Quality of Life Measurement Among Persons with Chronic Mental Illness: A Critique of Measures and Methods

Next | Previous | Table of Contents

2. Quality of Life Instruments, Design and Purpose

2.1 Domains of Life Quality

Health Related QoL, as distinct from general QoL, is conceptualized as those aspects of life quality or function which are impacted by one's health status. Within the context of the World Health Organization definition of health, namely that health is "a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity" (Mapper & Zembatu, 1986), QoL is indeed a broad construct, particularly as it pertains to mental health and mental illness. Further complicating matters, the impact of health status on QoL can be both quantitative and qualitative or subjective in nature. As will be discussed later, the relationship between changes in actual health status and individuals' appraisal of their QoL is heavily influenced by factors only indirectly related to health per se.

Nevertheless, most health theorists agree that depending on its intended use, a well rounded instrument should assess at least five life domains, including a person's biological, psychological, interpersonal, social and economic experience (cf., Rosenblatt & Attkisson, 1993). Accordingly, the most frequently utilized domains in our sample of quality of life instruments were Health, Self-Esteem/Well-being, Community/Productivity, Social/Love Relationships, and Leisure/Creativity (see Table 1 on the following page). Less frequently tapped were the domains of Family, Living Situation, Finances, Psychiatric Symptoms, and Religion. The sparse sampling by instruments in these latter categories was somewhat surprising to us given the dependence of chronic populations on financial and material support from both families and the larger society. Moreover, omission of life quality measurement in the area of spirituality and religion may reflect a certain scepticism concerning the importance of such beliefs in the lives of these individuals, and/or difficulty in evaluating this dimension of life.

It is worth noting that certain instruments, such as the Schedule for the Evaluation of Individual Quality of Life, utilize a unique methodology which allows patients to specify the life domains which impact most on their overall QoL. This approach is thought to maximize the personal relevance of the QoL measurement. It is not without its problems, however, particularly when one's intention is to compare results between people or programs, since the ratings refer to personally defined domains which naturally differ between individuals.

• Table 1 : Domains Measured by Selected QoL Instruments**

2.2 Measurement Specifications

Various factors influence the decision to select a particular instrument, particularly the resources required for its administration and the characteristics of its measurement. Table 2 (see next page) outlines the Data Collection Method, the Item Characteristics, the Data Source, and the approximate Completion Time for each of the 28 QoL instruments included in this review. Each of the Instrument Specifications Table headings are discussed in the sections which follow.

Administration Method: The two most widely used methods of data collection are the interview and the survey. Survey methods are less expensive to administer and can be conveniently managed (e.g., mailed out) but are not as versatile as the more expensive interview methods. Moreover, the response rate to mail-out surveys can be disappointingly low, to the point of casting doubt on any assumption, based on results, about the general population. Interviews, on the other hand, provide an opportunity for in-depth exploration of patient issues and attitudes, thereby improving the comprehensiveness and validity of findings. Subjects who hesitate or provide "snap" answers with little thought can be questioned more closely, and open ended items are likely to receive a fully considered response. An interviewer may also tailor an interview to suit a subject's particular life circumstances by skipping irrelevant items and explaining unclear terms or ideas. The latter is a particularly important consideration when working with a population with poor reading skills or impaired mental status. Overall, the flexibility and increased validity of interview methods is inevitably linked with costs associated with interviewer training and administration.

Nevertheless the suitability of an instrument, no matter what the method of administration, should be evaluated within the context of its use with a particular group of respondents and its ability to address questions intended to be answered by the study. Instruments ought to be pilot tested and difficulties which arise, either during the early stages of an interview based study or during initial survey completion, can be more thoroughly examined and steps taken to rework the assessment protocol. If cost containment is a concern, then plans to conduct an interview-based pilot and a survey-based implementation could cut costs by over 50% compared to those necessary for a full interview-based investigation and provide greater confidence in the instrument's suitability for use with its intended audience. One caveat is that while survey instruments can be used in an interview format the reverse is not always the case.

• Table 2 : Specifications Table**

Item Characteristics: Most theorists would agree that the domains which comprise life quality are numerous, since the overall construct of Quality of Life is multidimensional. There is less agreement, however, as to the role which individual preferences, perceptions of importance, and belief systems should play in the measurement of life quality. Jenkins (1992) contends that attempts should be made to avoid confounding subjective measures, such as personal expectations and desires, with more objective indicators of life quality. This position is disputed, however, by more subjectively oriented theorists who argue that individuals' belief systems are inextricably linked to the way in which illness impacts on patients' experiences of life quality. As we shall see, both perspectives have merit and raise important issues when it comes to evaluating life quality among the chronically mentally ill.

In order to maintain conceptual clarity between subjective and objective quality of life measurement and perhaps shed some new light on measurement issues, we have substituted the term "Subjective" with "Evaluative" so as to better reflect the role which thought and judgement play during patients' response to QoL items. For the purposes of this document, the terms "Objective" and "Evaluative" measures are defined as follows:

Objective Measures refer to items or scales which attempt to determine the amount or level of observable experiences within an individual's life. The assumption is that certain types of experiences (e.g., physical functioning, shelter, finances, social support etc.) are necessary components of life quality and measures of such experiences are direct indicators of patients' QoL.

