Depression is a highly heterogenous disorder, both with respect to risk factors and variability in symptom profiles (Fried, 2017; Fried et al., 2014). This fact has inspired researchers to characterize which specific symptoms in the depression syndrome are associated with specific risk factors (i.e., a phenotype) (Hilland et al., 2019; Moriarity et al., 2021; van Loo et al., 2018). Clarifying specific risk factor—psychopathology associations is, in turn, critical to advancing precision psychiatry insofar as it can enable the selection of treatments based on a person's symptom profile. For example, knowing that a certain group is more likely than other groups to have a symptom profile characterized by concentration difficulties can lead a mental health care provider to prioritize concentration-oriented therapies to maximize effectiveness. This work is also important to ensure the rigor of future research because characteristics that are differentially associated with symptoms within a syndrome cause an issue called measurement noninvariance when testing said characteristic as a predictor of symptom scores (e.g., sum score, latent variable) (Putnick and Bornstein, 2016; Moriarity et al., 2022). Briefly, measurement noninvariance is when a measure quantifies a construct differently across different groups. In this situation, standard analytic options (i.e., mean comparison) will not suffice because standard scoring procedures will not be equally valid for all groups. Therefore, phenotyping research is necessary both to maximize the translatability psychiatric research and enhance the measurement validity of assessments in diverse populations (Bauer, 2023).

Research examining individuals at high risk of developing depression has found that sexually diverse adults and youth experience depression at much higher rates than their heterosexual counterparts (Argyriou et al., 2021; Bostwick et al., 2010; Chakraborty et al., 2011; King et al., 2008; Pakula and Shoveller, 2013; Ploderl and Tremblay, 2015; Wittgens et al., 2022). However, across studies, the strength of this association is quite variable. For example, Marshal et al. (2011) conducted a meta-analysis of studies examining depression and suicidality in sexually diverse youth, and found that although sexually diverse youth consistently have higher rates of depression and suicidality, the strength of these associations varied across depression assessment methods (e.g., using a one item scale vs. a widely validated multi-item measure, such as the Center for Epidemiologic Studies–Depression Scale (CES-D) (Radloff, 1977). Although variability in the strength of findings is clearly impacted by the scale used to measure depression, we can also see variability due to the ways in which scholars use specific scales. For example, studies can test differences in depression symptoms as a singular diagnostic construct (e.g., total or latent symptom scores) or can test differences in individual symptoms to characterize symptom-specific phenotypes. These two approaches can yield very different conclusions. Careful consideration of how measures are used is particularly important for widely used scales like the CES-D, including how these scales might function differently across populations.

Research that takes a symptom-level approach, and that examines how individual constructs and items differ across groups, can provide far more specificity in our understanding of depression in specific populations as compared to simply examining latent scores. For example, Radusky et al. (2021) used the 20-item CES-D scale to investigate differential item functioning in a sample of diverse men with and without HIV. They found that regardless of HIV status, sexually diverse men were less likely to endorse factors “people were unfriendly” and “I felt like people disliked me,” and more likely to endorse the factor “I did not feel like eating,” than their heterosexual counterparts. This finding contradicts the theoretical prediction that those who face more stigma should have more negative interpersonal experiences (Meyer, 2003), and it highlights that sexually diverse men might be more prone to appetite suppression and malnourishment than non-sexually diverse peers. To date, however, Radusky et al. (2021) is one of—if not the only—study to have examined the differential functioning of individual items of the CES-D in sexually diverse populations.

Critically, the findings articulated above may not generalize to the larger sexually diverse population given that their sample was only men and was focused on assessing the impact of HIV status as well as sexual orientation. Therefore, research using more representative samples is needed. Additionally, the statistical techniques chosen were exclusively symptom-specific, failing to consider that sexual diversity might be simultaneously and uniquely associated with both depression generally and individual symptoms in a “hierarchical phenotype” (Moriarity et al., 2023). Without modeling both levels of depression measurement, it is impossible to falsify theories about whether sexual diversity is associated with depression and/or individual symptoms. To address these concerns, we leverage moderated nonlinear factor analysis (MNLFA) to explore the possibility of sexual diversity-centered hierarchical phenotypes of depression symptoms (i.e., that sexual diversity might be simultaneously and uniquely associated with depression generally as well as individual symptoms) in a diverse sample of 5065 adults.