Previous research on the depression scale of the Patient Health Questionnaire (PHQ-9) has found that different latent factor models have maximized empirical measures of goodness-of-fit. The clinical relevance of these differences is unclear. We aimed to investigate whether depression screening accuracy may be improved by employing latent factor model-based scoring rather than sum scores.

We used an individual participant data meta-analysis (IPDMA) database compiled to assess the screening accuracy of the PHQ-9. We included studies that used the Structured Clinical Interview for DSM (SCID) as a reference standard and split those into calibration and validation datasets. In the calibration dataset, we estimated unidimensional, two-dimensional (separating cognitive/affective and somatic symptoms of depression), and bi-factor models, and the respective cut-offs to maximize combined sensitivity and specificity. In the validation dataset, we assessed the differences in (combined) sensitivity and specificity between the latent variable approaches and the optimal sum score (⩾10), using bootstrapping to estimate 95% confidence intervals for the differences.

The calibration dataset included 24 studies (4378 participants, 652 major depression cases); the validation dataset 17 studies (4252 participants, 568 cases). In the validation dataset, optimal cut-offs of the unidimensional, two-dimensional, and bi-factor models had higher sensitivity (by 0.036, 0.050, 0.049 points, respectively) but lower specificity (0.017, 0.026, 0.019, respectively) compared to the sum score cut-off of ⩾10.

In a comprehensive dataset of diagnostic studies, scoring using complex latent variable models do not improve screening accuracy of the PHQ-9 meaningfully as compared to the simple sum score approach.

Although bipolar disorder (BD) is a fundamentally cyclical illness, a divided model of BD that emphasizes polarity over cyclicity has dominated modern psychiatric diagnostic systems since their advent in the 1980s. However, there has been a gradual return to conceptualizations of BD which focus on longitudinal course in the research community due to emerging supportive data. Advances in longitudinal statistical methods promise to further progress the field.

The current study employed hidden Markov modeling to uncover empirically derived manic and depressive states from longitudinal data [i.e. Young Mania Rating Scale and Montgomery–Asberg Depression Rating Scale responses across five occasions from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study], estimate participants’ probabilities of transitioning between these states over time (n = 3918), and evaluate whether clinical variables (e.g. rapid cycling and substance dependence) predict participants’ state transitions (n = 3229).

Analyses identified three empirically derived mood states (‘euthymic,’ ‘depressed,’ and ‘mixed’). Relative to the euthymic and depressed states, the mixed state was less commonly experienced, more temporally unstable, and uniquely associated with rapid cycling, substance use, and psychosis. Individuals assigned to the mixed state at baseline were relatively less likely to be diagnosed with BD-II (v. BD-I), more likely to present with a mixed or (hypo)manic episode, and reported experiencing irritable and elevated mood more frequently.

The results from the current study represent an important step in defining, and characterizing the longitudinal course of, empirically derived mood states that can be used to form the foundation of objective, empirical attempts to define meaningful subtypes of affective illness defined by clinical course.