Prone positioning of nonintubated patients with COVID-19 in Australian intensive care units Authors: Barry Johnston, Hannah Rotherham, Peinan Zhao, Aidan Burrell, Andrew Udy Published: 2024 ABSTRACT ObjectiveTo describe the use of and outcomes from awake prone positioning (APP) in nonintubated patients with COVID-19 in Australian intensive care units (ICUs) in comparison to those who did not receive APP, and to explore the temporal relationship between publication of APP research and changes in clinical practice.DesignMulticentre, observational cohort study.SettingSeventy-eight Australian ICUs participating in SPRINT-SARI Australia.ParticipantsAdult patients with confirmed COVID-19 admitted to ICU from 27 February 2020 until 30 June 2022.Main outcomes measuresProportion of patients receiving APP, rates of invasive ventilation, hospital length of stay (LOS), in-hospital mortality.Results4711 patients were included in the analysis, of whom 28.6% (1347/4711) underwent APP. Use of APP rapidly increased during the Delta wave and then subsequently declined. Over this period, there were a total of 30 publications on APP. APP patients received a median of 2 (IQR 1–4) days prone positioning, were less unwell (median APACHE-II 13.0 vs. 15.0, p < 0.001), and were less likely to require invasive ventilation (27.9% vs. 34.9%, p < 0.001). Overall, there was no difference in hospital LOS (median 14 vs. 13 days, P = 0.420) or in-hospital mortality (HR 0.95, 0.8–1.11) in those that did and did not receive APP. However, in patients requiring invasive ventilation after their first day in the ICU, not receiving APP was associated with earlier time to intubation (median 1 vs. 3 days, p < 0.001) and lower adjusted in-hospital mortality (HR 0.70, CI 0.54–0.90).ConclusionsAPP was rapidly adopted into practice within Australian ICUs during the COVID-19 pandemic at the same time as a growing number of publications on the topic. A lower frequency of invasive ventilation was noted with APP overall, but in those who eventually required this intervention, APP was associated with greater risk-adjusted in-hospital mortality. Introduction The benefits of prone positioning (PP) in mechanically ventilated patients with acute hypoxaemic respiratory failure (AHRF) include both improved oxygenation and an overall reduction in mortality.1,2 However, the use of awake prone positioning (APP) in nonintubated patients with AHRF have only recently been described. It was rapidly implemented into clinical practice during the COVID-19 pandemic. To date, COVID-19 in Australia has been characterised by a number of ‘waves. The original wave occurred between January 2020 and mid-June 2021, the Delta wave between mid-June 2021 and mid-December 2021, and the Omicron wave from mid-December 2021 to present. Over the time period of this study, there were 8,132,210 reported COVID-19 cases in Australia.3,4 Low-cost and noninvasive therapies for management of respiratory failure have a particular appeal during times of resource strain, such as the COVID-19 pandemic. APP is defined by an awake patient undergoing noninvasive respiratory support who is lying in the prone position. The duration of APP each day varies between studies, ranging from 1 h to 16 h per day.[5], [6], [7] There has been widespread support for this procedure due to its potential therapeutic benefits, perceived minimal risk, ease of implementation, as well as the ability to deliver it in the hospital outside of the ICU.[8], [9], [10] Early observational studies demonstrated physiological benefits, such as improvement in oxygenation, and an association with a reduction in intubation rates and mortality.6,11 However, later randomised control trials and meta-analyses demonstrated inconsistent findings and high-quality evidence to support its use remains lacking.12,13 The COVID-19 pandemic led to the rapid publication of clinical research, including the widespread dissemination of research findings prior to peer review. The international urgency to reduce the impact of COVID-19-related illness on patients and healthcare systems led to the rapid adoption of medical therapies and interventions despite the absence of robust high-quality evidence. Burrell et al. described the influence of the preprint release of the RECOVERY Trial on corticosteroid use in Australian ICUs and highlight the issue of rapid translation of the medical literature into practice and the potential implications.14 Prior to the Delta variant of COVID-19 in Australia, APP was not routinely practiced within the ICU. The nature of adoption and use of APP within Australian ICUs has not been described. Hence, the aims of this study were to: a) describe the use of APP in COVID-19 cases admitted to Australian ICUs between February 2020 and June 2022, b) describe any temporal association between APP use, and research publications on the topic and c) to explore the association between APP and patient-centred outcomes, such as the requirement for invasive ventilation, ICU and hospital length of stay, use of supportive therapies in the ICU and in-hospital mortality. Material and methods Study design, setting and participants Prospective, multicentre observational cohort study of adult patients with confirmed COVID-19 infection admitted to one of 78 Australian ICUs participating in the short period incidence study of severe acute respiratory infection (SPRINT-SARI) Australia project.15 All adult patients aged 16 years or older were included patients with proven incidental COVID-19 were excluded. Data for patients transferred between study ICUs were linked and counted as a single admission. This also applied to ICU re-admissions within the same hospitalisation episode. The study period was 27 February 2020 to 30 June 2022. Reporting of the study findings adheres to the Strengthening and Reporting of