Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
17 Assessing Health-Related Quality of Life Outcomes Donald L. Patrick The hope and promise of the Effectiveness Initiative to curtail escalating health care costs remind me of the time an airline pilot made an urgent announcement during a transatlantic flight. His voice suddenly came over the public address system, and he said, "Ladies and gentlemen, I have two pieces of news for you. One of them is good, and one of them is not so good. First I'll tell you the bad news. The bad news is that we are lost. We don't have any idea of where we are. But, as I told you, there is good news, too. The good news is that we have a 200-mile-an-hour tail wind. In other words, we don't know where we're going, but we're getting there awfully fast." ~ think this story nicely describes our current situation. Rapidly rising costs are hastening our attempts to use whatever means possible to find solutions. The Effectiveness Initiative is one means of turning bad news into good news. I wish to address four methodological issues involved in the assessment of health status outcomes: (a) selection of relevant outcomes; (b) use of generic and disease-specific measures; (c) progress toward short, reliable, valid, and responsive measures; and (d) methods for interpreting observed changes in measures. Before reviewing these issues, however, I would like to identify two challenges to our reliance on outcomes assessment for controlling health care costs. First, I would remind us that health services are only one determinant of health status (1~. When we talk about effectiveness of health care in terms of health status outcomes, we cannot forget that socioeconomic, political, and cultural systems have diverse and powerful influences on outcomes (2~. Effectiveness of medical care is our focus, but the larger sociocultural context influences both provider and patient reports of outcomes. Effectiveness is often in the eye of the beholder; patient expectations range from efficacious 137
138 EFFECTIVENESS AND O UTCOMES IN HEALTH CARE TABLE 1 Combinations of Cost and Effectiveness Outcomes Quality of Life Outcomes + Cost of Treatment + Worse quality of life Higher cost Better quality of life Higher cost Worse quality of life Lower cost Better quality of life Lower cost treatments that "cure" to "hugs from the doctor." Patient expectations, in fact, may well exceed our ability to provide the services that produce expected outcomes. A second cautionary note is that the increase in health technologies makes cost containment extremely difficult. Our decisions concerning cost and effectiveness outcomes can be described in Table 1. There are four different combinations of cost and effectiveness outcomes. Ideally, new technologies such as pharmaceuticals will produce better quality of life outcomes at reduced cost (lower right quadrant). Medical treatment for back pain and new drugs for benign prostatic hypertrophy might produce similar results. Rationing, capping reimbursement, and other methods of cost containment may produce outcomes in the lower left quadrant, that is, worse quality of life and lower costs. The upper left quadrant (higher cost, worse quality of life) is obviously to be avoided, although life-extending treatments might well fall into this category. Most technological innovations probably fall in the upper right quadrant, that is, better quality of life outcomes at higher cost. Technological advance clearly challenges our initiatives to maintain or lower health care costs. QUANTITY AND QUALITY OF LIFE Quantity and quality of life are distinct but related concepts used to evaluate the present and future state of a person or group of people (3~. Taken together, quantity and quality should represent a complete picture of the person or group. Quantity of life is assessed in terms of length of survival. For example, survival time is the number of days a patient lives after undergoing heart transplantation or while receiving drugs such as azidothymidine (AZT, a treatment for AIDS). Although such drugs may prolong the life of patients, they may produce concurrent toxic effects. It is
DEVELOPMENT AND USE OF OUTCOMES MEASURES 139 easy to assess quantity of life accurately in retrospective studies, but prog- nosis or duration of survival can only be estimated, often with considerable uncertainty. Furthermore, the value attached to a day of life differs from one person to another (4~. For some persons, death is preferable to the lowest states of functioning, such as coma or profound pain or depression. Quality of life has been assessed in a variety of different ways, including beauty in the landscape, close family life, environmental purity, and a spiritual understanding of existence. Health-related quality of life is more limited; it can be defined as the value assigned to the duration of life as modified by the social opportunities, perceptions, functional states, and impairments that are influenced by disease, injuries, treatments, or policy (5~. This definition covers five broad concepts on a continuum of health-related qual- ity of life, which is anchored at the top by an optimal