The in-hospital PHQ-2 depression scale showed poor to moderate predictive ability for post-discharge depression symptoms among critical illness survivors (AUC 0.74 for continuous vs 0.61 for binary).
Observational (n=164)
Yes
Does in-hospital PHQ-2 screening accurately predict post-discharge depression symptoms in survivors of critical illness?
The in-hospital PHQ-2 screening tool has poor predictive utility for identifying post-discharge depression symptoms among critical illness survivors.
Effect estimate: AUC 0.74 vs 0.61
Abstract Rationale Although post-discharge symptoms of depression are highly prevalent after critical illness, the best timing of and strategies for screening are unclear. The objective of this study was to determine the utility of using the Patient Health Questionnaire 2-item (PHQ-2) depression scale to detect elevated depression symptoms after hospital discharge, based on Patient Health Questionnaire 9-item (PHQ-9). Methods This is a secondary analysis of data from a factorial experimental trial designed to optimize a post-discharge distress intervention with 3-month follow among survivors of cardiorespiratory failure managed in intensive care units (ICUs) in North Carolina, Oregon, and Colorado. Depression symptoms were measured using continuous and binary PHQ-9 scores, with ≥5) indicating clinically important depressive symptoms. The primary predictor was the in-hospital PHQ-2, also assessed as a continuous and binary variable at different outpoint scoring thresholds. Predictive performance was assessed for each model using in-sample and cross-validated diagnostics such as Brier score and AUC for binary PHQ-9 and Root Mean Square Error, Mean Squared Error, and R2 for continuous PHQ-9. Additional sociodemographic and clinical variables were included in multivariable models using lasso penalized regression, and predictive performance was assessed accordingly. Results Among 164 patients included, 118 (72%) had a post-discharge PHQ-9 ≥5. Compared to the poor to moderate predictive ability of binary PHQ-2 scores alone, continuous PHQ-2 scores provided better discrimination (AUC 0.74 vs. 0.61) but similar prediction accuracy (Brier score 0.17 vs. 0.19). However, the R2values across all models (range 0.23-0.36) suggested weak explanatory power of the PHQ-2 score. Although the predictive performance of the PHQ-2 improved slightly with the addition of sociodemographic and clinical variables, it remained fair to moderate across all models (AUC range 0.77-0.83, Brier score range 0.15-0.17). The most powerful predictors in these models were financial stress, PHQ-2 score, and a surgical vs. medical diagnosis for ICU admission. While a PHQ-2 threshold ≥2 appeared to provide the best balance between sensitivity (57%) and specificity (80%), 42% with clinically important depression symptoms would be missed by this score cut-point and the overall accuracy was just 64% (Table). Though the specificity of the standard ≥3 cutoff score for the PHQ-2 was 91%, its sensitivity was only 31%. Conclusions The performance of the in-hospital PHQ-2 as a screening tool for post-discharge depression symptoms of clinical importance among critical illness survivors does not support its use for this purpose. This performance was not changed substantially by the addition of clinical characteristics. This abstract is funded by: NIH
Cox et al. (Fri,) conducted a observational in Critical illness (cardiorespiratory failure) (n=164). In-hospital Patient Health Questionnaire 2-item (PHQ-2) was evaluated on Elevated depression symptoms after hospital discharge (PHQ-9 ≥5) (AUC 0.74 vs 0.61). The in-hospital PHQ-2 depression scale showed poor to moderate predictive ability for post-discharge depression symptoms among critical illness survivors (AUC 0.74 for continuous vs 0.61 for binary).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: