Measuring Quality of Life in Stroke

Transcription

1 320 Measuring Quality of Life in Stroke R. de Haan, MS; N. Aaronson, PhD; M. Limburg, MD, PhD; R. Langton Hewer, MD, PhD; and H. van Crevel, MD, PhD Background and Purpose: Little attention has been focused on quality of life in stroke outcome research. The purpose of this review is to outline the meaning of the concept, describe important methodological issues and methods of assessment, review existing quality of life measures, and discuss criteria for selecting an appropriate instrument. Summary ofreview: The following 10 quality of life instruments were reviewed: COOP Charts; Euroqol; Frenchay Activities Index; Karnofsky Performance Status Scale; McMaster Health Index Questionnaire; Medical Outcomes Study 20-Item Short-Form Health Survey; Nottingham Health Profile; Quality of Life Index; Quality of Well-being Scale; and the Sickness Impact Profile. They were evaluated in terms of length, time needed to complete, content, scoring, and psychometric characteristics. Conclusions: Emphasis should be placed on further psychometric evaluation of existing quality of life measures rather than on generating new instruments. There is particular need for supplementary data on the responsiveness of the instruments to changes in patients' clinical status over time. The choice of a suitable quality of life instrument should be based not only on psychometric properties but also on careful consideration of the research question, the relevance to the objectives of the study, the feasibility of the instrument, and the specific characteristics of the stroke patients under investigation. (Stroke 1993;24: ) KEY WORDs * quality of life * stroke outcome Downloaded from by on November 3, 2018 D espitethegrowinginterestinqualityoflife (QL) issues in clinical research and practice, little attention has been paid to evaluating systematically the QL of stroke patients.1 QL investigations are useful not only in gaining a better understanding of patients' reactions to illness and for enhancing supportive care2 but also for evaluating the efficacy of therapeutic interventions. Although there is an association between neurological deficits and QL, they are not synonymous. If, with any given intervention, a clinical benefit has been obtained as demonstrated in terms of improved neurological function, an evaluation of the effects on patients' daily functioning, subjective health, and well-being is still highly relevant. The objectives of this paper are 1) to describe the QL concept as it is applied in health care settings; 2) to outline the main methodological problems in measuring QL; 3) to review existing QL measures; 4) to identify the most relevant criteria for selecting suitable instruments; and 5) to propose directions for future QL studies in the field of stroke research. From the Departments of Neurology (R. de H., M.L., H. van C.) and Clinical Epidemiology and Biostatistics (R. de H.), Academic Medical Center; the Department of Psychology (N.A.), University of Amsterdam; the Amsterdam World Health Organization Collaborating Center on Quality of Life (N.A.), Amsterdam, The Netherlands; and the Department of Neurology (R.L.H.), Frenchay Hospital, Bristol, UK. Address for correspondence: R. de Haan, Department of Neurology, Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. Received August 18, 1992; accepted November 3, The Concept of Quality of Life Over the years QL has been defined in many different ways. One speaks, for example, of need satisfaction,3 health-related subjective experiences,4 or psychosocial and physical well-being.5 One way of resolving the problem of definition is to leave this task to the individual patient by simply asking the question: "How would you rate your present QL?"6-8 A single-item QL measure, however, is in itself not very reliable or valid and is of little analytic value. More specific information is required to interpret global ratings of a concept as complex as QL. For this reason, most researchers today adopt a multidimensional approach to QL assessment.9 A broad consensus has emerged that at least four dimensions should be included in a QL assessment: physical, functional, psychological, and social health. The physical health dimension refers primarily to disease-related and treatment-related symptoms. Functional health comprises self-care, mobility, and physical activity level, as well as the capacity to carry out various roles in relation to family and work. Cognitive functioning, emotional status (especially poststroke depression), and general perceptions of health, well-being, life satisfaction, and happiness are the central components of the psychological life domain. Social functioning includes the assessment of qualitative and quantitative aspects of social contacts and interactions. These four QL domains are also reflected, in part, in the International Classification of Impairments, Disabilities, and Handicaps (ICIDH) of the World Health Organization.10 Impairments reflect organ dysfunctions or abnormalities of body structure. Disabilities refer to the consequences of impairments in terms of the pa-

2 tient's functional performance. Handicaps are concerned with the social disadvantages resulting from impairments and disabilities. This overlap between QL and the organ, person, and societal levels of the ICIDH is to be expected because both are closely linked to the World Health Organization definition of health as "a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity."'" Various studies of QL in stroke indicate that physical disabilities often have a negative impact on QL, but this is not always the case. Stroke patients with little or no physical dysfunction can also experience a compromised QL Psychological status appears to be as important as physical disability in altering an individual QL.6 In summary, we stress the importance of conceptualizing QL as a broad spectrum of consequences of disease, including elements of impairments, disabilities, and handicaps, as well as patients' perceived health status and well-being. Methodological Issues QL measures can be differentiated into 1) generic scales or health profiles, 2) disease-specific scales, and 3) scale batteries.15'6 Generic scales are not developed for a specific target population and may be suitable for use with many patient populations. A major strength of these generic scales is the possibility of detecting the relative effects of disease and treatment on different life domains. Another advantage is that they allow comparison of QL results across patient populations. Generic scales, however, are not always sufficiently focused on the specific problems of any given