The effect of surgery on the outcome of treatment for multidrug-resistant tuberculosis: a systematic review and meta-analysis

Table 2 Summary of the meta-analyses for TB treatment outcomes

Outcome	# studies	% (n/N) surgery	% (n/N) non-surgery	Summary OR (95 % CI)	P-value*	P-value**	I²
Successful treatment	14 [22–35]	81.9 % (371/453)^a	59.7 % (1197/2006)^a	2.62 (1.94, 3.54)	<0.001	0.08	37.3 %
Successful treatment^b	13 [22–30, 32–35]	84.1 % (328/390)^a	59.6 % (1108/1858)^a	2.81 (2.07, 3.81)	<0.001	0.1	30.8 %
Successful treatment^c	17 [22–35, 38–40]	81.7 % (379/464)^a	59.2 % (1304/2199)^a	2.52 (1.91, 3.32)	<0.001	0.2	26.0 %
Cure	5 [22, 23, 25, 28, 34]	75.2 % (118/157)	54.9 % (308/561)	3.03 (1.59, 5.78)	0.001	0.07	54.2 %
Death	5 [22, 23, 25, 28, 34]	5.8 % (11/191)	7.2 % (52/720)	0.82 (0.41, 1.64)	0.6	0.8	0.0 %
Loss to follow-up	4 [22, 23, 25, 28]	3.8 % (6/156)	12.6 % (77/613)	0.35 (0.15, 0.81)^d	0.01	0.7	0.0 %
Treatment failure	5 [22, 23, 25, 28, 34]	4.2 % (8/191)	11.4 % (82/720)	0.38 (0.18, 0.81)	0.01	0.99	0.0 %
Transfer out	2 [22, 23]	0 % (0/54)	2.0 % (6/350)	Not analysed

^aTwo studies only reported OR and 95%CI and so do not contribute to the denominator and numbers of outcomes reported in the table
^bFrom a sensitivity analysis which excluded one study [31] for which there may be some overlap of patients also reported in Kim et al. 2008 [32]
^cFrom a sensitivity analysis including three studies identified in Fox et al. which had <10 patients in the surgery arm. For two studies [39, 40] all patients in the surgery group had a successful outcome and so 0.5 was added to all cells so that OR and 95 % CI could be calculated
^dOne study [22] had zero patients lost to follow-up in the surgery arm, so 0.5 was added to all cells so that OR and 95 % CI could be calculated
*P-value for null hypothesis OR = 1; ** P-value for heterogeneity

ISSN: 1471-2334