The authors reply
First, the study populations included in our review were heterogeneous. In particular, the study by Masri et al (2) conducted enrolled a trauma population which was younger, included more males, and had a higher rate of tracheostomy than other studies. We did take this into account in our review and rated down the quality of evidence for two outcomes (constipation and duration of mechanical ventilation) for heterogeneity. Of note, the study by Masri et al (2) does not disproportionately contribute to statistical heterogeneity, and its removal interestingly increases, rather than decreases, heterogeneity for the outcome of duration of mechanical ventilation. Further, our review specifically excluded any trials of patients with primary gastrointestinal disorders, and although it is possible that a small number of such patients were enrolled in the included trials, this is unlikely to have had a major impact on the results of the review. We agree with Prat et al (1) to caution clinicians against applying the results of this meta-analysis to patients with primary gastrointestinal diagnoses.
Second, despite the use of lactulose in all included trials, we acknowledge the variable interventions representing bowel protocols that were tested. Cointerventions were not well described, and these may have differed across trials. Ideally, randomization and effective blinding (as was performed in two trials) would minimize these effects, but their overall impact is uncertain and unreported. We did prespecify the timing of protocols for a subgroup analysis; however, timing did not seem to explain any of the observed heterogeneity of effects, and all such comparisons were between-trial rather than within-trial subgroup comparisons, rendering interpretation challenging (3). Although the observational study by Guardioloa et al (4) suggests that the timing of a bowel protocol may influence its effect, we focused on higher level evidence by including only randomized trials in this review. In everyday practice, timely initiation of bowel care protocols poses a challenge, as found in a recent mixed-methods study conducted by one of the authors, which revealed the difficulties in achieving high clinician compliance in using bowel care protocols (Warren and Kent, unpublished data).
Finally, the definition of constipation that we chose was based upon its incorporation in observational studies and randomized trials, and recommendations from the European Society of Intensive Care Medicine Working Group on Abdominal Problems (5). Whether this definition will endure in future studies is unclear. As we suggested, a new definition may be clinimetrically developed, ensuring its reproducibility, utility at the bedside, and association with patient-important outcomes, thus making it a suitable target for therapeutic interventions in clinical trials.
We thank Prat et al (1) for highlighting these limitations in the existing evidence and agree that more research is needed to guide practice in this often neglected aspect of critical care.