Table 2. Stepwise multivariate logistic regression analysis of the diagnosis of MDD and BD-I.
Variable | Stepwise analysis | ||||
---|---|---|---|---|---|
B | S.E. | p-value | OR | 95% CI | |
Cyclothymic temperament | 3.68 | 0.87 | 0.000 | 39.52 | 7.14–218.75 |
Anxious temperament | –2.44 | 0.88 | 0.005 | 0.09 | 0.02–0.49 |
YMRS score | 0.13 | 0.06 | 0.049 | 1.13 | 1.00–1.28 |
Constant | –2.79 | 0.99 | 0.005 | 0.06 |
Fit index of this model: χ2 = 28.11 (p-value < 0.05), Cox-Snell R2 = 0.11, Hosmer–Lemeshow test p = 0.275, sensitivity = 0.17, specificity = 0.96, positive predictive value = 0.56, negative predictive value = 0.78, AUC of ROC = 0.71
dependent variable: the diagnosis of MDD (1) and BD-I (2)
7 independent variables: scores of 5 subscales of the TEMPS-A and the severity of depressive and manic symptoms (PHQ-9 and YMRS scores, respectively).
AUC, area under the curve; BD- I, bipolar I disorder; CI, confidence interval; MDD, major depressive disorder; OR, odds ratio; PHQ-9, Patient Health Questionnaire-9; ROC, receiver-operating characteristic; B, partial regression coefficient; S.E., standard error; TEMPS-A, Temperament Evaluation of Memphis, Pisa, Paris and San Diego-autoquestionnaire; YMRS, Young Mania Rating Scale.