Defining clinically meaningful change in health-related quality of life

https://doi.org/10.1016/S0895-4356(03)00044-1Get rights and content

Abstract

This article reviews current approaches to defining clinically meaningful change in health-related quality of life (HRQOL) and provides guidelines for their use. Definitions of clinically meaningful change are discussed. Two broad methods for identifying clinically meaningful change are contrasted: anchor-based methods and distribution-based methods. Anchor-based methods include cross-sectional approaches and longitudinal approaches. Distribution-based methods include those based on statistical significance, sample variability, and measurement precision. Anchor-based and distribution-based methods have advantages and limitations, and neither seems to be superior to the other. An integrated system for defining clinically meaningful change is recommended that combines anchor-based and distribution-based methods. We propose a new terminology for describing meaningful change derived from anchor-based and distribution-based methods.

Introduction

Health-related quality of life (HRQOL) assessments may benefit patients, clinicians, researchers, administrators, health maintenance organizations, and policy makers. Indeed, medical decision-making research has focused increasingly on HRQOL as an important variable. Furthermore, many randomized controlled trials now include HRQOL measures as valid and useful endpoints in addition to traditional clinical outcomes assessing mortality and morbidity. Although regulatory agencies do not currently require HRQOL for the approval of new drugs, pharmaceutical companies are increasingly including HRQOL data as part of their submissions for drug approval. Gotay and Moore [1] propose using quality of life (QOL) endpoints when (1) patients have chronic illness and need palliative care; (2) treatments are expected to be equivalent in efficacy, but one offers a QOL benefit; (3) a new treatment shows a small benefit that is offset by QOL deterioration; or (4) treatments differ in terms of short-term efficacy, but the overall failure rate is high. Authors of recent textbooks on QOL attribute the growing use of HRQOL assessment to several factors: (1) the aging of the population and its resulting increase in the prevalence of chronic disease [2]; (2) the more active role played by patients receiving medical care and their expressed interest in the nonclinical aspects of treatment, such as QOL [2]; and (3) the realization that many treatments for chronic diseases frequently fail to cure the disease, elevating the importance of QOL as a valuable outcome variable [3].

There are several well-known generic HRQOL instruments in wide use: SF-36 [4], the Sickness Impact Profile (SIP) [5], the Nottingham Health Profile [6], and the EuroQol [7]. In addition to generic measures of HRQOL, disease-specific measures of HRQOL have been developed for assessing HRQOL in a variety of diseases, including obesity [8], [9], [10], [11], [12], arthritis [13], [14], diabetes [15], [16], [17], [18], asthma [19], [20], [21], pulmonary disease [22], cancer [23], [24], [25], [26], epilepsy [27], and HIV [28]. Disease-specific measures of HRQOL are increasingly being used to evaluate medical treatments [29], [30], to make therapeutic decisions [31], and to allocate resources [32].

In its 1946 Constitution, the World Health Organization defined health as “a state of complete physical, mental, and social well-being and not merely the absence of disease or infirmity” [33]. This definition represented a departure from defining health solely in terms of death and disease. Early attempts to assess QOL, especially in the fields of cancer, arthritis, and heart disease, were designed to elicit evaluation about QOL from medical providers [34]. Today, most QOL instruments are based on ratings made by patients and have a wide range of applications.

A number of terms have been used interchangeably, including functional status, health status, quality of life, and health-related quality of life [35], [36]. Functional status typically refers to the performance of, or the capacity to perform, regular daily activities [37]. In contrast, health status is typically used to represent the patient's subjective (or perceived) appraisal of his/her state of physical and mental health [35].

A key distinguishing feature of QOL is the incorporation of patient values, judgments, and preferences [38]. Revecki et al. [35] characterize QOL as the subjective experiences, states, and perceptions relating to one's overall well being. It includes aspects of the physical, psychological, social, economic, and political environment. Testa and Simonson [39] define HRQOL as the “physical, psychological and social domains of health, seen as distinct areas that are influenced by a person's experiences, beliefs, expectations and perceptions” (p. 835).

An international group of investigators suggests six fundamental dimensions of HRQOL: physical functioning, psychological functioning, social functioning, role activities, overall life satisfaction, and perceptions of health status [40]. Testa and Simonson [39] argue that each domain of health can be measured in objective and subjective dimensions. Whereas the objective dimension serves to define a patient's degree of health, the patient's subjective evaluation serves to translate that health status into the actual QOL experienced. Thus, two patients with identical health status may have very different QOL depending on their subjective experiences, expectations, and perceptions regarding health.

When assessing physiologic measures such as blood pressure, experienced health care providers are usually adept at interpreting what constitutes a clinically meaningful difference. Repeated experience with measures, be they physiologic or measures of HRQOL, allows clinicians to make meaningful interpretations of results. However, the plethora of HRQOL instruments available, each with different units of measurement, makes for interpretation challenges. Juniper et al. [41] note, “the meaning of a change in score of 1.0 on a quality of life instrument is less intuitively apparent, not only because it has no familiar units, but also because health professionals seldom use quality of life measures in clinical practice” (page 81). Furthermore, clinically important differences may differ across groups of patients defined by diseases, conditions, levels of severity, socioeconomic status, and nationality [42].