value of 1.0 and at the bottom by a minimal value of 0. Specific dimensions of opportunity, perception, functional status, impairment, and survival fall along this continuum. See Table 2 for a more comprehensive description of the concepts and dimen- sions of health-related quality of life. Dimensions of the five concepts may be negatively or positively valued in relation to one another. The value assigned to the particular state of individuals or groups defines health-related quality of life. The time spent in that state or the probability of moving from one state to another (that is, prognosis) defines quantity of life. Thus, a complete representation of health-related quality of life involves specification of relevant states or combinations of dimensions, the values or preferences assigned to these states, and the duration or probability of duration in different states. This definition of health-related quality of life is similar to the health-state utilities approach developed over the last two decades (6~. SELECTION OF OUTCOMES The question arises whether there is a core set of outcomes that must be included in a quality of life assessment on theoretical, empirical, or judgmental criteria. Sol Levine has provided theoretical guidance in this area by focusing attention on two very important aspects of quality of life (74. The first is performance of the physical, psychological, and social functions and activities that people do in their everyday lives. The second is the satisfaction derived from performing these usual activities. Functional status and satisfaction with health are the core domains of health-related quality of life. There are many measures currently available to assess these health-re- lated quality of life dimensions. When we select these measures, we are attempting, in advance, to identify the potential effects of treatments as well as the potential side effects or unanticipated consequences of treatments. Table 3 contains a taxonomy of health-related quality of life measures.
140 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE TABLE 2 Concepts and Domains of Health-Related Quality of Life Concept and Domain Opportunity Social or cultural handicap Individual resilience Health perceptions Satisfaction with health function General health perceptions Functional status Social Limitations in usual roles Integration Contact Intimacy Psychological Affective Cognitive Physical Activity restrictions Fitness Impairment Subjective complaints S. 1gns Self-reported disease Psychological measures Tissue alterations Diagnoses Death and duration of life Definition/Indic ator Disadvantage because of health Capacity for health; ability to withstand stress; reserve Physical, psychological, social Self-rating of health; health concern, worry Acute or chronic limitations in social roles of student, worker, parent, household member Participation in the community Interaction with others Perceived feelings of closeness; sexual Psychological attitudes and behaviors, including distress and general well- being or happiness Alertness, disorientation; problems in reasoning Acute or chronic limitation in physical activity, mobility, self-care, sleep, communication Performance of activity with vigor and without excessive fatigue Reports of physical and psychological symptoms, sensations, pain, health problems, or feelings not directly observable Physical examination: observable evidence of defect or abnormality Patient listing of medical conditions or impairments Laboratory data, records, and their clinical interpretation Pathological evidence Clinical judgments after "all the evidence" Mortality; survival; longevity SOURCE: Patrick and Erickson (5~.
DEVELOPMENT AND USE OF OUTCOMES MEASURES TABLE 3 A Taxonomy of Health-Related Quality of Life Measures Approach Scores for analysis Single index number Useful for cost- effectiveness Profile of interrelated scores Battery of independent scores Wide range of outcomes Objective of application Generic: across conditions and populations Specific: disease, population, function, . . or cone Ton Weighting System Utility: preference weights from patients, . . prove .ers, or community Statistical: items weighted equally or from frequency of response 141 Strength Represents net impact May not be responsive Single instrument Effects on different outcomes possible Can select relevant outcomes Multiple comparisons possible Broadly applicable Summarize range of concepts May detect unanticipated effects More acceptable to respondents May be more responsive Interval scale Patient view incorporated Self-weighting samples More familiar techniques Appears easier to use Weakness Effects on different outcomes not possible May not be responsive Length often problem Cannot relate different outcomes to common scale Need to identify major outcome May not be responsive enough May not have focus of patient interest Length often problem Effects may be difficult to interpret Comparisons across conditions and populations not possible Difficulty obtaining weights May not differ from statistical weights that are easier to obtain May be influenced by prevalence SOURCE: Guyatt et al. (8~.
