Principal findings
In this systematic review and meta-analysis of 17 trials, awake prone positioning was associated with a decreased risk of endotracheal intubation compared with usual care in adults with hypoxemic respiratory failure due to covid-19. The evidence of reduction in endotracheal intubation with awake prone positioning was of high certainty and the results were consistent across multiple sensitivity and bayesian analyses. On average, awake prone positioning resulted in 55 fewer intubations per 1000 patients (95% confidence interval 87 to 19 fewer intubations). However, awake prone positioning probably had little to no effect on mortality, ventilator-free days, ICU length of stay, hospital length of stay, escalation of oxygen treatment, or mode of oxygen delivery. Awake prone positioning is generally safe, with infrequent adverse events that include unintentional catheter dislodgement, discomfort, nausea, and skin breakdown.
Comparison with other studies
As this systematic review represents a large number of patients and trials, the precision of the effect estimates is increased.1012 Including a larger number of trials addresses a limitation of previously published meta-analyses,1012 particularly by limiting any one trial from being excessively weighted in a meta-analysis. We also used two complementary statistical approaches (frequentist and bayesian) that supported the robustness of the results. The use of a bayesian approach allowed integration of prior information with our pooled data to determine a clinically useful summary of this information. Specifically, the bayesian approach provides probabilities of a benefit (or harm) with awake prone positioning given the observed data across varying previous beliefs (priors) about its effectiveness. For example, the posterior probability of a relative reduction of at least 5% in endotracheal intubation was high (≥0.90) across all degrees of prior beliefs about its effectiveness, given the data. In contrast, the posterior probability of a 5% relative reduction in mortality was 0.93 only if the prior beliefs about the effectiveness of awake prone positioning were strong (ie, using an enthusiastic prior). Many clinicians and patients would consider a 5% reduction in endotracheal intubation or mortality as clinically meaningful, particularly for a safe non-pharmacologic intervention.
The findings in this review were robust through a variety of different sensitivity analyses. The studies included in this systematic review differed from those in another recent meta-analysis,12 which included a randomized trial by Gad.38 We excluded that trial because it compared awake prone positioning without non-invasive ventilation with non-invasive ventilation, so the groups differed by the presence of prone positioning and by mode of respiratory support. In contrast, a trial by Hashemian and colleagues was incorporated in our review as it included non-invasive ventilation in both the usual care and the prone positioning groups.15 We a priori planned to include quasi-randomized trials in our analysis, anticipating a small number of eligible studies to be available for meta-analysis. One quasi-randomized trial was identified,16 in which allocation was based on patients’ medical record numbers, with even numbers receiving usual care and odd numbers receiving awake prone positioning. Owing to lack of concealed randomization, this study was assessed to be at high risk of selection bias. Although this quasi-randomized trial was not included in the primary analysis, when it was included in a sensitivity analysis the effect estimate did not change notably, further supporting the robustness of the results. The meta-analysis’s results are of important clinical relevance, as awake prone positioning is an inexpensive, non-pharmacological treatment that can be applied in a variety of hospital settings. In addition, awake prone positioning can be used in both low and middle income countries and high income countries, as shown by the geographic location of the studies in this systematic review.
Although we found no effect of awake prone positioning on mortality, a favorable effect cannot be excluded. Conversely, a reduction in the rate of endotracheal intubation was not associated with an increase in mortality, suggesting that patients were not put at risk by delaying intubation. To further support the safety of this intervention, the absolute rate of serious adverse events in the awake prone positioning group was low across trials. Also, downstream outcomes that could be associated with a reduction in endotracheal intubation, such as ventilator-free days and ICU and hospital length of stay were not statistically different between groups. Nevertheless, the effect estimates were consistently in the direction favoring awake prone positioning but with wide 95% confidence intervals. It may be that reducing intubation does not affect these outcomes, or that the lower number of studies reporting these secondary outcomes limited precision to detect small effect sizes.