patient population. Disease-specific QL scales exist for stroke,17 rheumatic disorders,18 cardiovascular diseases,19 and cancer.2 Such scales do not allow cross-disease comparisons but are often more sensitive to the QL issues particularly relevant to a specific population of patients. A battery of scales for measuring particular dimensions or aspects of health can also be used, such as instruments to assess cognitive functioning, depression, activities of daily living (ADL), or social functioning. The use of such a series of specific and often unidimensional measures allows one to assess each relevant life domain in depth. This approach has the disadvantages that results may not be comparable across studies and that the patient burden associated with completing such a battery may be unacceptably high. Most of the available QL instruments depend on patient self-reports. The data can be collected by either structured interviews or written questionnaires. These methods of data collection are less suitable for patients with serious cognitive, speech, and language disorders. It is not surprising, therefore, that in stroke outcome studies patients with severe impairments have often been excluded from QL assessments.12'13,20-22 This creates the problem of not evaluating QL in a subgroup of patients in which this is highly relevant. An alternative approach is to ask physicians to rate their patients' QL. However, this may create methodological difficulties. Some studies have shown low levels of interobserver reliability as well as disagreement between ratings provided by physicians and those of patients.23-26there are indications that patients' ratings of their overall QL reflect a broad range of physical, emotional, and social health concerns, whereas physicians' ratings are more de Haan et al Measuring Quality of Life 321 typically restricted to the physical health status of the patient.26 The QL of stroke patients with communication disorders can also be assessed by significant others such as partners or children. There are two variants of such measurements. The most straightforward is to ask the caregiver to rate the QL of the patient. The alternative, the so-called proxy measurement, involves asking the respondent to answer the questions as he or she thinks the patient would. Evidence regarding the quality of caregiver reports is mixed.27 Some suggest that they are reliable and valid.28 Others find evidence that the social network of the patient underreports health complaints,29 satisfaction, and emotional health.30 Studies on the interchangeability of patient and proxy responses, by definition, can be based only on score agreement between communicative patients and their caregivers. Thus, even in the case of complete concordance one cannot necessarily generalize the findings to the target population of noncommunicative patients. Surprisingly few QL studies have included control groups or other normative data in their designs or analysis.13"14'3' In stroke research, the use of agematched control groups should be viewed as an essential component of a QL investigation. Such a control group allows one to distinguish between QL effects related to the disease and its treatment and those attributable to the aging process per se. Alternative approaches to QL assessments involve the use of utility measures. Utility refers to the value attached to specific levels of health. For clarity of presentation we will focus only on the use of utility scales in clinical outcome studies (see References 32 and 33 for a discussion of utility analysis in clinical decision making). Utility scales, as a special class of generic instruments, focus on a broad range of life domains. The scales classify patients into various health states. The utilities of these health states typically have been assessed previously by a representative sample of the general public, by patient samples, or by a panel of experts and are expressed as a single numeric value on a scale ranging from 0.0 (death or worst possible outcome) to 1.0 (complete health or best possible outcome). These aggregated and weighted utility scores can be used as an evaluative end point in clinical trials. An attractive feature of the utility approach is that it yields a single summary score. This facilitates comparison of results across studies and the performance of cost-utility analyses. In these latter economic studies, survival data are combined with health ratings to yield quality-adjusted life years.34 This index reflects the gained years of life resulting from intervention, corrected by the utility of that life. By comparing alternative treatments on their costs and effects, cost-utility analysis can provide a useful basis for allocation of scarce health care resources.35,36 Various methodological problems in utility measurements have yet to be resolved. Torrance33 argues that individuals differ considerably in their health state preferences but that group mean utility values, required in most applications, are fairly stable. However, research is needed to clarify differences in utility values generated by different measurement methods. Moreover, differences in valuation of health states by different populations such as physicians, patients, and the

3 322 Stroke Vol 24, No 2 February 1993 general public, as well as the agreement of utility measures with other QL measures, merit further study-37 Ultimately, the decision as to whether to use a more classic, psychometric approach (e.g., questionnaires) or a utility approach to QL assessment depends on the specific research question being addressed. If one is interested primarily in generating a detailed picture of the impact of disease and treatment on patients' QL, then a multidimensional classic approach may be the most appropriate. If, on the other hand, the intent is to adjust survival data or to generate cost-utility information, then the utility approach to QL assessment would be preferred. Of course, the use of one approach does not preclude the other. With sufficient resources, both approaches could be incorporated within a single study. Review of Selective Quality of Life Instruments By means of a Medline search (key words: cerebrovascular disorders, stroke outcome, quality of life, health status, reliability, validity), reference tracing, and a recent bibliography,38 we identified nine generic QL measures that could be used in stroke research. These include the COOP Charts39,40; the McMaster Health Index Questionnaire41-45; the Nottingham Health Profile46-50; the Sickness Impact Profile51-56; the Medical Outcomes Study 20-Item Short-Form Health Survey57-59; the Karnofsky Performance Status Scale ; the Quality of