Minimal Clinically Important Difference in the Fibromyalgia Impact Questionnaire

DISCUSSION

We analyzed clinical data from over 2,000 patients with established, ACR-defined FM and clinically relevant pain as an inclusion criterion. The mean baseline total FIQ score of 62 was consistent with these patients’ moderately severe FM symptomatology. Patients were treated and followed for 3 months, enabling us to evaluate a clinically relevant improvement in the FIQ total score, and to obtain insight into severity categorization using the FIQ. The data were derived from 3 placebo-controlled clinical trials of pregabalin at fixed doses of 300, 450, and 600 mg/day. Analyses were undertaken to gauge the MCID of the FIQ total score and the FIQ stiffness item.

The MCID methodology is intended to quantify whether an individual patient within a particular treatment group has a clinically significant response and can be used to assess the mean percentage change within a particular treatment group. One advantage of using the MCID is that it accounts for the varying baseline levels of functional impairment. By calculating the percentage change in the FIQ total score from baseline for each patient and linking this to each patient’s PGIC category we were able to estimate the relationship between them. Using such an anchor-based interpretation of change in a functional score is a preferred method of evaluating clinical relevance, as it is a readily understood clinical phenomenon, and avoids elaborate statistical procedures¹⁹. To include a heterogeneous and diverse array of responses across the range of FIQ and PGIC scores to identify an accurate relationship, we combined responses across both treatment groups (pregabalin and placebo). The results from each of the 3 studies individually were generally similar and in close agreement. We found that a 14% change in the FIQ total score could be considered a MCID. Some researchers may prefer to use the absolute values instead of percentages, or in addition to percentages, to define a clinically meaningful difference. If we applied the same methodology for the absolute change as we did for the percentage change, then clinically meaningful differences of 8.1 (95% CI 7.6; 8.5) and 0.89 (95% CI 0.82; 0.96) would result for the FIQ total score and FIQ stiffness item, respectively.

We used an anchor-based approach — with PGIC as the anchor and FIQ as the targeted measure — to examine clinically important change. The anchor approach is consistent with that advocated by others for defining a clinically important change^13,14,23. The 2 essential characteristics of a good anchor are that it (1) be interpretable and (2) share an appreciable correlation with the target measure. Both of these characteristics were met in our investigation. Different ways of using the anchor approach have been proposed. For example, one way to define a clinically important improvement in pain has been to collapse categories on “much improved or very much improved” on PGIC²³. In the analyses presented in this article, we made each of the 7 categories on PGIC stand on its own merit and took the average difference between any pair of adjacent categories on PGIC as an estimate of the clinically important difference on the FIQ (which holds for clinical worsening as well as clinical improvement). In doing so, we avoided the arbitrariness of collapsing certain categories and capitalized on all available responses across the entire range of the PGIC.

One potential limitation is the assumption of a linear relationship between the FIQ and the 7 PGIC categories. In particular, FM does not appear to be a progressive disease and consequently the patients in these clinical trials are not likely to report deterioration. This may lead to a potential bias towards positive change on the investigated assessments, FIQ and PGIC. In the pooled analyses, 17% of patients reported any worsening and 17% reported no change on the PGIC. In a sensitivity analysis, in which the PGIC was used as a categorical rather than continuous variable, the relationship between the FIQ and PGIC and the estimated MCID was found to be similar to those reported with the PGIC as a continuous variable. Therefore the assumption of linearity was supported, and the estimate that a 14% change is clinically important is appropriate, and relevant in the direction of improvement or deterioration.

These findings are also plausible when taking into account the improvements observed in other clinical trials⁵ and experience in using the instrument in clinical practice. In this pooled analysis we found that the difference between the pregabalin 450 mg/day group and the placebo group in the percentages of patients who achieved an improvement of at least 14% was statistically significant, suggesting more patients receiving pregabalin treatment were likely to have a clinically important improvement in their FIQ total score. This finding is consistent with the observed efficacy of pregabalin in the treatment of FM^12,16,17.

We analyzed the FIQ stiffness item, in addition to the FIQ total score, as stiffness is a common and troublesome complaint among patients with FM^1,2,15. Morning stiffness is an early presenting symptom in rheumatoid arthritis and other inflammatory arthritides; its presence in FM, which is a noninflammatory pain disorder, complicates the differential diagnosis of both conditions. Thus it was of interest that a MCID of 13% for the FIQ stiffness item is similar to the MCID we observed in the FIQ total score. Given that the stiffness item contributes one-tenth to the total FIQ score⁴, its strong correlation with the overall score suggests that stiffness is a critical, yet unexplained, feature of FM.

As the FIQ includes items on pain, physical impairment, ability to work, restfulness, and mood, and may be influenced by pain levels, a significant correlation with the average daily pain score was expected. The overall correlation coefficient was 0.67 (range 0.5–0.7), a level appropriate for using pain severity cutpoints as an anchor to define FIQ severity bands^32,33. The analysis yielded categories in which a FIQ total score < 39 could be considered to represent a mild impairment, scores ≥ 39 to < 59 to represent a moderate effect, and a score ≥ 59 to represent a severe effect. This analysis is in quite good agreement with the approximations when the scale was originally developed, in which scores ≥ 70 were said to represent severe impairment. The severity bands can be useful in assessing treatment differences, as a criterion for study inclusion, and even to serve as an anchor to define change scores clinically important for other patient-reported outcomes^34,35. We found that most patients were rated as severe on the FIQ at baseline, which was not surprising given that the inclusion criteria required a score of 4 or greater (moderate to severe) on the pain VAS. At endpoint, there was a notable reduction in the percentage of patients rated as severe and an increase in the percentage of patients rated as mild.

The apparent gender bias in the FIQ, which results from it being developed in a predominantly female population and including items that are more likely to be performed by women⁵, does not affect the legitimacy of our findings as most of this sample also consisted of women. Our findings might not hold true in men with FM. As patients with severe depression or other psychiatric disorders were excluded from the studies, our findings may not extend to this subgroup. A further consideration is that these analyses were based on change over a 3-month period, but might not be applicable to longer periods of time; i.e., patients with a 14% improvement after 6 months or a year might not rate themselves as having improved.

These data were derived from FIQ changes in response to a medication; data from a nonpharmacological intervention (e.g., exercise) may yield a different MCID. Patients with severe depression or unstable psychiatric conditions were excluded from the studies; their inclusion may have yielded a different MCID. FM may occur in association with inflammatory disorders such as rheumatoid arthritis; changes in the FIQ have not been evaluated in combined disorders and it is likely that the MCID reported here would not be applicable in such situations. It should also be noted that a MCID may vary in other situations because of natural sampling variation, different study populations, type of anchor, time period of assessment, and other considerations³⁶; thus, our findings may not be generalizable. Moreover, the estimated MCID defined and derived here refers to a one-category difference on the PGIC, so values of MCID less than those estimated here would result for less than a one-category difference (e.g., a one-half category difference) on the PGIC.

The analyses reported here enhance the clinical utility of the FIQ in practice and assist in the interpretation of findings from clinical trials. In patients with moderate to severe impairment on the FIQ, a change of approximately 14% over 3 months is likely to be clinically important. A similar percentage change in stiffness is also likely to be clinically important.