assessment of treatment in heart disease. In:Teeling-Smith G, ed. Measuring health: a practicalapproach. Chichester: Wiley, 1988:191-210

35 Rosser RM, Kind P. A scale of valuations of states of illness:is there a social consensus? Int J Epidemiol1978;7:347-58.

36 Buxton M. Are we satisfied with QALYs? What are theconceptual and empirical uncertainties and what must wedo to make them more generally useful? In: Hopkins A,ed. Measures of the quality of life and uses to which suchmeasures may be put. London: Royal College ofPhysicians, 1992:41 -51

37 Von Neuman J, Morgenstern D. Theory of games andeconomic behaviour. 3rd ed. New York: Wiley, 1953.

38 Mohide A, Torrance GW, Streiner DL, Pringle DM,Gilbert R. Measuring the well-being of family caregiversusing the time trade-off technique. J Clin Epidemtiol1988;41:475-82.

39 O'Boyle CA, McGee H, Hickey A, O'Malley K, JoyceCRB. Individual quality of life in patients undergoing hipreplacement. Lancet 1992;339:1088-91.

40 Torrance GW, Boyle MH, Horwood SP. Application ofmulti-attribute utility theory to measure socialpreferences for health status. Operational Research1982;30:1043-69.

41 Churchill GA. Marketing research: methodologicalfoundations. 3rd ed. Chicago: Dryden Press, 1989.

42 SPSS. SPSS categories. Chicago: SPSS, 1990.43 Kruskal JB. Analysis of factorial experiments by estimating

monotone transformations of the data. Journal of theRoyal Statistical Society Series B 1965;27:251-63.

44 Patrick DL, Bush JW, Chen MM. Methods for measuringlevels of well-being for a health status index. HealthServices Research 1973;8:228-45.

45 Harwood RH, Rogers A, Dickinson E, Ebrahim S.Measuring handicap: the London handicap scale, a newoutcome measure for chronic disease. Quality in HealthCare 1994;3:1 1-6.

57 on A

pril 4, 2021 by guest. Protected by copyright.

http://qualitysafety.bmj.com

/Q

ual Health C

are: first published as 10.1136/qshc.3.1.53 on 1 March 1994. D

ownloaded from

http://qualitysafety.bmj.com/