TO THE EDITOR:
"A Self-Administered Questionnaire for Assessment of Symptoms and Function of the Shoulder" (79-A: 738—748, May 1997), by L'Insalata et al., presented an excellent questionnaire for patients who have problems related to the shoulder. Nevertheless, we have a number of concerns.
First, the authors were apparently unaware of other published work on shoulder questionnaires that were also designed for self-completion by patients and that have been subjected to various degrees of statistical evaluation2,3,7.
Second, the patients in each of these previous articles2,3,7 were, on the average, at least fifteen years older than the patients in the study by L'Insalata et al. (average age, forty years). We think that this is important since disorders of the rotator cuff, which are the shoulder conditions of greatest epidemiological importance, tend to affect older people and it is not at all clear that the newer questionnaire is either appropriate or acceptable for older patients (not to mention the difficulty that such patients may have with the use of touch-screen computers).
A third and related concern stems from the diagnostic heterogeneity of the sample described by L'Insalata et al. In particular, approximately one-third of the patients had shoulder instability.
During the in-depth-interview stage of the development of our own questionnaire3, it became apparent that people who have shoulder instability have distinct concerns that have as much to do with the anticipation of problems with specific activities as with any regular occurrence of pain or with actual disability. As a result of this observation, we excluded such patients from our own sample. The same observation has been made regarding the Constant (clinical) shoulder score5.
We believe that the inclusion of patients who have dislocation of the shoulder with patients who have a degenerative shoulder condition could threaten the overall reliability and validity of this new questionnaire. The scoring system would obscure this possibility by allotting average scores within domains where respondents deemed items not applicable or simply did not answer them. This has never been recommended practice on, for instance, the SF-36, as a missing response is often missing for a specific reason and it can never be assumed that the intended response would have been the average value for that domain. L'Insalata et al. also said that, at the developmental stage, they were obliged to omit a number of questions that were particularly pertinent to people who had instability. This seems an unnecessary compromise when a separate and more appropriate questionnaire designed specifically for this group of patients makes more sense.
Criticisms aside, we compliment L'Insalata et al. on the rigorous methodology employed in the design and testing of their questionnaire. As they stated, clinical measures rarely stand up to this level of scrutiny.
Jill Dawson, D.Phil.; Ray Fitzpatrick, Ph.D.: Department of Public Health, Division of Public Health and Primary Health Care, University of Oxford, Gibson Building, Radcliffe Infirmary, Oxford OX2 6HE, England
Andrew Carr, F.R.C.S.: Upper Limb Unit, Nuffield Orthopaedic Centre, Windmill Road, Headington, Oxford OX3 7LD, England
Dr. L'Insalata, Dr. Warren, Mr. Cohen, Dr. Altchek, and Dr. Peterson reply:
We appreciate the interest and comments of Dr. Dawson et al. In our paper, we wrote briefly about the existence of multiple scoring systems for the assessment of shoulder function. However, we did not attempt to provide a comprehensive review of the more than fifteen scales that have been used to assess such function1,4,6, and we limited our references to twenty-five at the request of the editors. We did read the recent article by Dawson et al.3 with much interest. However, we submitted our manuscript in 1995 and revised it before we were aware of their questionnaire.
Regarding the issue of the lower average age and the diagnostic heterogeneity of our sample, we did not preselect the patients for the study. Instead, all patients who were seen by the two senior ones of us (R. F. W. and D. W. A.) for evaluation of a disorder related to the shoulder were enrolled. We had hoped to have a heterogeneous cohort that would reflect the typical patients seen in an orthopaedic practice for problems related to the shoulder, and we view this heterogeneity as a strength of our study. As Dawson et al. note, the inclusion of patients who have multiple diagnoses potentially could have led to a scenario in which the questionnaire was found to be reliable, valid, and responsive for the overall group but failed to meet each of these criteria for one of the diagnostic groups or more. We anticipated this potential problem and thus analyzed the reliability, validity, and responsiveness of the questionnaire separately for each of four diagnostic groups: instability, impingement, complete tear of the rotator cuff, and osteoarthrosis and adhesive capsulitis. The questionnaire was found to be reliable, valid, and reproducible for each. The average age was twenty-eight years for the patients who had instability, forty-three years for those who had impingement, fifty years for those who had a tear of the rotator cuff, and fifty-one years for those who had osteoarthrosis or capsulitis. The average age of the patients in the latter two groups is comparable with that reported by others who restricted the diagnosis to impingement, tear of the rotator cuff, or osteoarthrosis of the shoulder2,3,7.
We disagree that a separate questionnaire should be designed for patients who have shoulder instability. There is already an abundance of shoulder-rating scales, and we believe that a single questionnaire that is valid, reliable, and responsive for multiple diagnoses is more practical than several diagnosis-specific questionnaires. Also, it may not always be apparent at the initial evaluation that the diagnosis is instability rather than impingement or some other diagnosis. A single questionnaire that is applicable to any diagnosis eliminates the potential of using the wrong questionnaire.
Like Dawson et al., we have found in our practice, as well as during interviews with patients for the creation of our questionnaire, that frequently the anticipation of an event that causes instability and not just the actual event concerns patients. As we stated, we had initially included two questions that specifically evaluated shoulder instability. These were eliminated from the final questionnaire mainly because, after further consideration and discussions with patients, we came to view instability more as a causal factor of dysfunction than as an actual measure of dysfunction. When they are first seen, most patients do not state that they are having instability. Rather, they say that they are having pain with particular activities or must restrict their activity because of pain or apprehension of additional pain or dislocations. Furthermore, we found that the questions regarding instability of the shoulder were difficult to validate clearly.
In contrast to the observations of Dawson et al., we have found that patients who have instability of the shoulder do have a measurable disability. When the initial 100 patients were grouped according to a diagnosis of instability or non-instability (all other diagnoses), we found similar levels of disability (defined as a low score on the questionnaire) in each domain. Moreover, the responsiveness of the questionnaire was very good, with an average standardized response of 1.4, for patients who had had a stabilization procedure on the shoulder.
John C. L'Insalata, M.D.: Orthopaedic Surgical Consultant, P.C., 9921 4th Avenue, Brooklyn, New York 11209
Russell F. Warren, M.D.; Steven B. Cohen, B.A.; David W. Altchek, M.D.; Margaret G. E. Peterson, Ph.D.: The Hospital for Special Surgery, 535 East 70th Street, New York, N.Y. 10021