Using Screening Scales for Bipolar Disorder in Epidemiologic Studies: Lessons Not Yet Learned

Area of Science:

• Psychiatry
• Epidemiology
• Psychometrics

Background:

• A large epidemiological study in England estimated bipolar disorder prevalence using the Mood Disorders Questionnaire (MDQ).
• The study reported similar prevalence rates to other regions and noted low treatment rates for diagnosed individuals.
• Concerns were raised regarding the statistical validity of the conclusions drawn from the MDQ's performance.

Purpose of the Study:

• To review diagnostic screening principles and statistics.
• To differentiate screening from case-finding.
• To evaluate the MDQ's performance in the general population and calculate its positive predictive value in the cited study.

Main Methods:

• Review of screening principles and statistics.
• Analysis of the MDQ's performance using reported sensitivity and specificity.
• Calculation of the positive predictive value (PPV) of the MDQ based on general population data and the study's prevalence estimate.

Main Results:

• The MDQ developers reported 28.1% sensitivity and 97.2% specificity at a cutoff of 7.
• Humpston et al. reported a 1.7% prevalence of "probable" bipolar disorder using this cutoff.
• The calculated PPV of the MDQ in this study was only 14.8%.

Conclusions:

• The MDQ's performance, particularly its low positive predictive value, is insufficient for validly identifying individuals with bipolar disorder in low-prevalence populations.
• The study by Humpston et al. likely overestimated the prevalence of bipolar disorder due to the MDQ's limitations.
• The findings suggest that the epidemiological conclusions drawn by Humpston et al. are not valid indicators of bipolar disorder correlates in England's general population.