Table 3.
Endpoint analyses adjusted for all individuals
| Variable | Intervention (n = 681) | Control (n = 663) | GEE adjusted difference between groups (95% c.i.)/P | Difference between groups, mean (95% c.i.)/P |
|---|---|---|---|---|
| Death within 3 months | 96 (14.1) | 138 (20.8) | OR 1.69 (1.22,2.35)/0.002 | 6.7 (2.5,10.9)/0.002 |
| Death within 12 months | 134 (19.7) | 184 (27.8) | OR 1.70 (1.26,2.28)/<0.001 | 8.1 (3.4,12.8)/<0.001 |
| Duration of hospital stay (days), mean(s.d.), median (range) | 10.2(13.3), 7 (0–175.6) (n = 681) | 11.9(13.0), 7.5 (0.1–112.9) (n = 663) | LS mean −1.81 (−3.18,−0.45)/0.009 | −1.71 (−3.13,−0.31)/0.017 |
| ICU care | 133 (19.5) | 145 (21.9) | OR 1.19 (0.88,1.60)/0.26 | 2.3 (–2.1,6.8)/0.32 |
| Duration of ICU care (days), mean(s.d.), median (range) | 3.12(5.97), 1.29 (0.02–53.54) (n = 133) | 5.40(8.34), 2.1 (0.03–62.02) (n = 145) | LS mean −2.36 (−4.08,−0.65)/0.007 | −2.28 (−4.01,−0.59)/0.006 |
| Readmission to the ICU | 22 (3.2) | 30 (4.5) | OR 1.63 (0.91,2.93)/0.10 | 1.3 (−0.9,3.5)/0.28 |
| Surgical complications | – | – | LS mean −0.16 (−0.23,−0.09)/<0.001 | 0.0001 |
| No complications | 1 (0.1) | 7 (1.1) | – | 0.9 (−0.1; 1.9)/0.064 |
| Clavien–Dindo I–IIIa | 494 (72.5) | 407 (61.4) | OR 0.53 (0.41−0.69)/<0.001 | −11.2 (−16.3; −6.0)/<0.001 |
| Clavien–Dindo IIIb–IVb | 115 (16.9) | 141 (21.3) | OR 1.39 (1.04−1.84)/0.024 | 4.4 (0.0; 8.7)/0.048 |
| Clavien–Dindo V | 71 (10.4) | 108 (16.3) | OR 1.80 (1.25−2.59)/0.002 | 5.9 (2.1; 9.6)/0.002 |
Values are n (%) unless otherwise indicated. The adjusted OR was analysed by GENMOD (General Mode) with the generalized estimating equation (GEE) model with binary outcome and link function logit adjusted for age, intestinal ischaemia, faecal/purulent/other peritonitis, chronic obstructive lung disease, ischaemic heart disease, congestive heart failure, chronic renal failure, diabetes, obesity, smoking, ASA classification, sex, and cancer. The OR was analysed by GENMOD with the GEE model with binary outcome and link function logit. For comparisons between groups, Fisher’s exact test (lowest one-sided P value multiplied by two) was used for dichotomous variables. The confidence interval for dichotomous variables was the asymptotic Wald confidence limits with continuity correction. LS, Least Squares.