Life Index51'63'64; the Euroqo165,66; and the Quality of Well-being Scale.51'55'67-71 Only one stroke-specific QL measure, the Frenchay Activities Index,17'72 was located. Each of these measures has been evaluated in terms of its reliability and validity. Two types of reliability can be evaluated: homogeneity (or internal consistency) and stability. Homogeneity refers to the statistical coherence of the scale items and is commonly assessed by the Cronbach's a coefficient.73 This coefficient is based on the average correlation of items within a scale and reflects the extent to which the items measure a common entity. The stability of a self-rating scale can be assessed by test-retest reliability, which is based on the measurement of the same person on two occasions with the same instrument. Interobserver reliability refers to the score agreement between different independent observers measuring a clinical phenomenon with the same instrument at the same time. Validity reflects the degree to which a scale measures what it is intended to measure. Three types of validity can be distinguished: content, criterion, and construct.71 Content validity involves a nonstatistical, subjective evaluation of the extent to which a measure covers the substantive issues of interest. For this review the content validity of the measures has not been evaluated. Rather, we have simply described the QL domains included in the scales. For criterion validity one examines the relation between QL scores and some superior criterion of QL. Because no golden standard exists for QL, this type of validity data is seldom available. More common are studies of the construct validity of a measure. Support of construct validity is provided if instruments purported to assess the same QL domains correlate significantly with one another (convergent validity), whereas instruments intended to assess different QL domains exhibit low correlations (discriminant validity). Finally, the clinical validity of an instrument can be assessed by examining the extent to which QL scores 1) are able to distinguish between patients with different diagnoses and between patients with differing degrees of disease severity within a given diagnosis; and 2) are able to detect clinically important changes in patients' health status over time. Although some authors consider this latter issue of responsiveness to be distinct from validity,74 others argue that it is simply another aspect (albeit an important one) of instrument validity.75 Table 1 provides a relatively detailed survey of each of the 10 QL measures, including 1) number of items, average time to complete, and self-completion versus observer rating; 2) QL domains assessed and scoring; and 3) psychometric data (i.e., reliability and validity). In the following paragraphs some global remarks will be made regarding each of these measures. The COOP scale comprises nine simple charts that are visually appealing because of the use of pictures. The scale was developed to routinely measure patients' perceived health and physical, emotional, and social functioning in a busy office practice. Studies have demonstrated its potential for generating new information on patients and for influencing communication and management in general practice and outpatient settings. Although the COOP scale is suitable in clinical situations, the use of one item to measure each life domain may limit its analytic value in stroke outcome studies. The McMaster Health Index Questionnaire (MHIQ) has 59 items measuring some physical symptoms (hearing, sight, headache), mobility, disabilities, quantitative aspects of social relations, social activities, and emotions (e.g., feelings of hopelessness and loss of control). Some characteristics of the MHIQ, such as the use of item skip patterns in the case of inapplicable questions and the use of different response categories, may be confusing for less-educated patients or patients with cognitive deficits. Not all reliability studies in terms of homogeneity and test-retest stability are convincing. In comparison to the MHIQ, the Nottingham Health Profile (NHP) is somewhat shorter and can be administered more quickly. The NHP consists of two parts. Part one primarily addresses serious health problems and particularly focuses on levels of disability, sleep disturbances, and subjective feelings of energy, pain, emotional distress, and qualitative aspects of social relations. The seven items of part two refer to the effects of health on seven life areas (e.g., social life, vocational function). The use of such a series of single items, in combination with the dichotomous response categories, restricts the range of possible scores and thus may be of limited analytic value. The subscales are in need of further validation. The Sickness Impact Profile (SIP) is a well-evaluated 136-item measure organized into 12 subscales. The subscales assessing communication, cognitive alertness, emotional behavior, and social functioning may be particularly relevant for use in stroke outcome studies. The physical dimension contains items measuring a broad range of ADL, mobility, and complex physical activities. Floor effects and especially ceiling effects, which are common to most specific disability scales, are therefore less likely. (Disability scales often address

4 only a narrow range of basic daily functions. As a result of this restricted range, health improvement or deterioration may not be detected because these changes can occur beyond the end points of the scale.) Because of its length the SIP is primarily suitable for cross-sectional studies. Since reliability data indicate poor reproducibility, it is recommended not to analyze item responses but to use aggregated category scores for description and analysis. The SIP defines QL only in behavioral terms, while other measures also incorporate aspects of subjective health. When using such a purely behavioral measure it may be advisable to add questions on feelings of well-being, health, happiness, and life satisfaction. The Medical Outcomes Study 20-Item Short-Form Health Survey (MOS-20) is intended for use in largescale outcome studies. The MOS-20 emphasizes mental health, health perception, and physical functioning. The mental health scale includes clearly defined items on distress and mood. In comparison with the other QL instruments, much attention is paid to general health perceptions and subjective evaluation of health status. The physical functioning scale primarily assesses complex physical activities. Only one item refers to ADL. Since limitations in self-care are important in stroke, this may be problematic in stroke outcome studies. When using the MOS-20 it