Table 1.

Analysis of publication bias.

Pathogen	Resistance	Number of studies	Egger’s test* [12]	Begg’s test [13]
E. coli	Carbapenems	60	t = −1.65, df = 58, p-value = 0.1039	z = 1.24, p-value = 0.2155
K. pneumoniae		50	t = −0.44, df = 48, p-value = 0.6638	z = −0.26, p-value = 0.7952
P. aeruginosa		56	t = −1.25, df = 54, p-value = 0.2177	z = −1.19, p-value = 0.2348
Enterobacter spp.		7	-**	-**
A. baumannii (complex)		36	t = −2.40, df = 34, p-value = 0.0221	z = −2.49, p-value = 0.0126
E. coli	Third-generation cephalosporins	58	t = −1.65, df = 58, p-value = 0.1039	z = 1.24, p-value = 0.2155
K. pneumoniae		48	t = −1.19, df = 44, p-value = 0.2401	z = −0.74, p-value = 0.4595
Enterobacter spp.		8	-**	-**
S. aureus	Methicillin	80	t = 0.47, df = 78, p-value = 0.6365	z = 0.61, p-value = 0.5411
	Vancomycin	39	t = 2.21, df = 37, p-value = 0.0337***	z = 4.92, p-value < 0.0001***

* Predictor: standard error, weight: inverse variance

** Number of studies too small to test for small study effects (k_min = 10)

*** The majority of studies reported proportions of 0 or close to 0. As stated above, funnel plots were found to be an inaccurate method of assessing publication bias in proportion studies with low proportion outcomes [8].