42 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE Measures can be classified according to the scores they produce for analy- sis, their objective in application, and the weighting system used in scoring (8~. SOURCES FOR ANALYSIS Indexes Measures such as the Quality of Well-being Scale (9), the Health Utilities Classification (6), and the Disability/Distress Scale (10) combine duration of life with specific dimensions of impairment, as well as physical, psychological, and social function. These measures yield a single index value, quality- adjusted life years (QALYs), that can be used to compare the cost per quality-adjusted life year gained from different health interventions. For example, the cost per QALY gained in 1986 U.S. dollars for coronary artery bypass surgery for left main coronary artery disease is $4,796, compared with $36,316 for neonatal intensive care for infants weighing 500 to 900 grams (6~. The effects of a particular treatment on a single index, such as QALYs, however, remain hidden. There is considerable controversy over whether such an index can represent health in a sufficient manner to detect changes and, indeed, interpret where those changes have taken place. Nev- ertheless, QALYs, or years of healthy life, are gaining acceptance, as exemplified by their inclusion in the Year 2000 Objectives for the Nation (11~. Profiles Other measures provide a profile of scores for different components of health-related quality of life. The Sickness Impact Profile (SIP) assesses sickness-related dysfunction in 12 different categories, producing a score for each category (12~. Various categories may be aggregated into a physical dimension score, a psychosocial dimension score, and an overall score with independent categories of work, eating, sleep and rest, home management, and recreation and pastimes. Similarly, Part I of the Nottingham Health Profile (NHP) contains 38 items that cover six domains of experience, yielding individual scores for each; Part II contains perceived problems in seven areas of daily life (13~. Unlike the SIP, however, the NHP does not yield an overall index score. The 59-item McMaster Health Questionnaire yields separate indexes for physical, emotional, and social function (14~. Measures developed originally at The RAND Corporation the 108-item Health Insurance Study battery and the 20- to 40-item Medical Outcomes Study short-form generic measures cover a wide spectrum of health concepts for use in general populations (15~. All these generic measures have been tested ex- tensively with different patient populations.
DEVELOPMENT AND USE OF OUTCOMES MEASURES Batteries 143 It is important to make a distinction between health profiles and batteries. Batteries are collections of health status measures with independent scores for each outcome. Specific measures of different health outcome domains are selected to make up an assessment battery. The recent evaluation of antihypertensive medications, sponsored by the Squibb Pharmaceutical Company, is an example of the battery approach (16~. These investigators selected, among others, the latest "best available" specific measures of general well- being, physical symptoms, and sexual dysfunction. Improvement in quality of life was assessed for three different anti-hypertensive agents on each of these independent measures. Luckily, all measures chosen for this study showed some improvement for the new drug under consideration. Often, our results are more mixed. The battery approach is an appealing assessment strategy because of the wide range and type of outcomes that can be assessed. A major outcome needs to be identified as the primary endpoint, however, to avoid conflicting findings and multiple comparisons of outcomes on different measurement scales. GENERIC AND SPECIFIC MEASURES Generic measures of health status are those that purport to be broadly applicable across types and severities of disease, across different medical treatments or health interventions, and across demographic and cultural subgroups. Visual analogue measures are designed to summarize a spectrum of the concepts of health or quality of life that apply to many different impair- ments, illnesses, patients, and populations. Disease-specific measures are those designed to assess specific diagnos- tic groups or patient populations, often with the goal of measuring responsiveness or "clinically important" changes. These are changes that clinicians and patients think are discernible and important, have been detected with an intervention of known efficacy, or are related to well-established physiological measures (such as grip strength for arthritis patients or spirometry for those with chronic obstructive lung disease) (17~. The term "disease-specific," here, refers to different adult patient populations with specific conditions or diagnoses. Not all specific measures are disease-related. They may be specific to conditions (for example, back pain or dyspnea), functions (for example, sexual or emotional function), or populations (for example, older adults or developmentally disabled children). Specific measures of single concepts or conditions are the most numerous of all within the health status field. These single-concept measures range from the assessment of specific symptoms
