Highlights
In this 37-hospital North American retrospective cohort, prone positioning among eligible mechanically ventilated adults with persistent moderate-to-severe hypoxemia increased from 11.0% before the pandemic to 51.9% during the pandemic, then declined to 25.6% after the pandemic.
Compared with the pre-pandemic period, the adjusted odds of proning were 7.6-fold higher during the pandemic. However, pandemic-period use remained 2.7-fold higher than the post-pandemic period, indicating incomplete retention of the practice gains achieved during COVID-19.
Hospital-level variation remained substantial in all periods, suggesting that local culture, staffing, workflow, and implementation capacity strongly influence whether eligible patients receive this guideline-recommended intervention.
During the pandemic, patients with SARS-CoV-2 infection were much more likely to be proned than SARS-CoV-2-negative patients, but even non-COVID pandemic patients experienced increased proning compared with pre-pandemic practice.
Background
Prone positioning is one of the few supportive interventions in acute respiratory distress syndrome, or ARDS, that has demonstrated a survival benefit when applied appropriately in patients with moderate-to-severe hypoxemia. By turning a mechanically ventilated patient from the supine to the prone position, clinicians can improve ventilation-perfusion matching, recruit dorsal lung units, reduce regional overdistension, and lessen ventilator-induced lung injury. These physiologic effects are particularly relevant in severe diffuse lung injury, where dependent atelectasis and heterogeneous stress distribution contribute to refractory hypoxemia and organ failure.
Despite this biologic rationale and the mortality benefit seen in landmark randomized trials, implementation of prone positioning has historically lagged behind guideline recommendations. Before the COVID-19 pandemic, underuse of proning in eligible ARDS patients was repeatedly documented. Barriers have been familiar to ICU clinicians: the labor required to safely turn an intubated patient, concern about adverse events such as endotracheal tube dislodgement or pressure injury, uncertainty about eligibility, and variable local expertise or staffing resources.
The COVID-19 pandemic changed the visibility of proning dramatically. Severe viral pneumonia with profound hypoxemia, intense public and professional attention to respiratory support, and the need to use evidence-based nonpharmacologic therapies at scale all likely accelerated adoption. Many ICUs built formal proning teams, checklists, and protocols during surges. What remained unclear, however, was whether this implementation gain reflected a durable change in critical care practice or a temporary pandemic-specific response.
The present study by Barker and colleagues addresses that question directly by examining peri-pandemic trends in proning across 37 North American hospitals. The study is clinically important because it tests a broader implementation hypothesis: when a crisis drives rapid uptake of a guideline-supported therapy, does that behavior persist after the crisis recedes?
Study Design
This was a retrospective cohort study evaluating prone positioning in mechanically ventilated adults treated at 37 North American hospitals. The investigators compared three time periods: pre-pandemic, defined as January 2018 through February 2020; pandemic, defined as March 2020 through February 2022; and post-pandemic, defined as March 2022 through December 2024.
The analytic cohort included patients receiving invasive mechanical ventilation who met prespecified hypoxemia criteria consistent with persistent moderate-to-severe oxygenation impairment: Pao2/Fio2 less than or equal to 150 mm Hg, Fio2 at least 0.6, and positive end-expiratory pressure at least 5 cm H2O. These thresholds align with the patient profile in which proning is generally considered most evidence-based and most likely to improve outcome.
The exposure of interest was the time period of ICU admission. The primary process outcome was receipt of prone positioning within 12 hours of meeting hypoxemia criteria. The investigators also examined hospital-level variation in proning behavior and compared use according to SARS-CoV-2 status during the pandemic period.
A total of 5,944 patients were eligible for proning. Because the study was observational and retrospective, it was designed to characterize implementation rather than to estimate the causal effect of proning on mortality or ventilator outcomes in this cohort. That distinction is important: the study asks whether clinicians used the therapy, not whether the therapy worked.