The mechanism for how awake prone positioning reduces endotracheal intubation remains uncertain. Adherence to longer duration of prone positioning may be an effect modifier on the outcome of endotracheal intubation. It has been hypothesized that longer duration of awake prone positioning may be more effective, similar to placing patients in the prone position who are receiving invasive ventilation.513 However, unlike patients receiving invasive ventilation who were placed in the prone positioning, awake patients are not sedated and not receiving neuromuscular blocking agents. This key difference may explain why none of the included trials that specified target durations for awake prone positioning met the prescribed dose in their intervention group. The intervention may be limited by patient tolerance as data suggest that awake patients may not cope well with long periods of prone positioning.33 Although many patients can place themselves in a prone position, others may need encouragement or assistance to do so for longer durations, which may require the availability of staff or other resources. Dedicated teams can increase adherence to prone positioning for intubated patients,3940 but data on the utility of this approach for non-intubated patients are limited. Other strategies to improve adherence, such as smart phone based guidance and reminders, did not result in better adherence in one trial.37 Thus, the benefits of awake prone positioning need to be weighed against the resources and staff needed to ensure safe adherence to the intervention. Thus, it remains uncertain whether better adherence to longer duration of awake prone positioning does modify the effect of the intervention. Our subgroup analysis suggested that in trials in which the median duration of awake prone positioning was ≥5 hours/day, the reduction in endotracheal intubation risk was relatively greater. However, the interaction test P value was not significant. Similarly, using meta-regression, the association between duration of awake prone positioning at the trial level and the effect size was not significant. Although these analyses suggest a potential association between duration of awake prone positioning and efficacy, they may be underpowered or potentially confounded since duration of prone positioning was not randomized and should be considered hypothesis generating. Even if an association exists between duration and efficacy, the optimal duration of awake prone positioning remains unknown. This question could be better evaluated in future randomized trials comparing various durations of prone positioning that are balanced with tolerability. In our other subgroup analyses, trials with more severe baseline hypoxemia, those performed in mixed hospital settings, and those performed in low to middle income countries tended to have larger effects. None of the interaction test P values were, however, significant, so we caution against over-interpretation of these findings. To most appropriately and efficiently allocate resources to deliver this intervention, future studies could aim to determine which patient subgroups, if any, benefit most from awake prone positioning.
Strengths and limitations of this study
This meta-analysis should be interpreted within the context of its limitations. First, although we explored potential effect modification in subgroup analyses based on trial level characteristics, lack of individual patient data limited the ability to evaluate effect modification more precisely. For example, while many of the included trials overlapped the pre-vaccine and post-vaccine eras of the pandemic, it is unknown whether covid-19 vaccination status modifies the effectiveness of awake prone positioning. This could not be evaluated with the available data, but effect modifiers could be better studied using individual patient data meta-analysis. Second, owing to differences between the targeted and achieved duration of awake prone positioning across studies, we are unable to conclude whether there is an optimal duration of prone positioning for patients to benefit. Third, some of the planned analyses were limited because of heterogeneity in the definition and reporting of certain outcomes such as oxygenation, missing trial level data for some outcomes in the prospective meta-analysis,11 or because a few studies reported some outcomes, limiting precision and certainty. Fourth, the decision to intubate a patient can vary, with no fixed criteria. Furthermore, factors influencing the decision to intubate a patient were likely variable between providers and institutions and may have changed over the course of the pandemic. Despite this variability, the meta-analysis suggests there is high certainty in this finding based on the wide range of study locations (14 trials conducted in 12 different countries), and this finding is further supported by a secondary bayesian analysis and multiple sensitivity analyses. Finally, studies that are still in progress or were unpublished at the time this meta-analysis was completed might not be included and could influence the results. Although given the size and number of studies included in this review, such an influence would be unlikely unless the unpublished study was large, had a large treatment effect, or had multiple studies showing alternative effects to what we found. Strengths of this study include the adherence to quality standards for meta-analysis, use of GRADE to assess the certainty of evidence, and duplicate review of the search strategy and analysis for the primary outcome. This report includes a larger number of trials and patients than previous meta-analyses, uses rigorous sensitivity analyses to challenge the robustness of the primary analysis, and uses complementary preplanned bayesian analyses with a priori assumptions in addition to the traditional frequentist approach.
Conclusions
Awake prone positioning compared with usual care reduced the risk of endotracheal intubation in adults with hypoxemic respiratory failure due to covid-19. Evidence on the effects of awake prone positioning on mortality or other secondary outcomes was, however, inconclusive. Adverse events related to awake prone positioning were uncommon, highlighting the safety of this intervention. However, adherence to the target duration of prone positioning was low in many trials. Thus, clinicians and patients must balance the goal of avoiding endotracheal intubation with the tolerability of awake prone positioning and availability of staff resources to encourage and assist patients. Future trials should aim to determine strategies to improve tolerability and adherence, assess the optimal duration of awake prone positioning, and determine the effect of awake prone positioning from other causes of hypoxemic respiratory failure.
What is already known on this topic
-
Awake prone positioning is an inexpensive, non-pharmacological treatment that can be applied readily and easily in a variety of hospital settings
-
The effect of awake prone positioning in patients with covid-19 related hypoxemic respiratory failure on endotracheal intubation and other outcomes remains uncertain
What this study adds
-
In this systematic review and meta-analysis of 17 randomized trials, awake prone positioning for hypoxemic respiratory failure due to covid-19 reduced the risk of endotracheal intubation, but evidence for the effect on mortality or other outcomes was inconclusive
-
Adverse events during awake prone positioning were uncommon and rarely serious