REVIEW. Methods of assessing health-related quality of life and outcome for plastic surgery

Transcription

1 British Journal of Plastic Surgery (1999), 52, The British Association of Plastic Surgeons REVIEW Methods of assessing health-related quality of life and outcome for plastic surgery R. Fitzpatrick, C. Jenkinson, A. Klassen* and T. Goodacre Institute of Health Sciences, University of Oxford, Oxford, UK; *Centre for Evaluation Sciences, University of British Columbia, Canada; and Department of Plastic and Reconstructive Surgery, Radcliffe Infirmary, Oxford, UK Keywords: outcome, plastic surgery, health, quality of life. Increasingly, evidence is required of the effectiveness of health services. A central form of evidence is the outcome of health care as assessed by the patient. There is considerable scope for improving the methodology of research in plastic surgery. 1,2 This paper is intended to provide an overview of the potential role of patient-based outcome measures. In recent years, a vast array of questionnaires have been developed to assess the patient s perspective. They are referred to here as patient-assessed outcome measures, but the literature variously terms them health-related quality of life, quality of life, health status, functional status, subjective health status, or sometimes just outcome measures. The main rationale for such measures is the fact that the majority of health care interventions have as either their primary or secondary purpose the improvement of one or more aspects of patients health-related quality of life (Table 1). To date, disproportionate attention in research in plastic surgery has focused on a narrow range of psychiatric and psychological dimensions of patients well-being. 3,4 However, one study indicated that patients receiving a range of plastic surgical interventions, when asked to identify areas of their lives most affected by their presenting problem, reported a wide range of problems including diverse physical symptoms, difficulties in personal relationships, problems in pursuing leisure activities, problems of self-concept and difficulties finding suitable clothes. 5 When this sample of patients also completed a standardised patient-assessed outcome measure (the SF-36 described below), a wide range of problems in physical, psychological and social function were identified prior to surgery and the same questionnaire provided unique evidence of improvements associated with surgery at follow-up. 5 There is now extensive evidence that proxy reports of patients health status and well-being, such as those provided by health professionals or carers, do not always agree with assessments patients make themselves; hence the need for patients own assessments. 6 There are four main uses for patient-assessed outcome measures. Firstly, they serve as measures of outcome in evaluative research. Here their clearest role is in the randomised trial, where potential biases in attributing changes in quality of life to interventions are minimised. A second and related use is in audit and quality assurance. Thirdly, they may be used in surveys to assess the health care needs of populations, either of geographical areas or of services such as patients of a particular clinic. The fourth application is to assist health professionals in individual patient care by providing a feasible and relevant set of information about patients progress. Background Patient-assessed outcome measures emerged because of a number of related developments in the field of health care and social policy more generally. Patrick and Guttmacher distinguish four circumstances that facilitated their emergence in the late 1960s and 1970s. 7 Firstly, the growing importance of chronic and disabling disease meant that measures of outcome more appropriate to interventions for chronic diseases were needed. Secondly, methods were needed to evaluate the expanding array of new medical technologies. Thirdly, increased levels of public expenditure on health care required increased attention to evidence of value for money. Fourthly, governments increasingly required measures that might assist in determining fair and appropriate allocations of resources in health care. In addition to these developments, influential bodies such as the World Health Organisation were advocating an expanded definition of health: a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity. 8 Lerner and Levine distinguish three separate traditions that contributed to the development of patientassessed outcome measures. 9 One tradition, termed the health indicators school by Lerner and Levine, began to develop measures of outcome for the field of health care, focusing initially on measuring health as something multi-dimensional, including health-related symptoms and physical function. A second tradition, informed by social scientists from the early 1970s, began to develop social indicators of quality of life such as level of well-being and life satisfaction to measure social progress and the impact of social policy generally. 251

