Development and validation of a risk score to predict mortality during TB treatment in patients with TB-diabetes comorbidity

Table 3 Prognostic score performance in patients with complete data for multivariate model (N = 1113)

Risk group	n (%)	Score, mean (±SD)	Mortality % (n)	P Value*
Low-risk group (< 12 points)	776 (63.7%)	4.8 (±3.8)	3.1% (24)	< 0.001
Medium-risk group (12–21 points)	233 (19.2%)	17.3 (±2.3)	12.9% (30)
High-risk group (≥22 points)	208 (17.1%)	27.7 (±7.3)	27.9% (58)
All patients with complete data (N = 1217)	1217 (100%)	11.1 (±10.0)	9.2% (112)
Discrimination assessment
AUC (95% CI), final model in development	0.83 (0.79, 0.87)
AUC (95% CI), final model in bootstrap validation	0.82 (0.78, 0.87)
Calibration assessment
Hosmer-Lemeshow test	Chi-square = 4.54, p = 0.81
Brier score	0.07

Abbreviation: SD standard deviation, AUC Receiver Operating Characteristic (ROC) curve
Score range: [0–68]; Comparisons of mortality between risk groups were conducted using Chi-square test; *Overall p-value. A p < 0.01 was also found for all pairwise comparisons among groups (i.e. low-risk vs. medium-risk, low-risk vs. high-risk and medium-risk vs. high-risk groups)
Brier score: ranged from 0 to 1, the lower the Brier score is, the better the model is calibrated; Hosmer-Lemeshow test: non-significant p-value indicates the model is well calibrated

ISSN: 1471-2334