TABLE 1.
Characteristics of the 26 studies that developed models with or without validation.
| Inpatient setting (n = 16) | Emergency department (n = 6) | Intensive care unit (n = 3) | Primary care (n = 1) | Overall (n = 26) | |
|---|---|---|---|---|---|
| Internal validation | |||||
| Random splitting | 3 | 3 | 1 | 0 | 7 |
| Bootstrapping | 6 | 1 | 1 | 1 | 9 |
| Cross‐validation | 0 | 0 | 0 | 0 | 0 |
| Combination of methods | 1 | 1 | 0 | 0 | 2 |
| None | 6 | 1 | 1 | 0 | 8 |
| Modelling method | |||||
| Logistic regression | 11 | 5 | 3 | 1 | 20 |
| Generalized linear model | 1 | 0 | 0 | 0 | 1 |
| Classification and regression tree | 2 | 1 | 0 | 0 | 3 |
| Machine learning | 1 | 0 | 0 | 0 | 1 |
| More than one method | 1 | 0 | 0 | 0 | 1 |
| Not reported | 0 | 0 | 0 | 0 | 0 |
| Handling of missing data | |||||
| Imputation | 4 | 1 | 0 | 1 | 6 |
| No missing values | 0 | 0 | 0 | 0 | 0 |
| Inappropriate handling | 2 | 0 | 0 | 0 | 2 |
| Not reported | 10 | 5 | 3 | 0 | 18 |
| Model discrimination | |||||
| C‐statistic | 16 | 6 | 3 | 1 | 26 |
| None | 0 | 0 | 0 | 0 | 0 |
| Model calibration | |||||
| Hosmer–Lemeshow test | 3 | 1 | 1 | 0 | 5 |
| Calibration plot | 1 | 0 | 0 | 1 | 2 |
| Hosmer–Lemeshow test and Calibration plot | 4 | 0 | 0 | 0 | 4 |
| Other | 4 | 0 | 0 | 0 | 4 |
| None | 4 | 5 | 2 | 0 | 11 |
| Model presentation | |||||
| Sum score | 9 | 4 | 1 | 0 | 14 |
| Decision tree | 1 | 1 | 0 | 0 | 2 |
| Nomogram | 1 | 1 | 0 | 0 | 2 |
| Equation | 2 | 0 | 1 | 1 | 4 |
| More than one method | 1 | 0 | 0 | 0 | 1 |
| None | 2 | 0 | 1 | 0 | 3 |