144 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE such as nausea and vomiting to more global concepts of life satisfaction. Mental health measures of depression, anxiety, and other emotional states, for example, are frequently used in clinical research for assessing individual concepts of psychological status. Numerical estimates of subjective pain, such as visual analogue scales, have gained a wide following, partly because of their high correlation with verbal rating scales and their simplicity (184. Visual analogue scales are also gaining popularity in the measurement of symptoms and functional status. Disease-specific measures, such as the Karnofsky Performance Status Scale for cancer (19), the American Rheumatism Association (ARA) functional classification for arthritis (20), and the New York Heart Association functional classification (21), have been used extensively over several decades. These measures were developed to meet the need for rapid classification of patients, and their sensitivity to small but clinically important change is limited. The ARA classification, for example, may detect large changes, such as those following hip replacement, but not smaller changes following drug therapy judged successful by other criteria. The popularity of disease-specific measures arises primarily from the need of clinical trials and practitioners to use scales that are most responsive to clinical changes that occur over time. Both discriminating improved from unimproved patients and accurately quantifying minimally important changes are particularly important measurement objectives for clinical research and clinical practice. Generic and disease-specific assessments alike are useful for clinical research, clinical practice, and policy analysis. Selection of different mea- sures depends on the objectives of measurement and the environment of the application. No single general-purpose measure is likely to meet all the needs of investigators and specific populations. Patients with different medical conditions have different concerns or place different emphasis on more generic concepts of health. Rather than develop disease-specific measures that incorporate generic concepts of health-related quality of life, the preferred strategy is to use standardized, generic instruments with disease-specific supplements. Generic measures permit the comparison of different populations and different programs, a most important objective for policy analysis and decision making. Use of generic measures is necessary for comparing benefits of different health interventions and allocating resources. Cumulative knowl- edge of health and quality of life outcomes using generic measures will establish the relative burden of different diseases and the relative merit of different interventions. By contrast, instruments specific to different diseases, conditions, and populations are critical for identifying important concerns of patients with particular conditions and for measuring small, clinically important changes from specific treatments. Experience with disease-specific measures to date
DEVELOPMENT AND USE OF OUTCOMES MEASURES 145 indicates their usefulness in discriminating among different conditions and in assessing changes. Rapid development of such instruments is to be expected for conditions and populations where few specific measures now exist, notably certain cardiovascular conditions, diabetes, gastrointestinal disorders, and sexually transmitted diseases including AIDS. SHORT, RESPONSIVE MEASURES The development of short-form generic measures for use in clinical practice is a welcome advance (22,23~. Undoubtedly these measures will be tested against more comprehensive and detailed generic instruments to identify information that may be lost using brief assessments. The need to measure an increasing array of physiological, physical, psychological, social, and general outcomes within a single investigation cannot be met unless measures are short, efficient to administer, and highly acceptable to investigators and respondents. Both generic and disease-specific measures will be assessed against these practical constraints. Self-administered, comprehensive measures that are sensitive to varia- tions in health care organization and medical practice are also needed. Short, generic health status measures (1 to 60 items) have been developed from longer versions based on minimal psychometric criteria for internal consis- tency reliability and content and construct validity (22,24~. The Short-Form Health Survey, derived from measures used in the Medical Outcomes Study, is currently being tested for its responsiveness, or ability to detect minimal changes of importance to interested parties. In the next decade, short-form generic measures need to be tested rigorously for their content validity, responsiveness, convergent and discriminant validity, and generalizability. "Short" and "comprehensive" can be conflicting goals for some applica- tions and populations. The full domain of health-related quality of life outcomes of interest to patients, providers, and payers simply cannot be represented in short measures. Some