Results of the fixed effects multivariate regression
| Effect on the proportion of patients waiting over 4 hours (in percentage points change) | Model (a) | 95% CIs | Model (b) | 95% CIs | |
| Hospital factors | Bed occupancy: | ||||
| 85% | 0 | (0.000 to 0.000) | |||
| 86% | 0.012 | (−0.001 to 0.025) | |||
| 87% | 0.009 | (−0.002 to 0.020) | |||
| 88% | 0.027*** | (0.015 to 0.038) | |||
| 89% | 0.023*** | (0.012 to 0.035) | |||
| 90% | 0.035*** | (0.022 to 0.047) | |||
| 91% | 0.035*** | (0.023 to 0.047) | |||
| 92% | 0.039*** | (0.026 to 0.052) | |||
| 93% | 0.043*** | (0.029 to 0.056) | |||
| 94% | 0.049*** | (0.036 to 0.063) | |||
| 95% | 0.052*** | (0.038 to 0.066) | |||
| 96% | 0.060*** | (0.046 to 0.074) | |||
| 97% | 0.062*** | (0.047 to 0.077) | |||
| 98% | 0.071*** | (0.056 to 0.086) | |||
| 99% | 0.084*** | (0.068 to 0.100) | |||
| 100% | 0.093*** | (0.075 to 0.111) | |||
| Admissions:discharges ratio | 0.037*** | (0.017 to 0.057) | |||
| L1.admissions:discharges ratio | 0.047*** | (0.034 to 0.059) | |||
| L2.admissions:discharges ratio | 0.054*** | (0.042 to 0.067) | |||
| Daily admissions ratio | 0.081*** | (0.063 to 0.100) | |||
| Total admissions | 0.000 | (−0.000 to 0.000) | |||
| Senior doctors | −0.047 | (−0.270 to 0.177) | −0.038 | (−0.272 to 0.197) | |
| Inpatient types | Long-stay patients | 0.069* | (0.008 to 0.130) | 0.128*** | (0.064 to 0.192) |
| Influenza | 0.384 | (−0.263 to 1.031) | 0.485 | (−0.125 to 1.095) | |
| Control variables | Yes | Yes | |||
| Observations | 11 790 | 11 544 | |||
| R 2 | 0.263 | 0.230 |
We analysed two adaptations of the model: model (a), which included indicator variables indicating the bed occupancy level in percentage points and the daily admissions ratio; and model (b), which omitted bed occupancy variables and instead included continuous variables measuring the patient flows that determined occupancy: total daily admissions and admissions to discharges ratio. Other control variables are the same in both models. 95% confidence intervals in brackets. Clustered standard errors. The results for the bed occupancy intervals below and equal to 85% are grouped in the 85% category—only 6% of observations of bed occupancy were below that level and the results of the regression on those intervals were not significant. Full tables can be found in the supplementary online material.
*p < 0.05, **p < 0.01, ***p < 0.001.