Key Findings
Overall use changed dramatically across the three periods
Of 5,944 proning-eligible patients, 2,155 patients, or 36.2%, received prone positioning within 12 hours. The raw rate changed strikingly over time. Before the pandemic, only 11.0% of eligible patients were proned. During the pandemic, that figure rose to 51.9%. After the pandemic, use declined to 25.6%.
These numbers convey two important messages. First, the pre-pandemic baseline was low, consistent with longstanding concerns that prone positioning remained underused even after supportive trial data and guideline endorsement. Second, the pandemic demonstrated that much higher uptake was operationally feasible across diverse hospitals. Yet the post-pandemic decline indicates that the system-level changes required to sustain this practice were either incomplete or not maintained.
Adjusted comparisons confirmed a strong pandemic effect
In adjusted analyses, the odds of proning during the pandemic were substantially higher than before the pandemic, with an adjusted odds ratio of 7.6 and a 95% confidence interval of 5.5 to 10.4. This is a large effect size for a change in clinical implementation behavior and strongly supports the conclusion that the pandemic period altered practice in a major way.
When compared with the post-pandemic period, the pandemic period still showed significantly greater proning use, with an adjusted odds ratio of 2.7 and a 95% confidence interval of 1.8 to 3.9. Thus, proning did not simply return to the pre-pandemic baseline, but neither did pandemic-era adoption persist at its peak level.
Hospital-to-hospital variation remained substantial
A major contribution of the study is its quantification of interhospital variation using median odds ratios. This metric estimates the median change in odds of receiving the intervention for similar patients treated at a lower-use versus higher-use hospital. The median odds ratio was 2.9 pre-pandemic, with a 95% credible interval of 1.9 to 5.3; 1.9 during the pandemic, with a 95% credible interval of 1.6 to 2.3; and 2.3 post-pandemic, with a 95% credible interval of 1.8 to 3.3.
Clinically, these values indicate that where a patient was admitted materially affected whether they were proned, even after accounting for measured patient factors. Variation narrowed during the pandemic, perhaps reflecting widespread protocolization and heightened awareness, but it did not disappear. The increase in variation again after the pandemic suggests that some hospitals retained strong implementation pathways while others reverted toward older practice patterns.
SARS-CoV-2 status influenced proning during the pandemic
Among pandemic-period patients, those testing positive for SARS-CoV-2 were much more likely to be proned than SARS-CoV-2-negative patients, with an odds ratio of 5.1 and a 95% confidence interval of 4.1 to 5.6. This likely reflects the centrality of hypoxemic COVID-19 respiratory failure in ICU care during the pandemic, along with disease-specific educational efforts and the perception that proning was especially important in this phenotype.
Importantly, the pandemic effect was not limited to COVID-19 patients. Pandemic-period patients without SARS-CoV-2 infection were still more likely to be proned than pre-pandemic patients, with an odds ratio of 3.8 and a 95% confidence interval of 2.7 to 5.2. This suggests spillover of practice change beyond viral pneumonia, an encouraging sign that ICUs learned and applied proning more broadly.
However, when pandemic SARS-CoV-2-negative patients were compared with post-pandemic patients, the difference was smaller and no longer statistically significant, with an odds ratio of 1.3 and a 95% confidence interval of 0.9 to 1.8. This finding implies that at least some of the broader implementation gains were retained after the pandemic, but at a lower and less consistent level than during the emergency phase.
What the numbers mean in practice
For front-line clinicians, the most sobering number may be the post-pandemic rate itself: only about one in four patients who met the study’s proning eligibility criteria received the intervention within 12 hours. Even allowing for legitimate clinical exceptions not fully captured in retrospective data, this suggests a substantial remaining evidence-to-practice gap.
For ICU leaders, the data point toward a classic implementation problem rather than a knowledge problem alone. During COVID-19, many units clearly demonstrated the capacity to identify candidates quickly and mobilize teams for prone turns. The later decline therefore suggests erosion of infrastructure, competing priorities, staff turnover, loss of dedicated proning teams, or reduced protocol adherence once the immediate crisis subsided.