concepts, for example, cognitive function, sleep and rest behaviors, recreation, and satisfaction with health, are seldom represented in short, generic measures. These omissions may not seriously compromise the usefulness of short-form measures in relatively well popu- lations, but outcomes assessment in specific populations such as older per- sons, mentally ill persons, and institutionalized persons may require long- form assessments. Responsiveness, how well short-form measures detect subtle changes in behavioral and subjective health status, also requires testing and comparison with clinical measures. We can be encouraged by the data from the Medical Outcomes Study indicating that generic measures are very strong in detect- ing stable scores among a clinically stable group. Greater emphasis needs
146 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE to be placed on the assessment of responsiveness in comparing generic and disease-specific outcomes. Responsiveness is an important consideration in all serial applications of health status measures. Including items sensitive to change is critical to such assessments. Responsiveness of health status measures has been assessed using the relative efficiency statistic (a ratio of paired t statistics) (25), correlation of scale changes with other measures (26), receiver-operating characteristic curves (26), and a responsiveness statistic (ratio of minimal clinically important differences to variability in stable subjects) (171. Disease-specific measures with items selected to assess particular concerns or worded to attribute change to the condition of interest, for example, back pain in the modified SIP, may be particularly sensitive to within-subject changes and thus more responsive than generic measures, which contain items unrelated to change. CHANGES IN MEASURES Analyzing and interpreting changes in health status measures are problems in all longitudinal studies. These may be observational case studies, cohort studies, clinical trials, or health services evaluations. Changes in physiological measures such as blood pressure or cholesterol level may be interpreted in terms of prognostic implications and well-established or agreed cutoff points. Changes in generic health measures are more difficult to interpret, although even small changes in portions of such measures may be quite useful (for example, changes in physical mobility or self-care are meaningful in disabled populations). Changes in scores on the most general measures, such as health perceptions or global physical and psychosocial dimension scores, can be even more difficult to interpret. The net changes observed may reflect a large number of different transi- tions or combinations of transitions within the population. Single-score or aggregated measures can make it difficult to identify which items or components are responsible for the change. Net changes must also be distinguished from random or systematic changes (learning effects, rumination) that may occur independently of an intervention. Although changes in these scores may reflect sensitive effects, the relative magnitude of the change may be difficult to assess. For example, is a S-point difference more meaningful than a 3-point difference? Changes in disease-specific measures may be easier to interpret because they are more specific or more closely associated with changes in clinical measures of disease activity such as blood pressure or joint inflammation. Clinician or patient assessments of improvement, which are common measures of change or effects, may be more closely associated with changes in disease- specific measures than with those in generic health status measures (27~.
DEVELOPMENT AND USE OF OUTCOMES MEASURES A THEORETICAL MODEL FOR HEALTH STATUS AND QUALITY OF LIFE 147 A major challenge facing developers and users of health and quality of life measures is to establish a testable theory of the expected relationships among the different concepts and domains of health-related quality of life. The problem is not confined only to the relationship between physiological measures and behaviors or perceptions, for example, blood pressure and functional status. Measures of various dimensions, such as symptoms, psy- chological function, and satisfaction with health have been shown to be only loosely associated or entirely dissociated within the same sample (2~. Figure 1 depicts hypothesized relationships among different health-related quality of life concepts in a simple linear progression. The concepts are bounded by environmental determinants that influence disease and its con- sequences and by prognoses for improvement, maintenance, or decline in health-related quality of life. The simple causal model suggested in Figure 1 does not represent the complexity or strength of the expected relationships among health-related quality of life dimensions. For example, persons can have an asymptomatic disease that affects prognosis without affecting functional status, perceptions, or opportunity. A person with hypertension or hypercholesterolemia may not have restrictions in activity but may be disadvantaged by fear of a stroke or death. Similarly, not all persons with impaired physiological capacity