Evaluative Measures refer to items or scales which attempt to measure individuals' evaluation of either the impact of illness on physical, social, and/or emotional functioning, or the satisfaction/importance of various life experiences. This approach differs from objective measurement in that evaluative judgements are appraisals of the subjective qualities of experiences and not quantification of actual experience. Our definition of evaluative measurement also includes the more global construct of well-being which is thought to tap both affective and cognitive (evaluative) domains (Pavot & Diener, 1993a).

Table 3 (see next page) lists the instruments which use at least one evaluative type of scale. Such scales allow respondents either to rate their subjective experiences in specific areas or evaluate their experiences on more global dimensions such as Social/Emotional Impact and Self-Esteem/Well-Being.

• Table 3 : Evaluative Dimensions of QoL Instruments**

Under this evaluative classification system, instruments tend to cluster into those using patient ratings of both Importance and Satisfaction/Fulfilment versus those using evaluative ratings of the impact of the illness on Physical/Function, Social/Emotion and Self-Esteem/Well-being dimensions. Almost without exception, instruments within the former cluster utilize a Gap Discrepancy Model to measure life quality. Methods based on a discrepancy model use a patient's rating of the personal importance of a domain and the adequacy of their current life experience in that domain to derive a subjectively weighted QoL score. Two other instruments use this model in a slightly different way by asking patients to rate their current health status in relation to their ideal state of health (SmithKline Beecham Questionnaire) or asking respondents to weight their satisfaction scores by rating the personal importance of the domains (QoL Index for Mental Health). Most other instruments listed use patients' evaluation of the impact of illness on their physical, emotional and social experiences without directly assessing the importance of these experiences to the patient. Instruments which do not use ratings of importance leave ambiguity about the personal significance of QoL measurement and the personal significance which any change in these areas might have on patients' lives.

Respondent: Typically, surveys and interviews are given to persons who are the subjects of inquiry. At times, however, it may be more advantageous to collect information from people who know the patient in either a personal or professional capacity. Such situations may occur when a patient's lucidity or judgement is impaired (Haneastad, 1990). While it seems likely that an evaluator would question the validity of results gathered from persons who are clearly unable to comprehend the purpose or meaning of evaluation items, the ways in which more subtle impairment of mental function influences patients' self-reports are not always as obvious. There is evidence that a patient's emotional state is highly correlated with evaluative QoL ratings (Atkinson, Zibin & Chuang, 1996; Moum, 1988). The overlap between affective state and certain QoL constructs poses some interesting issues since QoL measures may be confounded with the symptomatology of the mental illness itself.

Use of proxy measures may be one solution to problems associated with self-report methods. Proxy measurement, however, has its own set of difficulties. A common question is, how well do proxy reports actually reflect the experiences of persons for whom they are reporting? The correlation between mentally unimpaired subjects' scores and proxies' reports is moderate, generally around .6 to .7 for physical phenomena. This level of agreement, however, is observed primarily among proxies who know the patients well, interact with them on a regular basis and are asked to report on observable events. As proxies are asked to report on characteristics of an evaluative nature or about things which they have not observed, the convergence between patient and proxy reports declines significantly (Magaziner, Simonsick, Kashner & Hebel, 1988; Page, 1993; Rothman, Hendrick, Bulcroft, Hickman & Rubenstein, 1991). The same principle holds when using professional ratings. The use of professional proxies does not improve the precision of ratings if they do not observe the patient's daily experiences in the area on which they are being asked to report. As a final note, there is also a general tendency for proxies to under-rate measures of well-being and functioning in mentally healthy patients (Bassett, Magaziner & Hebel, 1990). It is unclear whether this trend holds for proxy reports of patients with a mental illness.

The choice to use proxies should be approached carefully and consideration should be given to the following: selecting proxy instruments which tap observable dimensions of patient experience and avoiding scales which rely on proxy evaluation; choosing proxies who are very familiar with the activities and experiences of the patient on whom they are being asked to report; providing clear instruction to proxies about what they are to include and to disregard during the rating process; and having a clear rationale for the use of proxy measurement.

If the purpose for using proxy reports is to gain a less biased rating for a group of patients, then the use of trained raters should be considered. Using several well trained raters is likely to provide more consistent and reliable ratings than gathering data with numerous untrained proxy informants. Again, however, the characteristics or events being rated must be observable and representative of daily life experiences, if observations are to be taken as indicators of experiences in daily life. When trained observers are used, the consistency among raters should be assessed using some trial cases, a technique which is more formally known as an assessment of "inter-rater reliability". Inter-rater reliability is typically reported either as the correlation between two or more raters' independent ratings or the percentage of categorical agreement between raters' rating(s) of the same event(s). Good inter-rater reliability should be above .8 or 80%, and assessed prior to the implementation of any data collection. Furthermore, if a study is expected to continue over a long period of time, or new raters are added, the consistency between raters should be re-assessed. It is also worth noting that similar principles apply to measurement tools which are administered by interview. Interviewers should be trained and variations in their approaches and methods should be minimized, if indeed it is desirable to reduce the effects of interviewer characteristics during the data collection process.

Next | Previous | Table of Contents