Table 2.

 Results of sensitivity analysis for postoperative complications

Variable	Postoperative myocardial infarction			Need for repeat revascularisation				Mortality at maximum follow-up
	Fixed effects model	Random effects model		Fixed effects model	Random effects model			Fixed effects model	Random effects model
Overall
Odds ratio (95% CI)	0.88 (0.41 to 1.88)	0.91 (0.38 to 2.19)		3.62 (2.38 to 5.51)	3.66 (2.35 to 5.41)			0.77 (0.35 to 1.71)	0.83 (0.35 to 1.97)
Heterogeneity (P value)*	6.25 (0.40)			3.49 (0.84)				2.44 (0.66)
No of adverse events†:
PTCS	16/1248			185/1295				18/620
MIDCAB	10/587			28/635				10/294
Randomised
Odds ratio (95% CI)	1.36 (0.59 to 3.12)	1.3 (0.51 to 3.32)		4.63 (2.52 to 8.51)	4.55 (2.47 to 8.37)			0.57 (0.17 to 1.89)	0.63 (0.16 to 2.40)
Heterogeneity (P value)*	4.10 (0.39)			1.05 (0.96)				2.14 (0.54)
No of adverse events†:
PTCS	14/378			57/425				3/179
MIDCAB	9/329			14/377				6/175
Follow-up >12 months
Odds ratio (95% CI)	0.89 (0.38 to 2.07)	0.91 (0.29 to 2.79)		3.7 (2.42 to 5.68)		3.65 (2.39 to 5.59)		0.77 (0.35 to 1.71)	0.83 (0.35 to 1.99)
Heterogeneity (P value)*	6.24 (0.28)			3.12 (0.79)				2.44 (0.66)
No of adverse events†:
PTCS	12/1129			182/1179				18/620
MIDCAB	8/517			27/567				10/294
>70 patients in each group
Odds ratio (95% CI)	0.59 (0.22 to 1.55)	0.59 (0.21 to 1.7)		3.29 (2.08 to 5.2)		3.24 (2.05 to 5.13)		NA	NA
Heterogeneity (P value)*	2.24 (0.52)			2.30 (0.51)				NA
No of adverse events†:
PTCS	8/1097			162/1094				NA
MIDCAB	8/436			24/434				NA

PTCS=percutaneous transluminal coronary artery stenting; MIDCAB=minimally invasive direct coronary artery bypass; NA=not available.

*χ² test.

†Patients may have more than one event.