Development and Validation of a Prediction Model for Co-Occurring Moderate-to-Severe Anxiety Symptoms in First-Episode and Drug Naïve Patients With Major Depressive Disorder

Abstract

Background: Moderate-to-severe anxiety symptoms are severe and common in patients with major depressive disorder (MDD) and have a significant impact on MDD patients and their families. The main objective of this study was to develop a risk prediction model for moderate-to-severe anxiety in MDD patients to make the detection more accurate and effective. Methods: We conducted a cross-sectional survey and tested biochemical indicators in 1718 first-episode and drug naïve (FEDN) patients with MDD. Using machine learning, we developed a risk prediction model for moderate-to-severe anxiety in these FEDN patients with MDD. Results: Four predictors were identified from a total of 21 variables studied by least absolute shrinkage and selection operator (LASSO) regression analysis, namely psychotic symptoms, suicide attempts, thyroid stimulating hormone (TSH), and Hamilton Depression Scale (HAMD) total score. The model built from the four predictors showed good predictive power, with an area under the receiver operating characteristic (ROC) curve of 0.903 for the training set and 0.896 for the validation set. The decision curve analysis (DCA) curve indicated that the nomogram could be applied to clinical practice if the risk thresholds were between 13% and 40%. In the external validation, the risk threshold was between 14% and 40%. Conclusion: The inclusion of psychotic symptoms, suicide attempts, TSH, and HAMD in the risk nomogram may improve its utility in identifying patients with MDD at risk of moderate-to-severe anxiety. It may be helpful in clinical decision-making or for conferring with patients, especially in risk-based interventions.

Figure 1

Flow diagram of study design. DCA, decision curve analysis; FEDN, first-episode and drug naïve; HAMA, Hamilton Anxiety Scale; HAMD, Hamilton Depression Scale; HL, Hosmer–Lemeshow; LASSO, least absolute shrinkage and selection operator; MDD, major depressive disorder; PANSS, Positive and Negative Syndrome Scale; ROC, receiver operating characteristic.

Figure 2

Variable selection by the LASSO binary logistic regression model. A coefficient profile plot was constructed against the log (λ) sequence. (a) Five variables with nonzero coefficients were selected by deriving the optimal λ. (b) Following verification of the optimal parameter (λ) in the LASSO model (dotted line on the left side of (b)), the partial likelihood deviance (binomial deviance) curve versus log (λ) is plotted and drawn based on 1 standard error criteria (dotted line on the right side of (b)). LASSO, least absolute shrinkage and selection operator.

Figure 3

A risk factors of suicide attempts, psychotic symptoms, HAMD, and TSH for the nomogram prediction model. HAMD, Hamilton Depression Scale; TSH, thyroid stimulating hormone.

Figure 4

ROC validation of the moderate-to-severe anxiety nomogram prediction. The y-axis indicates the true positive rate of risk prediction and the x-axis indicates the false positive rate of risk prediction. The thick blue line indicates the performance of the nomogram in the training set and validation set. ROC, receiver operating characteristic.

Figure 5

Calibration curves of the predictive moderate-to-severe anxiety risk nomogram. The y-axis represents actual diagnosed cases of moderate-to-severe anxiety, the x-axis represents the predicted risk of moderate-to-severe anxiety. The diagonal gray line represents a perfect prediction by an ideal model, and the dotted line represents the performance of the training set and validation set, with the results showing that a closer fit to the ideal line represents a better prediction.

Figure 6

Decision curve analysis for the moderate-to-severe anxiety risk nomogram. The y-axis measures the net benefit. The gray line represents the assumption that all patients have moderate-to-severe anxiety, the black line represents the assumption that all patients have no moderate-to-severe anxiety, and the red line represents the risk nomogram.