experience psychological dysfunction. Persons with rheumatoid arthritis and congestive heart failure may have high satisfaction with their health and positive well-being. Figure 1 also indicates that the causal relationships among concepts can be reversed; for example, functional limitations and perceived health can be viewed as influencing impairment or physiological measures of chronic disease (29~. Reversing the causal chain permits testing of the variable course of chronic disease, whereby impairments may become permanent and lead to changes in behavior and perceptions that, in turn, influence symptoms or level of impairment. The notion of an interplay between the psyche and the body is as old as medicine itself. Models of psychophysiological processes such as disruption in the regulation of blood volume and control of blood pressure by the kidneys can be invoked to explain sociobehavioral influ- ences on disease processes. This evidence may not be sufficient to convince the most skeptical biomedical researcher, but the hypothesis has moved well beyond mere speculation. At present, researchers tend to approach the relationship among end points inductively, by collecting data and examining the correlation among measures. Little hypothetical or deductive reasoning is involved in either the selection of measures or analysis of results. Head-to-head comparisons of different
48 EFFECTIVENESS AND OUTCOMES IN HEALTH CARE Duration of Life ! Disease and Injury l 1 Impairments Environment Genetic Personal Social Economic Cultural Physical Physical, Psychological, & Social Function 1 Health Perceptions Opportunity for Health Prognosis Improvement Maintenance Decline Variable FIGURE 1 Relationships Among Health-Related Quality of Life Concepts SOURCE: Adapted from Patrick and Bergner (28~. dimensions will be important for determining the association between spe- cific disease states or disorders and their behavioral, perceptual, and social consequences. Increasing our understanding of these relationships will help us realize the potential of health-related quality of life measures for identi- fying the intervention strategies that address the most important concerns of patients, their families, clinicians, and society in general. CONCLUSIONS The use of health-related quality of life measures, especially those based on function, is likely to increase during the next decade. This increase,
DEVELOPMENT AND USE OF OUTCOMES MEASURES 149 however, is most likely to occur in clinical research and clinical practice (301. Unless the necessary political will, resources, data, and policy researchers coexist, there will be relatively little advance in the use of health status measures for decision making and policymaking. Policy research tends to rely on available national data, and currently these data provide limited information about health status. The Effectiveness Initiative will be successful only if it motivates data collection and methods that incorporate a broad spectrum of health outcomes (such as death, impairment, functional status, and perceptions) into a single assessment. Health and quality of life outcomes are what count, and these outcomes cannot be determined without appropriate and inclusive measures of health-related quality of life. I hope that motivation and resources will be found to help resolve meth- odological issues in the measurement of health status and quality of life. I also hope that government agencies, employers, and private providers will begin to collect health-related quality of life data on the constituents and populations they serve. Even if these data are imperfect or primitive, the effects of improving accessibility and quality of health care can only be assessed adequately in terms of the health-related quality of life of the nation. REFERENCES 1. Levine, S., Elinson, J., and Feldman, J. Does Medical Care Do Any Good? Pp. 394-406 in Handbook of Health, [Iealth Care, and the Health Professions. Mechanic, D., ed. New York: Free Press. 1983. 2. Patrick, D.L., Stein, J., Porta, M., et al. Poverty, Health Services, and Health Status in Rural America. Milbank Quarterly 66~1~:105-136, 1988. 3. Patrick, D. and Elinson, J. Sociomedical Approaches to Disease and Treatment Outcomes in Cardiovascular Care. Quality of Life 1 :53-65, 1984. 4. Patrick, D., Bush, J., and Chen, M. Toward an Operational Definition of Health. Journal of Health and Social Behavior 14:6-23, 1973. 5. Patrick, D.L. and Erickson, P. What Constitutes Quality of Life? Concepts and Dimensions. Clinical Nutrition 7:53-63, 1988. 6. Torrance, G.W. Measurement of Health State Utilities for Economic Ap- praisal. Health Economics 5:1-30, 1986. 7. Levine, S. The Changing Terrains in Medical Sociology: Emergent Concern with Quality of Life. Journal of Health and Social Behavior 28:1-6, 1987. 8. Guyatt, G., Feeny, D., and Patrick, D. Issues in Quality-of-Life Measurement in Clinical Trials. Controlled Clinical Trials. In press. 9. Bush, J.W. Relative Preference Versus Relative Frequencies in Health-re- lated Quality of Life Evaluation. Pp. 118-139 in Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. Wenger, N.K., Mattson, M.E., Furberg, C.D., et al., eds. New York: LeJacq Publishing, Inc., 1984.