2 252 British Journal of Plastic Surgery Table 1 Dimension Typical dimensions of health-related quality of life assessed by patient-assessed outcome measures Illustrative content Physical function Symptoms Global judgements of health Psychological well-being Social well-being Cognitive functioning Role activities Personal constructs Satisfaction with care Mobility, range of movement, physical activity Activities of daily living: ability to eat, wash, dress Pain, fatigue, nausea Global self-rating of health Anxiety, depression, sense of control, self-esteem Family and intimate relations, social contact, integration, leisure activities Memory, alertness, ability to communicate Employment, financial concerns, household management Satisfaction with bodily appearance, stigma, life satisfaction, spirituality Satisfaction with treatment, communication, interpersonal relations These were applied to the field of health in the form of socio-medical indicators that went beyond conventional indicators such as mortality to assess the impact of medical care on variables such as life satisfaction and performance of social roles. 10 A third tradition, the medical ethicist school, at this time began to define aspects of the good life relevant to health care, such as happiness, self-determination and ability to pursue one s life goals. Although very disparate, these various traditions shaped the emerging expanded conception of health and how it should be measured. One of the most important early instruments to be developed was the Index of Activities of Daily Living (IADL), designed to describe older patients abilities in, for example, bathing, dressing and feeding. 11 Another influential instrument was the Health Status Index, more recently termed the Quality of Well-Being Scale. 12 This instrument assesses symptoms and medical problems, which are then assigned an overall weighting from members of society s judgements of relative severity and importance. An individual s global score can be adjusted to take account of prognosis. Instruments such as the IADL and the Quality of Well-Being Scale required trained interviewers to be administered. More recently, research has focused on measures in the form of questionnaires that respondents can complete by themselves. The first such instrument to become widely used across a range of health problems was the Sickness Impact Profile which, in its final form, contains 136 simple yes / no statements indicating changes in an individual s behaviour due to sickness. 13 Items aggregate into 12 separate scales assessing such dimensions as mobility, recreation and pastimes, emotional behaviour and communication. The last major type of instrument to be developed is referred to as a measure of utilities. Such instruments are distinctive because instead of simply recording the presence or absence of experiences, they assign weights (utilities) based on independent evidence of the overall values that individuals assign to given health states. An example of such an instrument is the Health Utilities Index, which assigns utilities to eight dimensions of health, for example vision, dexterity, pain and self-care. 14 Patient-assessed outcome measures began to be reported as main end-points of major clinical trials, thereby further stimulating their development and adoption in research As a result, there is now available an enormous array of patient-assessed outcome measures, differing in content, length, form of delivery, measurement properties and intended purpose. Several sources list and describe commonly used instruments. 18,19 The two main types of instruments that are commonly used are so-called generic and diseasespecific instruments. As the name implies, generic questionnaires are intended to be applicable to the widest range of health status problems. In principle, this should facilitate comparisons of the effectiveness of very different interventions in diverse patient groups. One of the most widely used of such instruments is the SF-36, a 36-item self-completed questionnaire assessing eight aspects of well-being, such as physical and social function, mental health and pain. 20 Disease-specific instruments are developed for use in one particular disease or a relatively narrow range of conditions, for example arthritis, or patients receiving hip-replacement surgery. The advantage of disease-specific instruments is that, in principle, they should be more sensitive to the specific problems of a particular condition. The Derriford Scale has been developed to be of specific relevance to patients receiving plastic surgery. 21 Since there is no single patient-assessed outcome measure that can be recommended for universal use in plastic surgery, it may be useful to have in mind some basic principles in judging a questionnaire s value in any given context. The following criteria for evaluating patient-assessed outcome measures were identified as a result of an extensive review of the available literature. 22 Appropriateness The first and most fundamental consideration in evaluating a potential outcome measure is to determine its appropriateness. For example, a trial of cosmetic surgery would need to consider whether it requires measures of psychological well-being that are generic and provide assessments of broader aspects of patients health status or, instead, more specific measures that focus on problems in relation to self-esteem and satisfaction with body image relevant to the patient group being investigated. Several studies that have reviewed samples of clinical trials have reported that the majority of patient-assessed measures used were inappropriate. 23,24 Validity In addressing whether an instrument is valid, that is whether it measures what it purports to measure, some