The minimal important difference has been defined as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate, in the absence of troublesome side-effects and excessive cost, a change in the patient's management” [43] (page 408). Osoba et al. [44] point out that determining clinically meaningful differences is especially important because small numerical differences in mean HRQOL scores might give statistically significant results when large sample sizes are used, but statistical significance is not equivalent to clinical significance.

From the point of view of the patient, a meaningful change in HRQOL may be one that results in a meaningful reduction in symptoms or improvement in function. In contrast, a meaningful change for the clinician may be one that indicates a change in the therapeutic treatment or in the prognosis of the disease. In defining clinically meaningful change, these perspectives may not always be in agreement. Osoba et al. [44] also describe the societal and institutional perspectives for defining clinically meaningful change. The societal perspective takes into account the degree of importance on a population level, where small differences might be important because of the large number of persons who might be affected. The institutional perspective may focus on that degree of change required to influence health care policies.

Gill and Feinstein [38] argue that unless assessments of HRQOL come directly from the patient, investigators are not measuring HRQOL. They recommend two practices that facilitate patient input into the assessment process: (1) Augment the instrument with supplemental items that patients can add so that issues important to them are addressed, and (2) invite patients to rate the importance of the problems in addition to the severity. Although the idea of “individualizing” HRQOL scales has a certain intuitive appeal, it can create substantial problems in terms of psychometric performance. Individual weighting of questions can substantially alter the correlation between scale items and thereby change the reliability of the scale. Further, Streiner and Norman [45] show that individual weights can change the magnitude and direction of the correlation of the scale with other measures.

An important issue with respect to perspective is whether inferences about clinically meaningful change are being made with respect to individuals or groups [46]. Samsa et al. [42] note that even groups with negligible mean changes in HRQOL likely contain individual patients whose improvement is noteworthy. Individual variability from the group perspective is seen as random variation associated with measurement error. Inferences at the group level are likely to be informative with respect to the comparison of treatments or decisions regarding public health policy. In contrast, inferences at the individual level are most relevant to individual clinical treatment decisions. Another potential difference between the individual perspective and the group perspective is in terms of the magnitude of change necessary to be considered meaningful. Relatively modest improvements at the individual level may be considered clinically important when considered at the group level.

Hageman and Arrindell [47] argue that the distinction between the individual and group perspective has implications in terms of how decisions about clinically meaningful change should be made. They propose a reliable change index for determining individual improvement or deterioration, called the RCINDIV. They state that the determination of outcome for the group should not be based on a simple aggregation of individual outcomes yielding some overall percentage because of the problem of inflated misclassification rates across multiple subjects. They contend that individual- and group-level analyses require different statistical approaches. They propose a group-based formula for determining the proportion of individuals that have changed and argue that this group-based percentage best reflects what the correct outcome of treatment would be at the group level.

Beaton et al. [48] emphasize the role of multiple perspectives in interpreting change. They propose a three-dimensional cube into which studies of responsiveness can be categorized based on (1) whether the analysis is at an individual or group level, (2) which scores are to be contrasted (i.e., within individuals or between-groups), and (3) what type of change is being assessed (e.g., minimum potentially detectable, observed change detectable given the measurement error of the instrument, or observed changes in those deemed to have had an important change by their own assessment or by the clinician's assessment).

Section snippets

Anchor-based approaches

One method of determining clinically meaningful changes in HRQOL is by comparing measures of HRQOL to other measures or phenomena that have clinical relevance. Lydick and Epstein [49] have called this method “anchor-based” interpretation. Anchor-based approaches have been used to determine clinically meaningful change via cross-sectional and longitudinal methods. A summary of anchor-based methods, along with the advantages and disadvantages associated with each, is presented in Table 1.

Recommendations

The question of whether to use anchor-based or distribution-based methods for determining clinically meaningful change has received considerable attention and debate. Arguments favoring one position or the other have too frequently been framed in the context of an “either or” argument. The proponents of one approach have occasionally given lip service to the merits of the other approach. However, there are only a handful of approaches that have attempted to systematically integrate information

Acknowledgements

Financial support for this project was provided by Bristol-Myers Squibb, Princeton, NJ.

References (110)

  • G Stucki et al.

    Relative responsiveness of condition specific and health status measures in degenerative lumbar spinal stenosis

    J Clin Epidemiol

    (1995)
  • J.A Husted et al.

    Methods for assessing responsiveness: a critical review and recommendations

    J Clin Epidemiol

    (2000)
  • G.R Norman et al.

    Methodological problems in the retrospective computation of responsiveness to change: the lesson of Cronbach

    J Clin Epidemiol

    (1997)
  • D.E Beaton et al.

    Evaluating changes in health status: reliability and responsiveness of five generic health status measures in workers with musculoskeletal disorders

    J Clin Epidemiol

    (1997)
  • K Wyrwich et al.