150 EFFECTIVENESS AND OUTCOMES IN HEALTlI CARE 10. Rosser, R.M. A Health Index and Output Measure. Pp. 133-160 in Quality of Life: Assessment and Application. Walker, S.R. and Rosser, R.M., eds. Lancaster, England: MTP Press, 1988 11. U.S. Department of Health and Human Services, Public Health Service. Promoting HealthlPreventing Disease: Year 2000 Objectives for the Nation. Wash- ington, D.C.: Government Printing Office. 1989. 12. Bergner, M.B., Bobbitt, R.A., Carter, W.B., et al. The SIP: Development and Final Revision of a Health Status Measure. Medical Care 19:787-805, 1981. 13. McEwen, J. The Nottingham Health Profile. Pp. 95-112 in Quality of Life: Assessment and Application. Walker, S.R. and Rosser, R.M. eds., Lancaster, En- gland: MTP Press, 1988. 14. Chambers, L.W. The McMaster Health Index Questionnaire: An Update. Pp. 113-131 in Quality of Life: Assessment and Application. Walker, S.R. and Rosser, R.M ., eds. Lancaster, England: MTP Press, 1988. 15. Stewart, A.L., Greenfield, S., Hays, R.D., et al. Functional Status and Well- Being of Patients with Chronic Conditions. Journal of the American Medical Asso- ciation 262:907-913, 1989. 16. Croog, S.H., Levine, S., Testa, M.A., et al. The Effects of Antihypertensive Therapy on the Quality of Life. New England Journal of Medicine 314:1657-1664, 1986. 17. Guyatt, G., Walter, S., and Norman G. Measuring Change Over Time: Assessing the Usefulness of Evaluative Instruments. Journal of Chronic Diseases 40:171-178, 1987. 18. Scott, P.J. and Huskisson, E.C. 2:175-184, 1976. Graphic Representation of Pain. Pain 19. Kamofsky, D.A., Abelmann, W.H., Craver, L.F., et al. The Use of Nitrogen Mustards in the Palliative Treatment of Cancer. Cancer 1:634-656, 1948. 20. Steinbrocker, O., Traeger, C.H., and Batterman, R.C. Therapeutic Criteria in Rheumatoid Arthritis. Journal of the American Medical Association 140:659-662, 1949. 21. Criteria Committee of the New York Heart Association, Inc. Disease of the Heart and Blood Vessels: Nomenclature and Criteria for Diagnosis, 6th edition. Boston: Little, Brown, 1964. 22. Stewart, A.L., Hays, R.D., and Ware, J.E. The MOS Short-Form General Health Survey: Reliability and Validity in a Patient Population. Medical Care 26:724- 735, 1988. 23. Nelson, E.C., Wasson, J.H., and Kirk J.W. Assessment of Function in Routine Clinical Practice: Description of the COOP Chart Method and Preliminary Findings. Journal of Chronic Diseases 40 Supplement 1:SSS-63S, 1987. 24. Nelson, E.C. and Berwick, D.M. The Measurement of Health Status in Clinical Practice. Medical Care 27~3) Supplement:S77-S90, l9B9. 25. Liang, M.H., Larson, M.G., Cullen, K.E., et al. Comparative Measurement Efficiency and Sensitivity of Five Health Status Instruments for Arthritis Research. Arthritis and Rheumatism 28 :542-547,1985. 26. Deyo, R.A. and Centor, R.M. Assessing the Responsiveness of Functional Scales to Clinical Change: An Analogy to Diagnostic Test Performance. Journal of Chronic Diseases 11 :897-906,1986. 27. MacKenzie, C.R., Charlson, M.E., DiGioia, D., et al. Can the Sickness Impact
73? Profile Demure Change? An Example of Scale AssessmenL a/ ~r ~- e~ 39:429-433, 1986. 28. Patrick, D.L. ad Berliner, H. Measurement of Heals Status in the 1990s. Ann~' R~f~ ~ f ~r ~6 1 1: 1 65 - 1 83, 1 990. 29. Patrick, D.L. Comma: Patient Retails of Heals Slalus as Predictors of Physiologic Heals ~ Chronic Diesease. Aura/ If I 40:37S-40S, 19R7. 30. Bergner, at. Qualm of Lid, Health Slams, and Clinical Research. Ifs/ cars 27:S 148-S 156, 1989.
