Objectives: The Nepean Dyspepsia Index (NDI) was specifically developed for trials in functional dyspepsia, but the smallest change on the total or subscale scores, which corresponds to a clinically meaningful change (minimal clinically important difference, MCID), has not been established. An MCID has been established for the SF-36 (> or =5 points on physical or mental health subscales); such information is critical for understanding clinical trial data. We aimed at calculating an MCID for the NDI to help guide the interpretation of future clinical trials.
Methods: Comprehensive clinical data were collected in outpatients (n=101) and community subjects (n=460), and the MCID for the NDI total score was examined in three ways. The first estimated the average change (over a 2-week period) in the NDI corresponding to a five-point change in the SF-36 mental and physical subscales, and the second repeated this using a 1-s.d. change in symptom level, whereas the third calculated Cohen's d effect size among individuals changing by at least five points on the SF-36 subscales. A separate cross-sectional study was used to obtain the receiver-operator characteristic curve for discriminating between dyspepsia and non-dyspepsia subjects.
Results: Among individuals improving by at least 1 s.d. on the patient-reported symptom score, the corresponding improvement in NDI quality of life (QoL) was an average of 18 points (s.d.=12) compared with only 7 points (s.d.=15) in those with no/minimal change in symptoms, yielding a Cohen d of 1.0 and a proposed MCID of 10 points. Although the same pattern was found using the SF-36 physical scale, the effect size was smaller (Cohen's d=0.25). Smaller effect sizes were also obtained using the SF-36 mental subscale (Cohen's d=0.1) and the physician global assessment (Cohen's d=0.33). In a separate cross-sectional community study, the NDI-QoL score was shown to provide good discrimination between individuals meeting and not meeting the Rome criteria for functional dyspepsia, with an area under the receiver-operator characteristic curve of 0.80 (95 % confidence interval: 0.75, 0.85).
Conclusions: A change of at least 10 points on the NDI total scale corresponds to a clinically meaningful change in patient status.