Context: Although meta-analyses of randomized trials have shown that selective digestive decontamination (SDD) prevents nosocomial pneumonia in critically ill patients, the influence of trial quality on the effectiveness of SDD has not been rigorously evaluated.
Objective: To assess the methodological quality of individual studies of SDD and its relation to the reported effects on pneumonia and mortality.
Design: Thirty-two studies were identified in a MEDLINE and reference list search and their methodological quality was assessed using a scoring system (range, 0-13 points) based on allocation and concealment, patient selection, patient characteristics, blinding of the intervention, and the definition of pneumonia.
Main outcome measure: Methodological quality of the primary trials and its effect on the relative risk reductions (RRRs) of SDD on pneumonia and mortality.
Results: The mean (SD) methodological quality score was 7.8 (2.9) (range, 1-11). The RRRs ranged from -0.1 to 1.0 for pneumonia and from -0.1 to 0.6 for mortality. The methodological quality score was associated with the RRR for pneumonia so that for each quality-point added, the RRR decreased by 5.8% (95% confidence interval, 2.4%-9.3%). No association between trial quality and the impact of SDD was found on mortality. Of the individual trial quality characteristics, patient selection, allocation of intervention, and blinding most strongly influenced the treatment effect.
Conclusions: The inverse relationship between methodological quality score and the benefit of SDD on the incidence of pneumonia may have resulted in overly optimistic estimates of SDD in prior meta-analyses. This emphasizes the importance of rigorous trial design in evaluating preventive interventions in the intensive care unit.