Present and future of prophylactic antibiotics for severe acute pancreatitis

Table 1 Recent meta-analyses on the outcomes of prophylactic antibiotics for severe acute pancreatitis

Meta-analysis	Year	No. of RCTs	Mortality		Infected necrosis		Recommendation
			Effect size	95% CI	Effect size	95% CI
Jafri et al[15]	2009	8	RR = 0.76	(0.49, 1.16)	RR = 0.79	(0.56, 1.11)	Unfavorable
Xu et al[16]	2008	8	RR = 0.76	(0.50, 1.18)	RR = 0.69	(0.50, 0.95)a	Pending
Bai et al[17]	2008	7	RR = 0.70	(0.42, 1.17)	RR = 0.81	(0.54, 1.22)	Unfavorable
Hart et al[18]	2008	7	OR = 0.71	(0.41, 1.23)	OR = 0.72	(0.45, 1.16)	Unfavorable
de Vries et al[19]	2007	6	ARR = 0.058	(-0.017, 0.134)	ARR = 0.055	(-0.084, 0.194)	Pending
Dambrauskas et al[20]	2007	10	RR = 0.76	(0.586, 0.976)	RR = 0.57	(0.418, 0.784)a	Pending
Villatoro et al[21]	2006	5	OR = 0.37	(0.17, 0.83)a	OR = 0.62	(0.35, 1.09)	Pending
Mazaki et al[22]	2006	6	RR = 0.78	(0.44, 1.39)	RR = 0.77	(0.54, 1.12)	Unfavorable
Xiong et al[23]	2006	6	RR = 0.54	(0.28, 1.04)	RR = 0.77	(0.48, 1.24)	Unfavorable

^aP < 0.05 antibiotic prophylaxis vs placebo/none; RCT: Randomized controlled trial; CI: Confidence interval; RR: Risk ratio; OR: Odds ratio; ARR:Absolute risk reduction.

Table 2 Mortality rates and related parameters between different periods

Period	Antibiotics1 (%)	Control2 (%)	ARR (%)	RRR (%)	NNT
Before 2000	5.26	18.18	12.92	71.07	8
From 2000	15.00	15.07	0.07	0.46	1429

¹The mortality rate of the subgroup before 2000 was significantly lower than the subgroup from 2000 (P = 0.019; by SPSS 13.0, Pearson χ² test, two-sided);

²No significant difference of mortality rate between two periods (P = 0.571; by SPSS 13.0, Pearson χ² test, two-sided). ARR: Absolute risk reduction; RRR: Relative risk reduction; NNT: Number needed to treat.

Table 3 Estimate of required sample size based on the pooling estimate from 2000

Δ	α	β	μα	μβ	p1	p2	|p1-p2|	Group	Study
0.10	0.05	0.10	1.9600	1.2816	0.1500	0.1507	0.0007	272	544
0.05	0.05	0.05	1.9600	1.9600	0.1500	0.1507	0.0007	1615	3230

Δ: Threshold of difference value; α: Possibility of type I error; β: Possibility of type II error; μα: Critical value corresponding to α; μβ: Critical value corresponding to β; p1: Possibility of mortality in prophylactic antibiotics group; p2: Possibility of mortality in placebo or blank control group; |p1-p2|: Absolute value of difference between two groups.

Table 4 Sensitivity analysis by adding a large-scaled randomized controlled trial

	Antibiotics		Placebo/none-treatment
	Mortality rate	No. of events	Mortality rate	No. of events
By best/worst data1	7.32%	40/544	27.78%	151/544
Meta-analysis3	RR = 0.76, 95% CI: 0.38-1.53, P = 0.44 (random effect model)
By pooling data2	15.00%	82/544	15.07%	82/544
Meta-analysis3	RR=1.00, 95% CI: 0.79-1.27, P = 0.99 (fixed effect model)

¹The mortality rate of antibiotic prophylaxis was obtained from the best outcome in a randomized controlled trial (RCT) after 2000, while that of placebo/none-treatment was from the worst outcome in another RCT after 2000;

²The mortality rates were from the pooling data of the subgroup from 2000 in the updated meta-analysis;

³The meta-analysis included both the 7 RCTs published from 2000 and an assumed large-scale RCT (by Cochrane RevMan 5.0, Mantel-Haenszel test, two-sided).