Measuring Health, Second Edition
A Guide to Rating Scales and Questionnaires

^{By Ian McDowell and Claire Newell
New York: Oxford University Press, 1996; 523 pp;
ISBN 0-19-510371-8; $85.95 (CAN)}

Despite the recent resurgence of infectious diseases, the modern history of health care is that its focus has been changing from saving lives to prolonging life. At least in the developed world, most medical conditions now are chronic ones, and the interventions are aimed at restoring the person's level of functioning, reducing disability and, at the end, offering palliation. As a consequence, it has become more difficult to evaluate the effectiveness of our procedures.

Simple counts of the number of survivors in one treatment condition or another are no longer sufficient, and both clinicians and researchers are turning to outcomes under such rubrics as "quality of life," "activities of daily living" or "quality adjusted life years." It has been only within the last decade, though, that people in the medical community have been able to measure these with any degree of precision.

Prior to this, those who were concerned with improving measurement techniques and clinician/researchers involved in health care delivery were living in two separate worlds that rarely intersected. Psychometricians were making advances in the areas of generalizability theory, item response theory, adaptive testing and confirmatory factor analysis, among others. These resulted in a new generation of tests to measure achievement, aptitudes, intelligence and other attributes of interest to psychologists: instruments that were much improved psychometrically, but that did not address the concerns of most physicians.

At the same time, those in the medical field were developing one scale after another to measure outcomes that were of interest to them, but, with rare exceptions, these were not based on sound principles of scale development. Much research, especially in the area of quality of life (QOL), was conducted using scales developed specifically for that study, and which had only rudimentary reliability and validity, at best, mainly because appropriate instruments did not exist. Compounding this bad situation, any good scales that did exist were most often difficult for the clinical researcher to find.

Fortunately, the picture has improved over the last decade or so. Clinical investigators are becoming aware that if their outcome measures are unreliable or have poor validity, their chances of finding significant differences between treatment conditions may be jeopardized. The flood of disease-specific QOL scales appears to have slowed to a trickle, as people are turning more and more to instruments with demonstrated psychometric properties, such as the Medical Outcome Study SF-36 and the Sickness Impact Profile. Also furthering this greater reliance on proven instruments is the availability of books that describe well-designed scales and that may include copies of the actual scales; this saves researchers from the tasks of tracking down the instrument, locating the supporting literature and then finding someone who can interpret what everything means.

When Measuring Health appeared in 1987, it was one of the first books outside psychology to which clinical researchers could turn to locate tests. The critiques of the measures were concise and balanced, based not only on the literature, but also on contacting the authors to clear up points of confusion. If the book had any shortcoming, it was the limited number of scales covered: only 50 were reviewed, grouped into six areas.

The second edition of this book significantly expands the scope of the instruments reviewed. There are now seven areas of tests: (1) physical disability and handicap, (2) social health, (3) psychological well-being, (4) depression (the area that is new to this edition), (5) mental status testing, (6) pain measurements, and (7) general health status and QOL. A few areas are specifically excluded, either because the focus of the tests is too narrow (e.g. disease-specific scales) or they have been reviewed in other books (child health, psychological health, unpublished tests). Within the areas covered, the authors do not attempt to be comprehensive, but to review only those tests that they consider to be the best.

A total of 88 different instruments are reviewed, divided roughly equally among the areas. Each review starts with a statement of the purpose of the test, including the intended population; an indication of the conceptual basis of the construct, when appropriate; a description of how the test is administered and scored; and a copy of the scale itself (when it is too long to reproduce in its entirety, a representative sample of items is given). This is followed by a review of the reliability and validity studies and the authors' comments about the strengths and weaknesses of the scale. Each of the seven sections has a table that summarizes and compares the scales in terms of their length, administration time, thoroughness of reliability and validity testing, and the results of these assessments.

Two questions immediately arise regarding the contents of the book: Are the tests chosen actually the best ones available, and Are the reviews balanced and fair? There is no objective way of answering these questions. But, before looking at the scales they selected and what they said about them, I made a list of the ones I considered best, and rank-ordered them. To the degree that the authors' judgements agreed with mine (by definition, the gold standard), they did an excellent job on both counts. They did not omit any scale that I would have included (and they added some I had not thought of), and their evaluations (or biases) were similar to mine.

While the scales make up the bulk of the book, there is also an introductory chapter, entitled "The Theoretical and Technical Foundations of Health Measurement" that is somewhat harder to review. Part of the difficulty is its scope. In just over 30 pages, the authors cover the history of health indicators, going back to the early studies of psychophysics, scaling methods and biases in subjective judgments, and ending with a discussion of validity and reliability assessment. There is an entire textbook of information contained in these pages, reflected by the fact that this chapter alone has 122 references, most of them key readings in psychometrics. My concern is that, while it may be too superficial for the psychometrician, it may be far too condensed for the neophyte, who may be overwhelmed by the sheer amount of content. However, anyone willing to read this chapter two or three times would be well rewarded for the effort.

If I have any quarrel with the content of the introductory chapter, it would be their reification of the "types" of tests (e.g. discriminative versus evaluative versus prognostic). Most psychometricians reject this, since instruments are often used in a variety of ways that may not have been foreseen by the developers. The key question is not What "type" of test is this, but the more general one, "Has this test been validated for the purpose for which I want to use it?" That would be carping, though. Overall, this is an excellent book, which should be chained to the bookshelf of everyone evaluating health care interventions-chained because it is one of the books that will constantly get "borrowed" by colleagues, never to reappear.

Overall rating:

Excellent

Strengths:

Comprehensive listing of the best tests. Fair and balanced reviews of the tests. Copies of the scales themselves

Weaknesses:

Very few. The chapter on psychometrics might be too overwhelming for the neophyte.

Audience:

Anyone evaluating the effects of health care, either at a program level or for the individual patient.