Sample size calculations in scleroderma: a national approach to choosing outcome measurements in scleroderma trials

Janet E. Pope
Nicholas Bellamy

Department of Medicine, Division of Rheumatology, University of Western Ontario, London, Ontario

(Original manuscript submitted 7/2/94; received in revised form 2/8/94; accepted 8/8/94)

Abstract

Subjects with both diffuse and limited scleroderma were studied to calculate the baseline characteristics of several commonly used outcome measurements in order to provide parameters for sample size calculations for scleroderma clinical trials. From these estimates, outcome measurements were chosen as potentially responsive to change in clinical trials if their sample sizes were not prohibitively large. Forty-five patients with scleroderma were systematically assessed to determine the means and standard deviations whereby sample size calculations can be performed using this information. Examples of sample sizes were determined for the entire group, and for 2 subsets: those with diffuse scleroderma and those with diffuse disease of recent onset. Many baseline characteristics were significantly different in patients with diffuse compared to limited systemic sclerosis. The baseline values were different for the Health Assessment Questionnaire (HAQ) disability score, Functional Index, grip strength, oral aperture, finger-to-palm distance, skin score, and physician global assessment. Sample sizes can vary widely depending upon the outcome measurement chosen and the range of deltas used within the different scleroderma subsets. Sample size requirements for many outcome measures are extremely large due to marked variability in the baseline measures. Primary outcome measures in scleroderma trials should be chosen which have adequate power to detect a minimal clinically relevant change in the primary outcome measurements at the sample sizes employed. All other outcome measures should be ranked as secondary. Skin scores, global assessments, and grip strength measurements require smaller sample sizes than the other outcome measurements which were studied. Sample sizes in future trials will vary depending upon the proportion of patients with diffuse and limited scleroderma who are included.

Clin Invest Med 1995; 18 (1): 1-10

Table of contents: CIM vol. 18, no. 1