Measurement of Vital Signs for Skin Diseases

Abstract

Well aware of Oscar Wilde’s admonition that ‘‘Consistency is the last refuge of the unimaginative’’ I realize that calls for improved measurement in dermatologic clinical research have been consistent, persistent, and long-standing (Marks et al, 1989; Chren, 1997, 2000; Charman et al, 2003). What is the basis for this emphasis? Have things improved? What are the current challenges and next steps for this field? Accurate measurement is particularly important for dermatologic clinical research. As in other fields, to document the effects of our care and to compare care with different therapies, we need accurate measurement tools for the ‘‘state’’ of a given skin condition. Measuring the state of the skin is difficult, however. In general, our patients are ambulatory, living with skin diseases rather than dying from them; moreover, these diseases do not often affect conventional laboratory values. Most importantly, we know instinctively that the severity of a skin disease is often not captured simply in its physical characteristics—its scale, induration, erythema—but also includes aspects that can be determined only by patients’ reports—pruritus, pain, effects on quality of life. Thus, our challenges are substantial: we must measure not only what we see clinically (using clinimetrics; Feinstein, 1987), but also patients’ reports of their experiences (using psychometrics). Tremendous progress has been made in clinical measurement in dermatology, and two papers in this issue of the Journal illustrate incremental advances in the field. Nijsten et al (2005) studied the responses of a large US sample with psoriasis to a common measure of disability from the disease, the psoriasis disability index (PDI). Previous work had documented that the instrument has evidence of validity. These investigators performed a variety of other psychometric analyses using classic methods as well as less common evaluations such as Rasch analyses, which are based on item response theory, a different approach to modeling, scale construction, and scoring (Ware, 2003). They conclude that the PDI may be suboptimal for widespread use in patients with mild-to-moderate psoriasis, in part because of its relative insensitivity to lower levels of disability, and the lack of evidence to support scoring the instrument with a single global score. In another paper, Hongbo et al (2005) provide evidence to aid the interpretation of quality-of-life scores. This work correlated patients’ scores on the Dermatology Life Quality Index (DLQI) to their responses to a global question about the effects of their skin disease on their lives. The analysis permits a valid qualitative interpretation of DLQI scores as indicating no, small, moderate, very large, or extremely large effects on patients’ lives. Attaching clinical meaning to psychometric scores is an important first step in making these tools useful in clinical encounters. These papers are important for several reasons. First, both studies examine instruments that were developed and refined by Andrew Finlay and his colleagues in Cardiff, highlighting the seminal and sustained effects of their work on improving the measurement of complex constructs in dermatology. Second, the psychometric scrutiny to which the instruments are subjected in these papers is emblematic of the overall increase in the scientific rigor being applied to dermatologic measurement in general. Finally, the papers provide an opportunity to think about the ‘‘next steps’’ in this field. Psychometric scrutiny may subject assays for clinical measurement to more compulsive standards than we use for other, more conventional clinical tests such as radiologic or histologic examinations (Sackett et al, 1991). On the other hand, these kinds of investigations raise legitimate questions and skepticism about some features of our clinical measurement instruments. For example, among the conclusions of the Nijsten work was that disabling effects of psoriasis are not unidimensional, questioning the validity and usefulness of a single score as a measure of these complex effects. This conclusion was based on data from their psychometric analyses, but it has substantial face validity. Single scores, especially those that are merely calculated as sums or products, are attractive because they are straightforward to compute, report, and compare among different diseases and patient groups (Chren and Weinstock, 2004). But they may be suboptimal for at least two reasons. First, they may be invalid for certain purposes, as suggested by the Nijsten work. As these investigators state, further validity testing is indicated to determine the dimensions and complexity of what the PDI measures. Second, single summative scores may also mask our understanding of quality-of-life effects of diseases. This point is illustrated by some of the data reported in the Hongbo paper. Because the various ‘‘domains’’ of qualityof-life effects are averaged for the DLQI score, a patient who responded that her skin was itchy ‘‘very much’’ but was Abbreviation: DLQI, dermatology life quality index