3 Methods of assessing health-related quality of life and outcome for plastic surgery 253 methods of assessing validity cannot realistically be applied. In particular, criterion validity requires that we have available a gold-standard measure against which a new instrument can be tested. This is rarely if ever the case and, were it the case, would minimise the need for a new measure. Instead, two broad and equally important approaches to validity have emerged. On the one hand, qualitative judgements can be made about the range and content of items in a questionnaire. This approach, content validity, addresses whether the range of aspects of a phenomenon have been covered. For example, does a scale of mobility contain items concerned with mobility in the home as well as in public spaces? Face validity addresses whether items appear, from their manifest expression, to measure what they purport to measure. These qualitative judgements can be structured in two ways. Most vitally, how extensive is the evidence that patients rather than experts participated in originally determining the content of an instrument? Secondly, patients, carers and other relevant interests can participate in further testing the validity of an instrument to a new field of application. The second broad approach is quantitatively to examine the construct validity of an instrument by examining patterns of relationships with a range of other variables. This does not depend on relationships with a single criterion variable. For example, we might expect poorer scores on a quality of life measure to be associated with greater severity and chronicity of the health problem, poorer psychological mood and more frequent use of health services. Reliability There are two distinct but related aspects to reliability as a requirement of patient-assessed outcome measures. Firstly, the majority of questionnaires assess a concept such as pain or self-esteem by means of several questionnaire items, because several related observations are expected to produce a more reliable estimate than one. For this to be the case, items all need to be sufficiently homogeneous, that is, all measuring aspects of a single concept. Thus questionnaire items intended to measure something such as pain should have high internal consistency, in that items should correlate adequately with each other. However, items of a scale should not be too homogeneous, because it is possible that the items are too similar and measure a very restricted aspect of, say, self-esteem. The second aspect of reliability is the requirement that instruments yield the same results on repeated applications, assuming the respondent s health status has not actually changed between measures. This property, reproducibility, is assessed by examining agreement between scores from two administrations of a questionnaire no more than a few days apart. Responsiveness Responsiveness denotes, in this context, the extent to which an instrument is sensitive to changes over time that are of importance to patients. In the study of various plastic surgical interventions cited earlier, questionnaires assessing the psychological well-being of patients differed markedly in the extent of improvement detected as an outcome of surgery. 5 Whilst increasingly accepted as a distinct and important requirement of a patient-assessed outcome measure, there is less agreement about how to assess responsiveness. Some studies have examined the degree of association of changes in the instrument of concern with changes over time in other parameters such as disease severity. 25 Other approaches identify as responsive those instruments that produce the largest amount of change when change is, for independent reasons such as the known effectiveness of an intervention, expected to occur. 26 Precision At one extreme on a spectrum of patient-assessed outcome measures are those that make a small number of simple and broad distinctions between health or quality of life states; at the opposite end are instruments that make many distinctions, most notably with continuous measures such as visual analogue scales. Unfortunately, it cannot be argued that instruments that are capable of more distinctions are more precise. Elegant studies have shown that commonly used health status instruments may not equally represent the full underlying range of severity of problems. Stucki et al, for example, argue from evidence with patients undergoing hip replacement surgery that the SF-36 over-represents items of underlying moderate severity in physical disability in such a way that change scores can exaggerate improvement for respondents with moderate disability at baseline. 27 Ceiling and floor effects may also be a problem with instruments not being capable of measuring either improvement or deterioration beyond particular points on the instrument. Such effects have been identified as a major problem for some patient-assessed outcome measures used in plastic surgery. 28 Interpretability Patient-assessed outcome measures produce numerical values that are not necessarily intuitively understandable. As patient-assessed outcome measures become more widely employed, repeated use will itself increase familiarity of the various audiences with the meaning of results. In addition, several different methods have been employed to increase interpretability. One approach is to correlate scores of such instruments with other information such as severe life events so that some common-sense or intuitive calibration of scores can be provided; a score of a particular magnitude on an instrument may, for example, be equivalent to the typical response to bereavement. 29 Another approach is to identify the smallest clinically meaningful scores for an instrument by relating change scores for that instrument to independent judgements of the scale or clinical significance of such experiences. 30 To date, interpretability is one of the least explored aspects of patient-assessed outcome measures.