is therefore worth considering the addition of a specific disability scale, such as the Barthel ADL Index.77 Future studies should establish the usefulness of the MOS-20 in stroke populations. Evaluations of a revised, 36-item version of the MOS instrument are currently underway.78 The Frenchay Activities Index (FAI) has been developed specifically for use with stroke patients. The FAI measures lifestyle in terms of more complex physical activities and social functioning. The index is easy to complete and can be administered in a few minutes. The item weights take potential sex differences in the scores into account. If one is primarily interested in measuring outcomes with an emphasis on disabilities, the FAI may be quite appropriate. However, additional studies of the reliability and validity of this instrument are necessary. The Karnofsky Performance Status Scale (KPSS) is a rating scale scored by an observer or interviewer. The KPSS was originally designed as an outcome measure in cancer research but has also been used in other chronically ill populations, including stroke.62 The ratings are narrowly based on physical disabilities and need for care. The measure incorporates death as a relevant end point. This is an important measurement feature, particularly in longitudinal stroke studies. If death is omitted, an increasing rate of mortality may be paradoxically related to a higher level of QL in the group of survivors.71 The scale lacks a standardized observational procedure, although some suggestions in this regard have been made.60 The QL-Index was developed to assess QL in cancer or other chronic diseases and encompasses the major life domains. Although this five-item scale, like the COOP Charts, is less precise, it may be useful in stroke research when one lacks resources for a more comprehensive patient-based evaluation or when the assessment of QL is not the main focus of the study. The reproducibility and the validation of the ratings with patients' self-reports need further attention. When us- de Haan et al Measuring Quality of Life 323 ing the QL-Index it is advisable to standardize the interview or observational procedure. The Euroqol is a utility scale that is still under development. The scale comprises five items to classify patients into various health profiles. The ordinal response scales are somewhat imbalanced, and the scale lacks the possibility of scoring the level of disability of patients in wheelchairs. Currently, the utilities of only some of the possible health profiles are assessed. The Quality of Well-being Scale (QWBS) is probably the most widely used utility scale. The utilities of health states and symptoms of the target patient group, measured respectively on three functional scales and a symptom/problem complex, have been assessed in open populations and in various (nonstroke) patient groups. The functional scales focus on disabilities and social functioning. The problem/symptom complex reflects the physical and, to some extent, the emotional dimension of QL and includes relevant stroke outcomes (e.g., cognitive dysfunctions, communication disorders, emotional distress, paralysis). Similar to the KPSS, the QWBS incorporates death as an end point. Although the scale requires extensive interviewer training, it has been used in a wide range of settings. Additional studies of reliability, however, are recommended. To summarize, with the exception of the QL-Index, the MOS-20, and, on the level of aggregate scores, the SIP, the QL measures reviewed are in need of further evaluation of their reliability and validity. Furthermore, with the exception of the QL-Index, none of the scales reviewed has been satisfactorily evaluated for its responsiveness to change in patients' clinical status over time. Although evidence of an instrument's responsiveness may be relatively unimportant in cross-sectional descriptive studies, it is critical in evaluating the appropriateness of a given QL measure for use in prospective clinical trials in stroke. Conclusions and Recommendations In this article we have reviewed nine generic QL measures and one stroke-specific instrument. Although it was not possible to incorporate all the literature pertaining to these instruments into our review, we are nevertheless confident that the studies reviewed provide a representative picture of the strengths and weaknesses of the QL measures. A well-considered choice of suitable instruments should be based not only on the psychometric properties of the measures but also on the specific research questions, the relevance to the objectives of the study, the available resources, and the specific characteristics of the stroke patients under study. In selecting a suitable QL instrument some trade-off between the level of detail provided and feasibility seems to be inevitable. There are measures, such as the SIP and the MHIQ, that assess a wide range of QL domains in depth. However, in selecting a lengthy instrument one also has to consider its feasibility in terms of the available resources as well as patient and staff burden. The shorter instruments, such as the QL-Index, the COOP Charts, and the MOS-20, provide less detail but may be more feasible for use in prospective clinical trials. If the objective of the study is to assess general perceptions of health and well-being, measures such as the SIP and the FAI, which focus on the behavioral performance of the

5 324 Stroke Vol 24, No 2 February 1993 TABLE 1. Summary of Psychometric Properties of 10 Selected Quality of Life Scales* Clinical Reliability validity Scale Length/time (min)/rater Domains/scoring H TR IO RS DIF COOP Charts items/2-3/patient or Physical, emotional, social, NR interviewer role, global health, health change, pain, overall QL, social supportt/9 scale scores McMaster Health 59 items/20/patient Functional, emotional, ± NR ±.. ±.. Index social/3 summated subscale Questionnaire41-45 scores; standardized index value (item weights) Nottingham Health Profile items/5-15/patient Part 1: Physical, emotional, social, pain, energy, sleep/6 summated scale scores (item weights) Part 2: Effects on 7 life areas/scoring not clear (Part 1) (Part 2) NR Downloaded from by on November 3, 2018 Sickness Impact Profile51-56 Medical Outcomes Study57-59 Frenchay Activities Index17,72 Karnofsky Performance Status Scale25'60-63 Quality of Life Index51,63,64 Euroqol65,66 Quality of Well-being Scale51,55, items/20-30/patient 20 items/3-4/patient 15 items/<5/patient or caregiver 11-point rating scale/1/observer or interviewer 5 items/1-2/observer or interviewer 6 items (with 2 or 3 ordinal steps)/patients: <5; valuation of health states: 20/patient; health states are valued by