Multivariable association between predictors and outcome events after correction for optimism, including the model constant
| Predictor | Adjusted regression coefficient*† (95% CI) | P value‡ |
| Number of comorbidities | 0.125 (0.0663 to 0.184) | <0.001 |
| Estimated glomerular filtration rate/10 (ml/min/1.73 m2) | −0.0308 (−0.0628 to 0.0012) | 0.059 |
| White cell count (109/L) | 0.0234 (−0.0007 to 0.0476) | 0.057 |
| Number of medicines | 0.0347 (0.0063 to 0.0630) | 0.016 |
| Previous allergy§ | 0.272 (0.0591 to 0.484) | 0.012 |
| Nervous system and mental disorders§ | 0.354 (0.0156 to 0.693) | 0.040 |
| Respiratory system§ | −0.234 (−0.493 to 0.0253) | 0.077 |
| Gastrointestinal system§ | −0.533 (−0.911 to −0.156) | 0.006 |
| Aminoglycosides and glycopeptides§ | 0.331 (−0.0457 to 0.708) | 0.085 |
| Other systemic antimicrobials§ | 0.311 (0.0777 to 0.545) | 0.009 |
| Epilepsy medicines§ | 0.385 (0.0950 to 0.675) | 0.009 |
| Constant | −1.674 |
*Relationship between the independent and dependent variable (amount of increase in predicted log odds of the outcome event that would be predicted by a one unit increase in the independent variable).
†Original regression coefficients corrected by uniform linear shrinkage factor (0.855).
‡Test for difference between admissions with and without occurrence of outcome event. Obtained from multivariable regression modelling.
§Categorical exposure variable. For the purposes of calculating the predicted risk for individual patients, categorical variables are coded as ‘one’ if present and ‘zero’ if absent.