    Further evidence supporting an SEM-based criterion for identifying meaningful intra-individual changes in health-related quality of life

    J Clin Epidemiol

    (1999)
  • N.S Jacobson et al.

    Psychotherapy outcomes research: methods for reporting variability and evaluating clinical significance

    Behav Therapy

    (1984)
  • N.S Jacobson et al.

    Toward a standard definition of clinically significant change

    Behav Therapy

    (1986)
  • L Christensen et al.

    A method of assessing change in a single subject: an alteration of the RC index

    Behav Therapy

    (1986)
  • D Cella et al.

    What is clinically meaningful change on the Functional Assessment of Cancer Therapy-Lung (FACT-L) questionnaire? Results from Eastern Cooperative Oncology Group (ECOG) Study 5592

    J Clin Epidemiol

    (2002)
  • G Stucki et al.

    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)
  • C.C Gotay et al.

    Assessing quality of life in head and neck cancer

    Qual Life Res

    (1992)
  • R Berzon

    Understanding and using health-related quality of life instruments within clinical research studies

  • P.M Fayers et al.

    Quality of life: assessment, analysis and interpretation

    (2000)
  • J.E Ware et al.

    The MOS 36-item short-form health survey (SF-36)

    Med Care

    (1992)
  • M Bergner et al.

    The Sickness Impact Profile: development and final revision of a health status measure

    Med Care

    (1981)
  • S.D Mathias et al.

    Assessing health-related quality-of-life and health state preference in persons with obesity: a validation study

    Qual Life Res

    (1997)
  • R.L Kolotkin et al.

    Assessing impact of weight on quality of life

    Obes Res

    (1995)
  • R.L Kolotkin et al.

    Development of a brief measure to assess quality of life in obesity

    Obes Res

    (2001)
  • C Le Pen et al.

    ‘Specific’ scale compared with ‘generic’ scale: a double measurement of the quality of life in a French community sample of obese subjects

    J Epidemiol Commun Health

    (1998)
  • J.F Fries et al.

    Measurement of patient outcome in arthritis

    Arthritis Rheum

    (1980)
  • R.F Meenan et al.

    The content and properties of a revised and expanded Arthritis Impact Measurement Scale Health Status Questionnaire

    Arthritis Rheum

    (1992)
  • G.S Hammond et al.

    Measurement of health status in diabetic patients: diabetes impact measurement scales

    Diabetes Care

    (1992)
  • J.T Fitzgerald et al.

    Development and validation of the Diabetes Care Profile

    Eval Health Prof

    (1996)
  • K Meadows et al.

    The Diabetes Health Profile (DHP): a new instrument for assessing the psychosocial profile of insulin requiring patients–development and psychometric evaluation

    Qual Life Res

    (1996)
  • The DCCT Research Group

    Reliability and validity of a Diabetes Quality of Life Measure (DQOL) for the Diabetes Control and Complications Trial (DCCT)

    Diabetes Care

    (1988)
  • E.F Juniper et al.

    Evaluation of impairment of health related quality of life in asthma: development of a questionnaire for use in clinical trials

    Thorax

    (1992)
  • P.W Jones et al.

    A self-complete measure of health status for chronic airflow limitation

    Am Rev Respir Dis

    (1992)
  • G.H Guyatt et al.

    A measure of quality of life for clinical trials in chronic lung disease

    Thorax

    (1987)
  • N.K Aaronson et al.

    The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology

    J Natl Cancer Inst

    (1993)
  • D.F Cella et al.

    The Functional Assessment of Cancer Therapy scale: development and validation of the general measure

    J Clin Oncol

    (1993)
  • J.C.J.M De Haes et al.

    Measuring the quality of life of cancer patients: the Rotterdam Symptom Checklist (RSCL): a manual

    (1996)
  • O Devinsky et al.

    Development of the quality of life in epilepsy inventory

    Epilepsia

    (1995)
  • S.A Bozzette et al.

    A Perceived Health Index for use in persons with advanced HIV disease: derivation, reliability, and validity

    Med Care

    (1994)
  • R Jaeschke et al.

    Quality of life instruments in the evaluation of new drugs

    Pharmacoeconomics

    (1992)
  • M.A Testa et al.

    Interpreting pharmacoeconomic and quality-of-life clinical trial data for use in therapeutics

    Pharmacoeconomics

    (1992)
  • D.L Patrick et al.

    Health status and health policy: quality of life in health care evaluation and resource allocation

    (1992)
  • Constitution of the World Health Organization

    (1946)
  • D Karnofsky et al.

    The clinical evaluation of chemotherapeutic agents in cancer

  • D.A Revicki et al.

    Recommendations on health-related quality of life research to support labeling and promotional claims in the United States

    Qual Life Res

    (2000)
  • M Bergner

    Qualify of life, health status, and clinical research

    Med Care

    (1989)
  • Cited by (1030)

    View all citing articles on Scopus

    Portions of this paper were presented at the Sixth Annual International Meeting of the International Society for Pharmacoeconomics and Outcomes Research (Arlington, Virginia), May 2001.

    View full text