general public 3 functional scales with 3 (or 5?) ordinal steps; 27 (no consistent reports) symptoms/patients: 12-17; general public: 7-13/patient or caregiver; health states are valued by general public or patient populations Ambulation, mobility, body care, social, emotional, communication, alertness, sleep, eating, home management, recreation, employment/12 summated category scores; 2 summated dimension scores; 1 summated total score (item weights) Physical, emotional, social, pain, role, global health/6 summated scale scores (item weights) Domestic chores, outdoor activities, leisure and work/3 subscale scores; 1 summated total score (item weights) Physical disability, need of carel1 score Physical, emotional, social/1 summated total score (subscale analysis possible) Utility scale: Mobility, self-care, daily activities, family and leisure activities, pain and discomfort, mood/1 weighted summarized score (health profiles of patients in terms of steps on the 6 domains; health profiles are valued by public preferences) Utility scale: Mobility, physical activity, social activity, symptom/problem complex/i weighted summarized score (patients are classified on the functional scales and the symptom/problem scale; health states are valued by public or patient preferences) (Aggregate) (Item level)... NR NR... NR NR, Substantial indications: reliability coefficients >0.80 for the homogeneity (H), test-retest (TR), and interobserver (IO) agreement (if a scale contains few items a coefficient >0.60 is also sufficient). When 10 is assessed by K, a coefficient >0.60 is considered as substantial.76 The responsiveness (RS) of the scale is demonstrated by the ability to detect health changes in patients over time. The scale is able to differentiate (DIF) patient groups. Convergent (CON) validity is demonstrated by significant correlations between the scale scores and instruments measuring the same or closely associated quality of life (QL) domains. Discriminant (DIS) validity is shown by low correlations between the scale scores and instruments measuring different QL dimensions.

6 de Haan et al Measuring Quality of Life 325 TABLE 1. Continued Construct validity CON DIS Further supportt Remarkst Positive relation with number of symptoms IO reliabilities show wide fluctuations (r= ). Charts are and chronic diseases. sensitive for effects of several diseases, but further studies are recommended Relation with global health, work absenteeism, place of residence, length of hospital stay. Sex differences in score profiles. Factor analytic support for construct validity. Relations with clinical measures and scores across demographic strata. Not all reliability studies are convincing (in population with rheumatoid disease: social scale, H=0.48, TR=0.51; physical scale, H=0.53). Scores on physical scale differentiate age groups and patient groups. Scale is claimed to be responsive to change, but supportive data are reported only for functional scale. Cut-off point for social scale is suggested (sensitivity=0.72, specificity=0.77). In part two TR<0.60 in 2 social subscales (evaluated in osteoarthrosis population). Average score in stroke population did not change despite improvements in physical ability. Items focus on serious health problems, which probably limits the responsiveness of the scale among less ill patients. Only convergence data for emotion scale. Reliability of scores is high but poor on item level (r=0.50). Measures relatively gross changes on group level, is probably more responsive to deterioration than to improvement, but is not sensitive enough for subtle individual changes. Health perception is related to several sociodemographic variables. Floor effect among severely ill patients may limit responsiveness to health deterioration. Two overlapping items on role functions. Downloaded from by on November 3, Factor analytic support for the domains.... Related to health perception, number of symptoms, and pain. Scores predict duration of survival in lung cancer.... Can value health states in the community by postal surveys (although a number of inappropriate responses have to be taken into account). Results of 3 studies show close relations between absolute values and ranking orders of health states. On group level there is logical consistency of valuation. Well-being scores are logically related to number of chronic conditions, symptoms, physician contacts, dysfunctional status, age, subjective well-being, and employment status. Public preferences for various health states do not change over time and show no systematic differences between demographic groups, providers, students, and administrators. Poststroke scores in comparison with prestroke levels of function showed health deterioration on the various subscales. Change scores were based on retrospective data. Scale scores discriminated male but not female patients according to severity of their stroke. Two studies in noncancer populations failed to meet K standards and showed substantial disagreement between ratings of physicians and patients. Lacks standardized observational procedure. There is no standardized observational procedure. Scoring categories are very broadly defined. Scale is not suitable for relatively healthy patients. A self-administration version is available. Scale is in development. Translation of measurements into quality-adjusted life years has not been operationalized. Fourteen of the theoretically 216 health states have been valued by the general public. In 1 study (only) 2 health states were presented twice, with high TR correlation. Five symptoms lack empirical weights. Scale is well evaluated on validity, but further analysis of 10 reliability and homogeneity is recommended. Data do not consistently support responsiveness of the scale to change. Scale allows calculation of quality-adjusted life years., Moderate indications: data of study reviewed are not consistent, results are summarily presented, or studies suffer from methodological flaws; -, no clear indications: data do not support the criteria; NR, not relevant in view of the characteristic of the scale;..., no data available. *For clarity and brevity the table is presented without original data of the studies. tadditional and explanatory remarks are only made in case the studies reviewed do not (consistently) meet the psychometric criteria. *Since there are a confusing number of other names for validity (e.g., biological, concurrent, statistical) and many researchers do not use the same classification system of validity types, we have summarized additional evidence of validity in neutral terms so that the reader can make his own judgment of what type of validity is supported.