4 254 British Journal of Plastic Surgery Acceptability It is essential that the burden to patients of completing outcome measures is minimised so that distress is as far as possible avoided and the refusal rate and rate of incomplete data for outcomes are reduced. In general terms, the acceptability of patient-assessed outcome measures will be determined by their complexity, the distress arising from items, the overall layout, appearance and legibility of the questionnaire and the time required to complete the task. Amongst patients who had undergone hip replacement surgery and were invited to return by post a 12-item questionnaire about hip pain and a 36-item generic questionnaire, the completion rate for the shorter instrument was 98% and for the longer instrument 73%. 31 It is useful to have available evidence about time to complete different instruments of the kind provided by Read et al who found that three patient-assessed outcome measures required 11, 18 and 22 min respectively. 32 Given the evidence that patients with poorer health status are less likely to complete quality of life measures, there is a considerable risk of bias if this aspect is neglected. 33 Feasibility In addition to the possible burden to patients, it is necessary to estimate the likely impact of instruments on clinical staff and researchers in collecting, coding and analysing patient-assessed data. At one extreme are brief self-completed questionnaires that impose minimal disruption. At the other extreme are instruments that require lengthy training of staff and lengthy interviews with patients to obtain data. Whilst investigators need to anticipate feasibility of data collection, they should not consider this a completely fixed property of instruments. Bernhard et al conducted a qualitative study of the collection of quality of life data from patients in a clinical trial and concluded that staff attitudes to the value of such information were a key determinant of success in obtaining data. 34 Conclusions Many problems presented by patients in plastic surgery involve subjective experiences such as embarrassment, impaired social life and distress in relation to appearance. The new array of questionnaires offer potentially simple, valid and standardised means of assessing outcomes of these aspects of patients experiences As it becomes increasingly necessary to marshal multi-centre trials to detect the moderate benefits that may be gained from improvements in surgical techniques, patient-based outcome measures actually make such work more feasible. However, clearly at present considerable judgement is required in order to select an instrument. Also patient-assessed outcome measures do not have fixed properties such as reliability and validity in some universal sense; strictly these are properties relative to a specific use. In selecting an instrument it may be necessary to trade-off the criteria discussed in this review; an instrument that may have very substantial validity in collecting very detailed information about patients experiences of an illness may not be acceptable and feasible because of length. Overall, patient-assessed outcome measures provide an important source of evidence complementing existing clinical, laboratory and other data. References 1. Wengle HP. The psychology of cosmetic surgery: a critical overview of the literature Part 1. Ann Plast Surg 1986; 16: Coady MSE. Measuring outcomes in plastic surgery. Br J Plast Surg 1997; 50: Goin MK, Rees TD. A prospective study of patients psychological reactions to rhinoplasty. Ann Plast Surg 1991; 27: Wright MR, Wright WK. A psychological study of patients undergoing cosmetic surgery. Arch Otolaryngol 1975; 101: Klassen A, Jenkinson C, Fitzpatrick R, Goodacre T. Patients health related quality of life before and after aesthetic surgery. Br J Plast Surg 1996; 49: Sprangers MA, Aaronson NK. The role of health care providers and significant others in evaluating the quality of life of patients with chronic disease: a review. J Clin Epidemiol 1992; 45: Patrick D, Guttmacher S. Socio-political issues in the use of health indicators. In: Culyer A (ed.), Health Indicators. Oxford: Martin Robertson, 1983; World Health Organisation. Constitution of the World Health Organisation. Basic Documents. Geneva: WHO, Lerner D, Levine S. Health-related quality of life: origins, gaps, and directions. In: Albrecht G, Fitzpatrick R (eds), Quality of Life in Health Care. Advances in Medical Sociology, Vol 5. Connecticut: JAI Press, 