Clinical Interpretation
The study’s findings are highly plausible and align with what many clinicians observed in practice. Before COVID-19, proning was often seen as a specialized, burdensome maneuver performed mainly in highly protocolized ARDS care settings. During the pandemic, it became mainstream. High patient volumes, repeated exposure to severe hypoxemia, and urgency around nonpharmacologic interventions likely shifted ICU norms. In effect, the pandemic may have functioned as a large-scale implementation experiment.
From a translational standpoint, the study demonstrates that underuse of proven supportive care is not necessarily fixed. Practice can change quickly when institutions align staffing, education, operational workflows, and expectations. That is encouraging. At the same time, the post-pandemic decline underscores a harder truth in implementation science: adoption is easier than sustainment.
The persistence of between-hospital variation is especially instructive. It indicates that proning uptake is shaped not only by patient severity but by institutional factors such as nurse-to-patient ratios, respiratory therapy availability, clinician champions, multidisciplinary training, formal protocols, pressure-injury prevention pathways, and local beliefs about workload and risk. These are modifiable features of care delivery.
Current ARDS guidelines have generally supported prone ventilation for patients with moderate-to-severe ARDS, particularly when Pao2/Fio2 is below 150 mm Hg despite lung-protective ventilation. The present study does not challenge that evidence base; rather, it shows that real-world performance against guideline expectations remains inconsistent. That is exactly the kind of gap that quality improvement and implementation research should target.
Strengths and Limitations
Strengths
The study has several notable strengths. It includes a large multicenter cohort spanning 37 hospitals and a long observation window that captures pre-pandemic, pandemic, and post-pandemic practice. The eligibility criteria are clinically meaningful and focus on patients most likely to be considered for proning under contemporary ARDS care standards. The investigators also go beyond simple descriptive reporting by quantifying adjusted temporal associations and hospital-level heterogeneity.
Another strength is the inclusion of SARS-CoV-2-stratified analyses during the pandemic. This helps distinguish disease-specific behavior from broader ICU practice change and suggests that the pandemic had both targeted and general implementation effects.
Limitations
As with any retrospective observational study, several limitations deserve emphasis. First, proning eligibility was based on physiologic criteria available in the dataset, but some patients may have had legitimate reasons not to undergo prone positioning, such as hemodynamic instability, spinal precautions, open abdomen, recent surgery, elevated intracranial pressure concerns, or staffing limitations. Residual confounding in implementation studies is difficult to eliminate.
Second, the study evaluated proning within 12 hours of meeting hypoxemia criteria. That is a clinically sensible window, but some patients may have been proned later, which would not count toward the primary outcome. Similarly, the analysis does not provide detail here on duration of prone sessions, repeat sessions, adherence to lung-protective ventilation, neuromuscular blockade strategies, or other co-interventions that may have influenced decision-making.
Third, the study was not designed to estimate patient-centered outcome benefits associated with proning in this cohort. Therefore, one should not infer from these data alone whether the rise or fall in proning altered mortality, ventilator-free days, or ICU length of stay in the included hospitals.
Fourth, although the cohort is multicenter and North American, generalizability may still be limited by the participating institutions’ case mix, electronic data infrastructure, and ICU organization. Community hospitals, smaller systems, or non-North American settings may show different implementation patterns.
Finally, the abstract does not report funding details, and because this was a retrospective cohort study, there is no ClinicalTrials.gov registration number expected for an interventional trial. These are not defects, but they are relevant for transparency and interpretation.
Implications for Practice and Policy
The central practical message is straightforward: proning rates can be improved, but improvement is not self-sustaining. Hospitals that want to preserve gains should treat prone positioning as a system capability rather than an individual clinician preference. That means building durable structures around identification, execution, and safety monitoring.