7 326 Stroke Vol 24, No 2 February 1993 patient, may be less suitable than scales such as the MOS-20 and the NHP. In some cases, especially in longitudinal studies, it may be desirable to incorporate survival data in the measurement and analysis strategy. Stroke patients are typically elderly and have a range of comorbid conditions.79 Considering the frailty and advanced age of this population, QL measures should be evaluated in terms of their simplicity and their capacity to differentiate the effects of age from those of illness. A thorough analysis of item content is therefore also a prerequisite for the selection of a suitable scale. For example, in the functional domain it is sometimes difficult to differentiate physical disabilities resulting from stroke from dysfunctions related to the normal aging process. A possible contamination effect can also occur in subscales measuring psychological distress or depression. Items directed to the physical manifestations of emotions can also reflect somatic outcomes of stroke, and symptoms such as insomnia, anorexia, and anergia may not only be indicators of an affective state but can also be related to the aging process. Isolation or reduced social activities can be present as a result of physical disability without the presence of depression. To overcome the conceptual and methodological problems surrounding the assessment of QL in stroke, we propose the following directions for future discussion and study: 1) the conceptualization of QL and its relation to the ICIDH should be further clarified; 2) emphasis should be placed on further psychometric evaluation of existing generic QL measures rather than on generating new instruments; 3) in view of the importance of incorporating QL measures in intervention studies, supplementary data should be generated on the responsiveness of the instruments to within-patient changes over time; 4) stroke outcome studies are also in need of a practical disease-specific QL scale that focuses on the specific problems of stroke patients. This instrument must be able to differentiate the effects of stroke from those of aging and to avoid contamination effects of somatic disturbances on measures of emotional distress; 5) for any QL instrument used, data on a nonstroke control sample with an age distribution similar to that of the stroke group under investigation should be collected; 6) with regard to the utility approach to QL assessment, important methodological issues concerning the methods for generating utilities and convergence with classic QL measures still need to be resolved. Acknowledgment The authors wish to acknowledge Patrick Bossuyt of the Department of Clinical Epidemiology and Biostatistics of the Academic Medical Center, University of Amsterdam for his helpful advice in preparing this manuscript. References 1. Task Force on Stroke Impairment, Task Force on Stroke Disability, and Task Force on Stroke Handicap: Symposium recommendations for methodology in stroke outcome research. Stroke 1990; 21(suppl II):II-68-II Van Knippenberg FCE, de Haes JCJM: The quality of life of cancer patients: A review of the literature. Soc Sci Med 1985;20: Hornquist JO: The concept quality of life. Scand J Soc Med 1982; 10: Guyatt GH, Jaeschke R: Measurements in clinical trials Choosing the appropriate approach, in Spilker B (ed): Quality of Life Assessments in Clinical Trials. New York, Raven Press, Publishers, 1990, pp Wenger NK, Mattson ME, Furberg CD, Elinson J: Preface, in Wenger NK, Mattson ME, Furberg CD, Elinson J (eds): Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. Washington DC, Le Hacq, 1984, pp xi-xv 6. Ahlsio B, Britton M, Murray V, Theorell T: Disablement and quality of life after stroke. Stroke 1984;15: Aaronson NK: Quality of life assessments in clinical trials: Methodological issues. Controlled Clin Trials 1989;10:195S-208S 8. Gough IR, Furnival CM, Schilder I, Grove W: Assessment of the quality life of patients with advanced cancer. Eur J Cancer Clin Oncol 1983;19: Aaronson NK: Quality of life: What is it? How should it be measured? Oncology 1988;2: World Health Organization: International Classification of Impairments, Disabilities and Handicaps. Geneva. World Health Organization, World Health Organization: The first ten years of the World Health Organization. Geneva, World Health Organization, Lawrence L, Christie D: Quality of life after stroke: A three-year follow-up. Age Ageing 1979;8: Viitanen M, Fugl-Meyer KS, Bernspang B, Fugl-Meyer AR: Life satisfaction in long term survivors after stroke. Scand J Rehabil Med 1988;20: Labi MLC, Phillips TF, Gresham GE: Psychosocial disability in physically restored long-term stroke survivors. Arch Phys Med Rehabil 