Table 1. Probabilities of outcomes and mean LOS 9 .
| Description | Value |
|---|---|
| Probability of being transferred from the ICU to a general ward | 79.7% |
| Probability of being discharged directly from the ICU to home | 1.3% |
| Probability of dying in the ICU | 19.0% |
| Probability of dying in a general ward | 7.9% |
| Probability of new infection acquired in the ICU | 11.4% |
| LOS, pre-ICU, days (s.d.) | 5.60 (9.91) |
| LOS ICU—alive patients, days (s.d.) | 6.10 (11.40) |
| LOS ICU—dead patients, days (s.d.) | 8.80 (16.04) |
| LOS, post-ICU—alive patients, days (s.d.) | 23.20 (23.70) |
| LOS, post-ICU—dead patients, days (s.d.) | 18.10 (24.10) |
Abbreviations: ICU, intensive care unit; LOS, length of stay; s.d., standard deviation.
Note: To mathematically represent LOS variables in the model, a Weibull distribution was fitted to these data sets, yielding a satisfying ‘goodness of fit'. Other distributions were used appropriately according to the type of data (Dirichlet distributions for probability of transfer from ICU to general ward, discharge and death in the ICU; beta distributions for probability of death in a general ward and of new infection in the ICU).