Several implementation strategies follow naturally from these data. First, ICUs can embed automated electronic alerts or dashboards to identify patients meeting proning thresholds in near real time. Second, standardized protocols and checklists can reduce the cognitive and operational burden of the maneuver. Third, recurring multidisciplinary simulation and bedside training may counteract staff turnover. Fourth, designated proning teams or rapid-response staffing models could help maintain capacity during nights, weekends, and high-acuity periods. Fifth, audit-and-feedback programs comparing unit-level proning rates for eligible patients may reduce unwarranted variation.
For quality leaders and health systems, proning may also be a useful marker of broader ARDS care reliability. Units that struggle to sustain proning uptake may also have gaps in low tidal volume ventilation, conservative fluid strategies, or evidence-based sedation practices. In that sense, proning is both a therapy and a signal of ICU implementation maturity.
Conclusion
This multicenter North American study shows that prone positioning for mechanically ventilated adults with persistent moderate-to-severe hypoxemia was dramatically underused before COVID-19, rose sharply during the pandemic, and then declined after the pandemic, though not fully back to baseline. The findings suggest that the pandemic temporarily closed an important evidence-to-practice gap, but the gains were only partially sustained.
Perhaps the most actionable lesson is that the barriers to proning are not immutable. Many hospitals demonstrated that high uptake was achievable when urgency, training, and systems support aligned. The challenge now is to convert that temporary success into stable standard care for eligible ARDS patients. Future work should identify which organizational strategies best sustain proning implementation and whether improved delivery translates into measurable gains in survival and recovery at the health-system level.
Funding and ClinicalTrials.gov
Funding information was not reported in the provided abstract. This was a retrospective cohort study rather than a prospective interventional trial; no ClinicalTrials.gov registration number was provided in the source citation.
References
1. Barker AK, Nishimura A, Nuppnau M, Buell KG, Lyons PG, Liao WT, Park-Egan B, Schmid BE, Ingraham NE, Chaudhari V, Gao CA, Ortiz AC, Weissman GE, Chhikara K, Rojas JC, Amaral ACKB, Parker WF, Iwashyna TJ, Hager DN, Sjoding MW, Hochberg CH, Common Longitudinal ICU data Format (CLIF) Consortium. Prone Positioning in a North American Cohort of Hypoxemic Patients on Mechanical Ventilation. Critical Care Medicine. 2026-05-22. PMID: 42171428. Available at: https://pubmed.ncbi.nlm.nih.gov/42171428/
2. Guérin C, Reignier J, Richard JC, Beuret P, Gacouin A, Boulain T, Mercier E, Badet M, Mercat A, Baudin O, Clavel M, Chatellier D, Jaber S, Rosselli S, Mancebo J, Sirodot M, Hilbert G, Bengler C, Richecoeur J, Gainnier M, Bayle F, Bourdin G, Leray V, Girard R, Baboi L, Ayzac L; PROSEVA Study Group. Prone positioning in severe acute respiratory distress syndrome. New England Journal of Medicine. 2013;368(23):2159-2168.
3. Fan E, Del Sorbo L, Goligher EC, Hodgson CL, Munshi L, Walkey AJ, Adhikari NKJ, Amato MBP, Branson R, Brower RG, Ferguson ND, Gajic O, Gattinoni L, Hess D, Mancebo J, Mehta S, McAuley DF, Ranieri VM, Rubenfeld GD, Rubenfeld G, Seckel M, Slutsky AS, Talmor D, Thompson BT, Wunsch H, Uleryk E, Brodie D. An Official American Thoracic Society/European Society of Intensive Care Medicine/Society of Critical Care Medicine Clinical Practice Guideline: Mechanical Ventilation in Adult Patients with Acute Respiratory Distress Syndrome. American Journal of Respiratory and Critical Care Medicine. 2017;195(9):1253-1263.
4. Munshi L, Del Sorbo L, Adhikari NKJ, Hodgson CL, Wunsch H, Meade MO, Uleryk E, Mancebo J, Pesenti A, Ranieri VM, Fan E. Prone position for acute respiratory distress syndrome. A systematic review and meta-analysis. Annals of the American Thoracic Society. 2017;14(Supplement_4):S280-S288.