1980;61: Patrick DL, Deyo RA: Generic and disease-specific measures in assessing health status and quality of life. Med Care 1989;27(suppl 3):S217-S Guyatt G, Feeny D, Patrick D: Issues in quality-of-life measurement in clinical trials. Controlled Clin Trials 1991;12(suppl): 81S-90S 17. Holbrook M, Skilbeck CE: An activities index for use with stroke patients. Age Ageing 1983;12: Liang MH, Katz JN, Ginsburg KS: Chronic rheumatic disease, in Spilker B (ed): Quality of Life Assessments in Clinical Trials. New York, Raven Press, Publishers, 1990, pp Wenger NK, Furberg CD: Cardiovascular disorders, in Spilker B (ed): Quality of Life Assessments in Clinical Trials. New York, Raven Press, Publishers, 1990, pp Niemi M, Laaksonen R, Kotila M, Waltimo 0: Quality of life 4 years after stroke. Stroke 1988;19: Smith DS: Outcome studies in stroke rehabilitation: The South Australian Stroke Study. Stroke 1990;21(suppl II):II Robinson RG, Bolduc PL, Kubos KL, Starr LB, Price TR: Social functioning assessment in stroke patients. Arch Phys Med Rehabil 1985;66: Lomas J, Pickard L, Mohide A: Patient versus clinician item generation for quality-of-life measures. Med Care 1987;25: Elam JT, Beaver T, El Derwi D, Applegate WB, Graney MJ: Comparison of sources of functional reports with observed functional ability of frail older persons. (abstract) Gerontologist 1989; 29(suppl):308A 25. Hutchinson TA, Boyd NF, Feinstein AR, Gonda A, Hollomby D, Rowat B: Scientific problems in clinical scales, as demonstrated in the Karnofsky Index of Performance Status. J Chron Dis 1979;32: Hall JA, Epstein AM, McNeil BJ: Multidimensionality of health status in an elderly population: Construct validity of a measurement battery. Med Care 1989;27:S168-S Sprangers MAG, 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: McCusker J, Stoddard AM: Use of a surrogate for the SIP. Med Care 1984;22: Clarridge BR, Massagli MP: The use of female spouse proxies in common symptom reporting. Med Care 1989;27: Epstein AM, Hall JA, Tognetti J, Son LH, Conant L: Using proxies to evaluate quality of life: Can they provide valid information about patients' health status and satisfaction with medical care? Med Care 1989;27(suppl 3):S91-S Evans RL, Noonan WC, Bishop DS, Hendricks RD: Caregiver assessment of personal adjustment after stroke in a Veterans Administration medical center outpatient cohort. Stroke 1989;20: Habbema JDF, Bossuyt PMM, Dippel DWJ, Marshall 5, Hilden J: Analysing clinical decision analyses. Stat Med 1990;9:

8 33. Torrance GW: Utility approach to measuring health-related quality of life. J Chron Dis 1987;40: Eisenberg JM: Clinical economics: A guide to the economic analysis of clinical practices. JAMA 1989;262: Stoddard GL, Drummond MF: How to read clinical journals: 7. To understand an economic evaluation (part A). Can Med Assoc J 1984;130: Feeny D, Labelle R, Torrance GW: Integrating economic evaluations and quality of life assessments, in Spilker B (ed): Quality of Life Assessments in Clinical Trials. New York, Raven Press, Publishers, 1990, pp Drummond MF: Discussion: Torrance's 'utility approach to measuring health-related quality of life.' J Chron Dis 1987;40: Tilson H, Spilker B: Quality of life bibliography. Med Care 1990; 28(suppl):DS1-DS Nelson E, Wasson J, Kirk J, Keller A, Clark D, Dietrich A, Stewart A, Zubkoff M: Assessment of function in routine clinical practice: Description of the COOP Chart method and preliminary findings. J Chron Dis 1987;40(suppl 1):55S-63S 40. Nelson EC, Landgraf JM, Hays RD, Wasson JH, Kirk JW: The functional status of patients: How can it be measured in physicians' offices? Med Care 1990;28: Sackett DL, Chambers LW, MacPherson AS, Goldsmith CH, McAuley RG: The development and application of indices of health: General methods and a summary of results. Am J Public Health 1977;67: Chambers LW, Sackett DL, Goldsmith CH, MacPherson AS, McAuley RG: Development and application of an index of social function. Health Serv Res 1976;11: Chambers LW, MacDonald LA, Tugwell P, Buchanan WW, Kraag G: The McMaster Health Index Questionnaire as a measure of quality of life for patients with rheumatoid disease. J Rheumatol 1982;9: Chambers LW, Haight M, Norman G, MacDonald L: Sensitivity to change and the effect of mode of administration on health status measurement. Med Care 1987;25: Fortin F, Kerouac S: Validation of questionnaires on physical function. Nurs Res 1977;26: Hunt SM, McKenna SP, McEwen J, Backett EM, Williams J, Papp E: A quantitative approach to perceived health status: A validation study. J Epidemiol Community Health 1980;34: Hunt SM, McKenna SP, Williams J: Reliability of a population survey tool for measuring perceived health problems: A study