TABLE 1.

Characteristics of studies^a

Characteristic	Result for:
	Observational studies (no intervention)	Studies of VAP prevention
		Nonpolymyxin studies	Topical polymyxin studies
			NCC	CC
Study characteristics
Source	Table S1	Table S2	Table S3	Table S4
Total no. of studies	93	60	17	25
No. of studies with origin from systematic reviewb	43	47	5	22
No. of studies originating from North American ICUsc	25	15	2	3
No. of studies with MV for >48 h for <90%d	6	1	1	4
No. of studies originating from trauma ICUse	21	13	2	8
No. of studies with bronchoscopic samplingf	57	25	6	6
Study publication yr (range)	1986–2015	1987–2016	1975–2017	1987–2007
Overall VAP prevention effect sizeg	NA	0.74; 0.67–0.82	0.40; 0.31–0.52	0.35; 0.29–0.43
Group characteristics
No. of patients per study group (median; IQR)h	290; 135–660	76; 60–149	84; 60–228	42; 31–74
VAP incidence per 100 patients (mean %; 95% CI) (no. of groups)
Cohort	20; 18–22 (102)	NA	NA	NA
Control	NA	21; 19–24 (60)	18; 11–30 (9)	33; 26–41 (24)
Intervention	NA	15; 13–18 (54)	10; 6–15 (15)	14; 11–18 (25)
PsVAP incidence per 100 patients (mean %; 95% CI) (no. of groups)
Cohort	4.6i; 4.0–5.3 (102)
Control		4.8j; 4.0–5.8 (60)	6.4k,l; 5.1–7.9 (9)	9.9l,m,n; 7.6–12.8 (24)
Intervention		3.5o; 2.8–4.3 (54)	1.6k,p,q; 1.0–4.5 (15)	3.8m,n,p; 2.9–5.0 (25)

^aNote that several studies had more than one control and/or intervention group. Hence, the number of groups does not equal the number of studies. Abbreviations: NCC, nonconcurrent control; CC, concurrent control; NA, not applicable.

^bStudies that were sourced from 20 systematic reviews (references available in supplemental material).

^cStudies originating from an ICU in Canada or the United States.

^dStudies for which fewer than 90% of patients were reported to receive >48 h of MV.

^eA trauma ICU was arbitrarily defined as an ICU with >50% of admissions for trauma.

^fBronchoscopic instead of tracheal sampling for VAP diagnosis.

^gSee Fig. S1 and S2 in the supplemental material.

^hData are the median and interquartile range (IQR).

ⁱThis value is the PsVAP benchmark as derived in Fig. S3.

^jSee Fig. S4.

^kSee Fig. S6.

^lThe incidences of PsVAP within control groups from CC and NCC design studies of topical polymyxin were statistically different (P = 0.049).

^mThe incidence of PsVAP within CC control groups of studies of topical polymyxin in which observer blinding was achieved by using a topical placebo was 10.9% (7.0 to 16.0%; n = 11) compared to 8.9% (6.2 to 12.7%; n = 14) in those not using a topical placebo.

ⁿSee Fig. S7.

^oSee Fig. S5.

^pThe incidences of PsVAP within intervention groups from CC and NCC design studies of topical polymyxin were statistically different (P = 0.017).

^qThe incidence of PsVAP among intervention groups of NCC studies of topical polymyxin after exclusion of two pre-1980 North American studies (37, 39) was 1.6% (1.0 to 2.8%).