1994; Elinson J. Toward sociomedical health indicators. Social Indicators Research 1974; 1: Katz S, Ford A, Moskowitz R, Jacobson B, Jaffe M. The Index of ADL: a standardised measure of biological and psychosocial function. JAMA 1963; 185: Fanshel S, Bush J. A health status index and its application to health service outcomes. Operations Research 1970; 18: Bergner M, Bobbitt RA, Pollard WE, Martin DP, Gilson BS. The sickness impact profile: validation of a health status measure. Med Care 1976; 14: Torrance G, Furlong W, Feeny D, Boyle M. Multi-attribute preference functions: Health Utilities Index. Pharmaco- Economics 1995; 7: Croog SH, Levine S, Testa MA, et al. The effects of antihypertensive therapy on the quality of life. New Engl J Med 1986; 314: Bombardier C, Ware J, Russell IJ, Larson M, Chalmers A, Read JL. Auranofin therapy and quality of life in patients with rheumatoid arthritis: results of a multicenter trial. Am J Med 1986; 81: Rorabeck CH, Bourne RB, Laupacis A, et al. A double-blind study of 250 cases comparing cemented with cementless total hip arthroplasty: cost-effectiveness and its impact on healthrelated quality of life. Clin Orthop 1994; 31: Bowling A. Measuring Disease. Buckingham: Open University Press, McDowell I, Newell C. Measuring Health. Oxford: Oxford University Press, Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care 1992; 30: Ward CM. Rationing and resource management. Br J Plast Surg 1994; 47: Fitzpatrick R, Davey C, Buxton M, Jones D. Evaluating patient-based outcome measures for use in clinical trials. Health Technology Assessment 1998; 2 (14). 23. Guyatt GH, Veldhuyzen Van Zanten SJ, Feeny DH, Patrick DL. Measuring quality of life in clinical trials: a taxonomy and review. Can Med Assoc J 1989; 140:

5 Methods of assessing health-related quality of life and outcome for plastic surgery Gill TM, Feinstein AR. A critical appraisal of the quality of quality-of-life measurements. JAMA 1994; 272: Meenan RF, Anderson JJ, Kazis LE, et al. Outcome assessment in clinical trials: evidence for the sensitivity of a health status measure. Arthritis Rheumatism 1984; 27: Liang MH, Fossel AH, Larson MG. Comparisons of five health status instruments for orthopedic evaluation. Med Care 1990; 28: Stucki G, Daltroy L, Katz J, Johannesson M, Liang M. Interpretation of change scores in ordinal clinical scales and health status measures: the whole may not equal the sum of the parts. J Clin Epidemiol 1996; 49: Klassen A, Fitzpatrick R, Jenkinson C, Goodacre T, The value to patients of cosmetic surgery. J Epidemiol Community Health (in press). 29. Testa M, Simonson D. Assessment of quality of life outcomes. N Engl J Med 1996; 334: Juniper EF, Guyatt GH, Willan A, Griffith LE. Determining a minimal important change in a disease-specific quality of life questionnaire. J Clin Epidemiol 1994; 47: Dawson J, Fitzpatrick R, Murray D, Carr A. Comparison of measures to assess outcomes in total hip replacement surgery. Quality in Health Care 1996; 5: Read JL, Quinn RJ, Hoefer MA. Measuring overall health: an evaluation of three important approaches. J Chronic Dis 1987; 40 Suppl 1: 7S 26S. 33. Hopwood P, Stephens RJ, Machin D. Approaches to the analysis of quality of life data: experiences gained from a Medical Research Council lung cancer working party palliative chemotherapy trial. Qual Life Res 1994; 3: Bernhard J, Gusset H, Hurny C. Quality-of-life assessment in cancer clinical trials: an intervention by itself? Support Care Cancer 1995; 3: Cole RP, Shakespeare V, Shakespeare P, Hobby JAE. Measuring outcome in low-priority plastic surgery patients using Quality of Life indices. Br J Plast Surg 1994; 47: Klassen A, Fitzpatrick R, Jenkinson C, Goodacre T. Should breast reduction surgery be rationed? A comparison of the health status of patients before and after treatment: postal questionnaire survey. Br Med J 1996; 313: Shakespeare V, Cole RP. Measuring patient-based outcomes in a plastic surgery service: breast reduction surgical patients. Br J Plast Surg 1997; 50: The Authors R. Fitzpatrick C. Jenkinson Division of Public Health and Primary Health Care, Institute of Health Sciences, Oxford University, Old Road, Headington, Oxford OX3 7LF, UK. A. Klassen, Centre for Evaluation Sciences, University of British Columbia, Canada. T. Goodacre, Department of Plastic and Reconstructive Surgery, Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE, UK. Correspondence to Professor R. Fitzpatrick. Paper received 19 May Accepted 12 January 1999.