of patients with osteoarthrosis. J Epidemiol Community Health 1981; 35: Hunt SM, McKenna SP, McEwen J, Williams J, Papp E: The Nottingham Health Profile: Subjective health status and medical consultations. Soc Sci Med 1981;15a: Hunt SM, McEwen J, McKenna SP, Backett EM, Pope C: Subjective health of patients with peripheral vascular disease. Practitioner 1982;226: Ebrahim S, Barer D, Nouri F: Use of the Nottingham Health Profile with patients after a stroke. J Epidemiol Community Health 1986;40: Tandon PK, Stander H, Schwarz RP: Analysis of quality of life data from a randomized, placebo-controlled heart-failure trial. J Clin Epidemiol 1989;42: Bergner M, Bobbitt RA, Carter WB, Gilson BS: The Sickness Impact Profile: Development and final revision of a health status measure. Med Care 1981;19: Deyo RA, Inui TS: Toward clinical applications of health status measures: Sensitivity of scales to clinically important changes. Health Serv Res 1984;19: MacKenzie CR, Charlson ME, DiGigioia D, Kelley K: Can the Sickness Impact Profile measure change? An example of scale assessment. J Chron Dis 1986;39: Read JL, Quinn RJ, Hoefer MA: Measuring overall health: An evaluation of three important approaches. J Chron Dis 1987; 40(suppl 1):7S-21S de Haan et al Measuring Quality of Life Brooks WB, Jordan JS, Divine GW, Smith KS, Neelon FA: The impact of psychologic factors on measurement of functional status. Med Care 1990;28: Stewart AL, Hays RD, Ware JE: The MOS short-form general health survey: Reliability and validity in a patient population. Med Care 1988;26: Stewart AL, Greenfield S, Hays RD, Wells K, Rogers WH, Berry SD, McGlynn AA, Ware JE: Functional status and well-being of patients with chronic conditions: Results from the Medical Outcomes Study. JAMA 1989;262: Bindman AB, Keane D, Lurie N: Measuring health changes among severely ill patients: The floor phenomenon. Med Care 1990;28: Schag CC, Heinrich RL, Ganz PA: Karnofsky Performance Status revisited: Reliability, validity and guidelines. J Clin Oncol 1984;2: Hyde L, Wolf J, McCracken S, Yesner R: Natural course of inoperable lung cancer. Chest 1973;64: Grieco A, Long CJ: Investigation of the Karnofsky Performance Status as a measure of quality of life. Health Psychol 1984;3: Monahan ML: Quality of life of adults receiving chemotherapy: A comparison of instruments. Oncol Nurs Forum 1988;15: Spitzer WO, Dobson AJ, Hall J, Chesterman E, Levi J, Shepherd R, Battista RN, Catchlove BR: Measuring the quality of life of cancer patients: A concise QL-index for use by physicians. J Chron Dis 1981;34: The Euroqol Group: Euroqol: A new facility for the measurement of health-related quality of life. Health Policy 1990;19: Essink-Bot M, Bonsel GJ, Van Der Maas PJ: Valuation of health states by the general public: Feasibility of a standardized measurement procedure. Soc Sci Med 1990;11: 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: Anderson JP, Bush JW, Berry CC: Internal consistency analysis: A method for studying the accuracy of function assessment for health outcome and quality of life evaluation. J Clin Epidemiol 1988;41: Anderson JP, Kaplan RM, Berry CC, Bush JW, Rumbaut RG: Interday reliability of function assessment for a health status measure: The quality of well-being scale. Med Care 1989;27: Kaplan RM, Anderson JP: The general health policy model: An integrated approach, in Spilker B (ed): Quality of Life Assessments in Clinical Trials. New York, Raven Press, Publishers, 1990, pp Kaplan RM, Bush JW, Berry CC: Health status: Types of validity and the Index of Well-being. Health Serv Res 1976;11: Wade DT, Langton Hewer R, Skilbeck CE, David RM: Stroke: A Critical Approach to Diagnosis, Treatment and Management. London, Chapman and Hall, 1985, p Cronbach LJ: Coefficient alpha and the internal structure of tests. Psychometrika 1951;16: Guyatt G, Walter S, Norman G: Measuring change over time: Assessing the usefulness of evaluative instruments. J Chron Dis 1987;40: Hays RD, Hadorn D: Responsiveness to change: An aspect of validity, not a separate dimension. Qual Life Res 1992;1: Landis RJ, Koch GG: The measurement of observer agreement for categorical data. Biometrics 1977;33: Wade DT, Collin C: The Barthel ADL Index: A standard measure of physical disability? Int Disabil Studies 1988;10: Ware JE, Sherbourne CD: The MOS 36-item short-form health survey (SF-36): 1. Conceptual framework and item selection. Med Care (in press) 79. Herman B, Leyten ACM, van Luijk JM, Frenken CWGM, Op de Coul AAW, Schulte BPM: Epidemiology of stroke in Tilburg, the Netherlands: The population-based stroke incidence register: 2. Incidence, initial clinical picture and medical care, and three-week case fatality. Stroke 1982;13: