Risk of bias summary based on Prediction study Risk Of Bias Assessment Tool (PROBAST) (permission from Wolff R, personal communication)
| Population | General surgery | T&O | General (medical and surgical) | Heart failure | |||||||
| Model, year | Kheterpal et al, 200733 | Kheterpal et al, 200934 | Bell et al, 201535 | Drawz et al, 200838 | Matheny et al, 201037 | Koyner et al, 201640 | Bedford et al, 201639 | Forni et al, 201336 | Forman et al, 200443 | Breidthardt et al, 201141 | Wang et al, 201342 |
| Study participants | ? | ? | + | + | + | ? | + | + | ? | ? | ? |
| Predictors | ? | ? | ? | ? | – | ? | – | + | – | – | ? |
| Outcome | – | – | + | – | – | – | + | + | – | – | – |
| Sample size and missing data | – | + | + | – | – | ? | ? | ? | – | – | ? |
| Statistical analysis | – | – | + | – | + | – | – | ? | – | – | – |
| Overall judgement of bias | – | – | + | – | – | – | – | + | – | – | – |
| Overall judgement of applicability | – | – | ? | – | – | – | + | + | – | – | – |
| Usability of the model | + | + | + | + | + | + | + | + | + | + | + |
Study participants domain: design of the included study, and inclusion and exclusion of its participants; predictors domain: definition, timing and measurement of predictors (also assesses whether predictors have not been measured and were therefore omitted from the model); outcome domain: definition, timing and measurement of predicted outcomes; sample size and missing data domain: number of participants in the study and exclusions owing to missing data; statistical analysis domain: methods (eg, appropriate presentation of discrimination and calibration). Red=‘high’, green=‘low’ or amber=‘unclear’ risk of bias.