Purpose: Scale-recalibration can occur in patients who undergo a change in their lives, for instance an improvement in health. This can be illustrated by the fact that such patients retrospectively rate their health status worse than they actually did before the change. Until now scale-recalibration has only been demonstrated in patients who respond to treatment. We hypothesized that scale-recalibration is outcome dependent, meaning that non-responders retrospectively rate their health status the same or better than they do before intervention and responders rate their health status worse than they do before intervention.

Method: The baseline and three month follow-up data from a daily clinical practice cohort of patients with rheumatoid arthritis starting on TNFα blocking agents was used. Patients were classified as responders or non-responders according to an objective, disease-specific clinical outcome measure. A visual analogue scale (range 0-100 mm) for general health (GH) was used as a patient self-reporting outcome measure. At three months after starting on anti-TNFα, baseline GH was also retrospectively rated by means of a then-test. Differences were calculated between baseline and three months values and between then-test and baseline values. Significant differences were identified between responders and non-responders using a student's t-test.

Results: At three months 45.1% of the patients were classified as responder and 54.9% as non-responder. Mean baseline GH values were 62 for responders and 59 for non-responders, these values did not differ between the two groups (p=0.494). Responders reported a 30 (95% CI: 22; 39) VAS GH improvement compared to an improvement of 10 (95% CI: 3; 17) for non-responders. The mean difference between the then-test and baseline value was equal (p=0.561) for both responders (mean: 10, 95% CI: 3; 16) and non-responders (mean: 7, 95% CI: 1; 13), values being in favour of baseline. Therefore, the alternative hypothesis that outcome dependency would be present, in the sense that the differences between the then-test and the baseline value would be different for responders and non-responders, was rejected.

Conclusion: Both responders and non-responders retrospectively rated their general health worse than they did at baseline. Scale-recalibration was not dependent on the outcome of the intervention. This finding questions the validity of a then-test for assessing adaptation or correction for scale-recalibration.