Table 3.
Results of sensitivity and subgroup analyses for 30-day in-hospital mortality
| Number of patients | Difference | |||
| Analyses | ICU | HDU | (95% CI) | P value |
| Overall cohort | ||||
| Propensity score matching | 4,982 | 4,982 | -7.2 (-10.0 to -4.4) | < .001 |
| Multivariable regression | 7,528 | 7,331 | -7.1 (-8.7 to -5.5) | < .001 |
| Overlap weight | 7,528 | 7,331 | -7.1 (-9.9 to -4.4) | < .001 |
| Instrumental variable | 7,528 | 7,331 | -10.1 (-14.3 to -5.8) | < .001 |
| Subgroup 1: Excluding patients who were admitted to hospitals with only ICU beds or only HDU beds | ||||
| Propensity score matching | 4,180 | 4,180 | -6.3 (-9.5 to -3.1) | < .001 |
| Multivariable regression | 5,911 | 5,648 | -7.2 (-9.1 to -5.4) | < .001 |
| Overlap weight | 5,911 | 5,648 | -7.3 (-10.5 to -4.1) | < .001 |
| Instrumental variable | 5,911 | 5,648 | -14.1 (-20.3 to -7.8) | < .001 |
| Subgroup 2: Excluding patients who were admitted to hospitals with both ICU beds and HDU beds | ||||
| Propensity score matching | 913 | 913 | -5.3 (-11.0 to 0.5) | .075 |
| Multivariable regression | 1,617 | 1,683 | -4.7 (-8.0 to -1.5) | .005 |
| Overlap weight | 1,617 | 1,683 | -4.8 (-11.1 to 1.5) | .13 |
| Instrumental variable | 1,617 | 1,683 | -3.7 (-8.5 to 1.1) | .14 |
| Subgroup 3: Excluding patients who died on the day of admission | ||||
| Propensity score matching | 4,935 | 4,935 | -6.5 (-8.4 to -4.7) | < .001 |
| Multivariable regression | 7,061 | 6,602 | -6.3 (-7.8 to -4.7) | < .001 |
| Overlap weight | 7,061 | 6,602 | -6.3 (-8.2 to -4.5) | < .001 |
| Instrumental variable | 7,061 | 6,602 | -9.3 (-14.0 to -4.6) | < .001 |
| Subgroup 4: ICUs with ICU management fees 1 or 2 vs. HDUs | ||||
| Propensity score matching | 1,026 | 1,026 | -8.7 (-12.9 to -4.5) | < .001 |
| Multivariable regression | 1,105 | 7,331 | -8.6 (-11.4 to -5.9) | < .001 |
| Overlap weight | 1,105 | 7,331 | -8.9 (-12.3 to -5.5) | < .001 |
| Instrumental variable | 1,105 | 7,331 | -12.6 (-22.8 to -2.5) | .014 |
| Subgroup 5: ICUs either with emergency and critical care unit management fees 2 or 4 or with ICU management fees 3 or 4 vs. HDUs | ||||
| Propensity score matching | 4,546 | 4,546 | -6.6 (-9.5 to -3.8) | < .001 |
| Multivariable regression | 6,423 | 7,331 | -6.7 (-8.4 to -5.1) | < .001 |
| Overlap weight | 6,423 | 7,331 | -6.8 (-9.6 to -3.9) | < .001 |
| Instrumental variable | 6,423 | 7,331 | -9.3 (-13.7 to -4.9) | < .001 |
| Subgroup 6: ICUs vs. HDUs with emergency and critical care unit management fees 1 or 3 | ||||
| Propensity score matching | 2,800 | 2,800 | -7.8 (-12.1 to -3.4) | < .001 |
| Multivariable regression | 7,528 | 4,634 | -8.4 (-10.6 to -6.3) | < .001 |
| Overlap weight | 7,528 | 4,634 | -8.4 (-12.6 to -4.1) | < .001 |
| Instrumental variable | 7,528 | 4,634 | -13.1 (-17.9 to -8.2) | < .001 |
| Subgroup 7: ICUs vs. HDUs with HDU management fees 1 or 2 | ||||
| Propensity score matching | 2,369 | 2,369 | -5.5 (-8.2 to -2.9) | < .001 |
| Multivariable regression | 7,528 | 2,697 | -5.2 (-7.3 to -3.2) | < .001 |
| Overlap weight | 7,528 | 2,697 | -5.7 (-8.2 to -3.2) | < .001 |
| Instrumental variable | 7,528 | 2,697 | -15.6 (-7.8 to 4.7) | .62 |
| Subgroup 8: Pathogen-confirmed pneumonia | ||||
| Propensity score matching | 2,714 | 2,714 | -6.7 (-9.3 to -4.1) | < .001 |
| Multivariable regression | 4,167 | 3,818 | -6.7 (-8.7 to -4.6) | < .001 |
| Overlap weight | 4,167 | 3,818 | -6.6 (-9.1 to -4.1) | < .001 |
| Instrumental variable | 4,167 | 3,818 | -9.0 (-15.0 to -3.0) | .004 |
| Subgroup 9: Pathogen-nonconfirmed pneumonia | ||||
| Propensity score matching | 2,223 | 2,223 | -8.4 (-12.7 to -4.1) | < .001 |
| Multivariable regression | 3,361 | 3,513 | -7.6 (-9.8 to -5.5) | < .001 |
| Overlap weight | 3,361 | 3,513 | -7.7 (-12.1 to -3.4) | 0.001 |
| Instrumental variable | 3,361 | 3,513 | -12.2 (-18.2 to -6.3) | < .001 |
ICU, intensive care unit; HDU, high-dependency care unit; CI, confidence interval