Interpretation and power of a pooled index.

OBJECTIVE: To describe, illustrate, provide statistical rationale and give clinical examples of the concept of a pooled index. METHODS: No explicit design. Randomized control trials of patients diagnosed with rheumatoid arthritis from Rheumatic disease clinics, tertiary care hospitals and communities are used as examples to illustrate the pooled index technique. Interventions in rheumatic disease unit inpatient and outpatient care, occupational therapy and physiotherapy provided by Arthritis Society Therapists in the community were compared. Variables included active joint count, morning stiffness in min, mean of right and left grip strength in mm Hg, erythrocyte sedimentation rate in mm/h, a functional change score, and a pooled index. RESULTS: The pooled index in each study detected a clinically important effect while individual measures were generally not able to declare the effect statistically significant. CONCLUSIONS: The pooled index is a more powerful clinical outcome than any individual measure provided it is computed from at least 2 relevant clinical outcome measures that have low correlation with each other.