Estimating posttraumatic stress disorder severity in the presence of differential item functioning across populations, comorbidities, and interview measures: Introduction to Project Harmony

Abstract

Multiple factor analytic and item response theory studies have shown that items/symptoms vary in their relative clinical weights in structured interview measures for posttraumatic stress disorder (PTSD). Despite these findings, the use of total scores, which treat symptoms as though they are equally weighted, predominates in practice, with the consequence of undermining the precision of clinical decision‐making. We conducted an integrative data analysis (IDA) study to harmonize PTSD structured interview data (i.e., recoding of items to a common symptom metric) from 25 studies (total N = 2,568). We aimed to identify (a) measurement noninvariance/differential item functioning (MNI/DIF) across multiple populations, psychiatric comorbidities, and interview measures simultaneously and (b) differences in inferences regarding underlying PTSD severity between scale scores estimated using moderated nonlinear factor analysis (MNLFA) and a total score analog model (TSA). Several predictors of MNI/DIF impacted effect size differences in underlying severity across scale scoring methods. Notably, we observed MNI/DIF substantial enough to bias inferences on underlying PTSD severity for two groups: African Americans and incarcerated women. The findings highlight two issues raised elsewhere in the PTSD psychometrics literature: (a) bias in characterizing underlying PTSD severity and individual‐level treatment outcomes when the psychometric model underlying total scores fails to fit the data and (b) higher latent severity scores, on average, when using DSM‐5 (net of MNI/DIF) criteria, by which multiple factors (e.g., Criterion A discordance across DSM editions, changes to the number/type of symptom clusters, changes to the symptoms themselves) may have impacted severity scoring for some patients.

Cumulative exposure to traumatic events and craving among women in residential treatment for substance use disorder: The role of emotion dysregulation and mindfulness disposition

“…Lastly, our findings identifying evident differences by raceethnicity across key study constructs (i.e., mindfulness, emotion dysregulation, and craving) may also be a function of the psychometric properties of the utilized scales. Recent studies have revealed significant differences in item functioning across scales assessing affect and trauma among underserved groups such as racial and ethnic minoritized populations (i.e., the Clinician-Administered PTSD Scale for DSM-IV (CAPS-IV; Ruglass et al, 2020;Morgan-López et al, 2022) Future studies are needed to examine if these similar psychometric issues are present in other widely used scales such as those utilized in the present study. Indeed, these differences may likely have implications for treatment interventions that target racial-ethnic minoritized groups with substance use dependence.…”

Differential symptom weighting in estimating empirical thresholds for underlying PTSD severity: Toward a “platinum” standard for diagnosis?

“…Notably, there is no particular prioritization of any one symptom over any other symptom within each PTSD criterion, as each symptom is treated as though it has equal priority in reflecting underlying PTSD severity and diagnosis. However, it has been noted that symptom counts can give a distorted picture of underlying disorder severity, because the relative weight of each symptom in relation to the disorder is not taken into account (He et al., 2014; Morgan‐López, Hien et al., 2022; Silverstein et al., 2020).…”

“…However, it has been long understood that total scores can be problematic from a psychometric perspective (McNeish & Wolf, 2020) as well as a clinical perspective (Saavedra et al., 2022; Saavedra, Morgan‐López, Hien, Back et al., 2021). Total scoring is primarily thought of as a “data management problem”, often with little‐to‐no consideration for what is implied psychometrically by its use (Morgan‐López, Hien et al., 2022; Morgan‐López, Saavedra et al., 2022). For any latent variable model that is structured under the “equal weights” model (e.g., 1‐parameter logistic item response theory (IRT), nonlinear factor analysis [NLFA] with equal factor loadings), the total score is a sufficient statistic for maximum likelihood estimation of the latent variable (Andrich, 1978, 2018; McNeish & Wolf, 2020).…”

“…In practice however, it has been consistently demonstrated that PTSD symptoms fit to IRT or NLFA models require discrimination parameters/factor loadings that vary across symptoms (Contractor et al., 2018; Morgan‐López, Saavedra et al., 2020; Ruglass et al., 2020; Saavedra, Morgan‐López, Hien, Back, et al., 2021, 2022; Silverstein et al., 2020). The underlying clinical conclusion from this collection of advanced psychometric work is that some symptoms “matter” more than others (Bourne et al., 2013; Morgan‐López, Hien et al., 2022).…”

“…The underlying clinical conclusion from this collection of advanced psychometric work is that some symptoms "matter" more than others (Bourne et al, 2013;Morgan-López, Hien et al, 2022).…”