Incidences of Pseudomonas aeruginosa-Associated Ventilator-Associated Pneumonia within Studies of Respiratory Tract Applications of Polymyxin: Testing the Stoutenbeek Concurrency Postulates

Abstract

Regimens containing topical polymyxin appear highly effective at preventing ventilator-associated pneumonia (VAP) overall and, more so, VAP caused by Gram-negative bacteria. However, Stoutenbeek's postulates that VAP incidences within studies of topical antibiotics depend on the context of whether the component (control and intervention) groups of each study were concurrent versus nonconcurrent remain untested. The literature was searched for concurrent control (CC) versus nonconcurrent control (NCC) designed studies of respiratory tract applications of topical polymyxin to mechanically ventilated (MV) patients that reported incidences of Pseudomonas-associated ventilator-associated pneumonia (PsVAP). Studies of various interventions other than topical polymyxin (nonpolymyxin studies) served to provide additional points of reference. The PsVAP incidences within the component groups of all studies were benchmarked against groups from observational studies. This was undertaken by meta-regression using generalized estimating equation methods. Dot plots, caterpillar plots, and funnel plots enable visual benchmarking. The PsVAP benchmark (and 95% confidence interval [CI]) derived from 102 observational groups is 4.6% (4.0 to 5.3%). In contrast, the mean PsVAP within NCC polymyxin intervention groups (1.6%; CI, 1.0 to 4.5%) is lower than that of all other component group categories. The mean PsVAP within CC polymyxin control groups (9.9%; CI, 7.6 to 12.8%) is higher than that of all other component group categories. The PsVAP incidences of control and intervention groups of studies of respiratory tract applications of polymyxin are dependent on whether the groups were within a concurrent versus nonconcurrent study. Stoutenbeek's concurrency postulates are validated.

Keywords: Pseudomonas; antibiotic prophylaxis; intensive care; mechanical ventilation; polymyxin; selective digestive decontamination; study design; ventilator-associated pneumonia.

FIG 1

Search method, screening criteria, and resulting classification of eligible studies and subsequent derivation of component groups. The arrows numbered 1 to 6 indicate the following steps. (1) An electronic search for systematic reviews containing potentially eligible studies was performed using the search terms “ventilator-associated pneumonia,” “mechanical ventilation,” and “intensive care unit,” each combined with either “meta-analysis” or “systematic review,” up to December 2015. (2) Studies were streamed into one of four categories: studies in which there was no intervention (observational studies), studies with topical polymyxin with or without a CC design, and studies of nonpolymyxin methods of VAP prevention. The studies of nonpolymyxin methods of VAP prevention encompass a broad range of methods. (3) The studies were screened against inclusion and exclusion criteria. (4) A hand search was undertaken for additional studies. (5) Eligible studies were then collated, and any duplicate studies were removed. (6) The component groups, being control (rectangles), intervention (ovals), and observation (diamond) groups, were extracted from each study. Note that the total numbers do not tally, as some systematic reviews provided studies in more than one category and some studies provided groups in more than one category.

FIG 2

Incidence of PsVAP versus the benchmark. Results for the component (Ob, observational [open symbols]; C, control, and I, intervention [closed triangles]) groups of studies of either topical polymyxin- or nonpolymyxin-based methods of VAP prevention are shown (95% CI of summary incidence). The PsVAP benchmark is the summary mean (central unbroken vertical line) derived from the observational studies. Note that the x axis is a logit scale and groups with a zero count have a continuity correction (N + 0.5) to enable them to appear in the plot. These data are displayed in more detail as caterpillar plots (see Fig. S3 to S7 in the supplemental material).

FIG 3

Pseudomonas VAP incidence per 100 patients versus year of publication. Symbols: open gray circles, observational groups; open triangles, control group from nonpolymyxin studies; solid gray squares, control group from NCC polymyxin studies; solid black triangles, control group from CC polymyxin studies. Note that the y axis is a logit scale and that groups with a zero count have a continuity correction (N + 0.5) to enable them to appear in the plot. The regression line is derived from only the observational groups.