(PS3-D84) Comparison of Hypomania Checklist Psychometrics in Pre-covid Pandemic and Pandemic-era Samples

The COVID-19 pandemic accelerated changes in access to psychological assessment. Remote-access screening helped to address this need, but little is known about the effect of pandemic conditions on the reliability and validity of these screening assessments. The Hypomania Checklist (HCL-32-R1) can be administered remotely to screen for bipolar disorders among individuals displaying symptoms of major depressive disorder, but to our knowledge, no study has examined its validity during pandemic conditions. Pandemic conditions may have influenced item relationships with the latent factor structure of the assessment and discrimination overall, reducing the usefulness of traditional indicators in hypomania screening. We hypothesized differences in factor loadings, item discrimination, and test sensitivity would emerge in items reflecting lifestyle habits due to pandemic restrictions, such as items relating to driving, physical activity, and clothing choice, or items reflecting risk-taking behaviors exacerbated during isolation, such as drinking or drug use. 

Participants (N=10,800) between 18-88 years of age accessed the HCL-32-R1 remotely between 2017-2022 through the Depression and Bipolar Support Alliance. Data were split into pre-pandemic and pandemic groups using the WHO declaration of pandemic on March 11, 2020 as the cutoff date. We compared maximum likelihood exploratory factor analysis (ML EFA) models to examine changes in factor structure and item relationships between these groups. Item response theory (IRT) modeling of “Sunny'' and “Dark” subscales visualized item responses, assessed sensitivity, and determined subscale reliability. 

Using a minimum loading of 0.4, ML EFA favored a 2-factor model corresponding to Sunny and Dark subscales in both conditions. Despite some item loadings inconsistent with the assignments suggested for clinical use, most items loaded onto the expected latent factors. Items’ factor loadings between groups were similar, with most loading differences between 0.00-0.05.  Item 8 (“I spend too much money”) did not load onto Dark in the pandemic group, whereas “I smoke more cigarettes” did. This may reflect widespread lifestyle changes due to lockdowns, though it is unclear why other items (e.g. relating to driving, physical activity, clothing, drinking) were largely unaffected. The similarity of factor structure and loadings between groups did not suggest a need for confirmatory factor analysis. 

IRT Information Criterion Curves also demonstrated comparable discrimination and sensitivity on Sunny and Dark subscale items for pre-pandemic and pandemic groups. IRT reliability curves showed similar reliability above threshold (0.8) for both groups in the Sunny subscale (between 𝜃 ~ 0 and 2) and the Dark subscale (between 𝜃 ~ 0.5 and 1). Our findings suggest the HCL-32-R1 remained a reliable screening tool during pandemic conditions. Further, the HCL-32-R1 can be accessed for free and remotely. Its reliability and accessibility make it a clinically useful tool to respond to mental health needs in emergency situations. Limitations and future directions will be discussed.