Observational Studies in Epidemiology (STROBE) statement.16 Data collection Data were collected by research coordinators at each individual site using the case report form previously described and were entered into a web-based REDCap data management system.15 The data collected included demographics, clinical features, ICU interventions, COVID-19 therapies and hospital outcomes. Definitions Awake prone days were defined as any ICU admission day that had a recorded episode of prone positioning when invasive mechanical ventilation did not occur. Patients were categorised as an awake prone admission if they had one or more awake prone days during their ICU admission. The primary outcome of this study was the proportion of cases receiving APP over the study period. This was then correlated with published APP reports overtime. Secondary outcomes included rates of invasive mechanical ventilation, ICU and hospital length of stay (LOS), the use of supportive therapies in the ICU, and in-hospital mortality. Literature search To describe the uptake of APP in Australian ICUs in relation to publications of clinical research, a literature review was undertaken using PubMed. Key words used in the search included ‘intensive care’, ‘awake prone’, and ‘COVID-19’. Only studies in English were included. Statistical analysis Descriptive statistics for all variables were calculated. Continuous variables are reported as medians (with interquartile ranges) and categorical variables as percentages. Statistical comparisons were performed using the student's t-test for continuous variables and chi-square test for categorical variables, where appropriate. Multi-variable analyses (MVA) were performed using the Cox proportional hazards model to assess the hazard ratio of in-hospital mortality and invasive ventilation of patients who underwent APP vs. patients who had no APP. Key variables including age, BMI, Acute Physiology and Chronic Health Evaluation (APACHE) II score, number of comorbidities, use of non-invasive ventilationor high flow nasal prongs, use of extracorporeal membrane oxygenation, use of continuous renal replacement therapy, and time period of ICU admission were included as covariates. Missing data were imputed using multiple imputation models (continuous variables - predictive mean matching, binary variables–logistic regression; categorical variables with more than two levels–polytomous regression) prior to MVA. Statistical significance was considered as a P value < 0.05. All statistical analysis was conducted using R v.4.0.4. Ethics approval The Human Research Ethics Committee of Alfred Health approved data collection and their analysis for this report and waived the requirement for informed consent by individual patients (HREC/16/Alfred/59). Results Details of cohort Between 27 February 2022 and 30 June 2022, 4711 patients were recorded in the SPRINT-SARI Australia database from 78 participating ICUs, of whom 28.6% (1347/4711) underwent at least one episode of awake prone positioning. There was a low incidence of repeat admissions between groups, 0.5% (7/1347) of APP patients vs. 0.3% (11/3364) of non-APP patients were re-admitted to the ICU during their index hospitalisation. Literature review Prior to the COVID-19 pandemic, we identified <5 publications involving APP.[17], [18], [19] During the study period, there was a total of 30 publications on APP. The number of research publications involving APP increased steadily over time, largely representing the Northern Hemisphere experience. Early observational studies tended to suggest benefit, but this outcome was not replicated in the later publication of randomised controlled trials and meta-analyses. Further details concerning these papers are provided in the Supplementary Appendix Table 1.Table 1A comparison of the baseline characteristics, interventions and outcomes of all COVID-19 patients who underwent awake prone positioning versus those who did not.Table 1APP (N = 1347)No APP (N = 3364)Overall (N = 4711)P-valueBaseline characteristicsAge at admission Mean (SD)56.6 (15.5)58.3 (16.8)57.8 (16.4)0.002Age group at admission <60704 (52.3%)1605 (47.7%)2309 (49.0%)<0.001 60 - 69345 (25.6%)785 (23.3%)1130 (24.0%) 70 - 79231 (17.1%)679 (20.2%)910 (19.3%) >8067 (5.0%)295 (8.8%)362 (7.7%)Sex Male873 (64.8%)2083 (61.9%)2956 (62.7%)0.072 Female474 (35.2%)1279 (38.0%)1753 (37.2%)Any prone positioning Yes1347 (100%)518 (15.4%)1865 (39.6%)<0.001Received COVID-19 vaccine Yes475 (35.3%)1569 (46.6%)2044 (43.4%)<0.001 No833 (61.8%)1358 (40.4%)2191 (46.5%)Pregnant Yes19 (1.4%)91 (2.7%)110 (2.3%)0.016 No402 (29.8%)1012 (30.1%)1414 (30.0%)APACHE-II on day 1 Mean (SD)13.1 (5.20)15.9 (7.18)14.9 (6.69)<0.001Number of Comorbidities 0369 (27.4%)1207 (35.9%)1576 (33.5%)<0.001 1397 (29.5%)674 (20.0%)1071 (22.7%) 2280 (20.8%)555 (16.5%)835 (17.7%) ≥3301 (22.3%)928 (27.6%)1229 (26.1%)BMI at admission Median [IQR]30.8 [26.6, 35.8]30.1 [25.6, 36.1]30.4 [26.0, 36.0]0.798BMI groupNormal weight156 (11.6%)451 (13.4%)607 (12.9%)<0.001Obese - Class I299 (22.2%)520 (15.5%)819 (17.4%)Obese - Class II148 (11.0%)297 (8.8%)445 (9.4%)Obese - Class III160 (11.9%)389 (11.6%)549 (11.7%)Overweight345 (25.6%)640 (19.0%)985 (20.9%)Underweight2 (0.1%)58 (1.7%)60 (1.3%)InterventionsTocilizumab Yes163 (12.1%)289 (8.6%)452 (9.6%)<0.001 No1184 (87.9%)3075 (91.4%)4259 (90.4%)Neutralizing antibodies Yes49 (3.6%)115 (3.4%)164 (3.5%)0.130 No189 (14.0%)325 (9.7%)514 (10.9%)Remdesivir Yes141 (10.5%)104 (3.1%)245 (5.2%)<0.001 No1206 (89.5%)3260 (96.9%)4466 (94.8%)Invasive mechanical ventilation at any time during ICU admission Yes376 (27.9%)1175 (34.9%)1551 (32.9%)<0.001 No971 (72.1%)2189 (65.1%)3160 (67.1%)Invasive mechanical ventilation after day 1 Yes343 (25.5%)680 (20.2%)1023 (21.7%)<0.001 No33 (2.4%)495 (14.7%)528 (11.2%)Noninvasive mechanical ventilation Yes786 (58.4%)965 (28.7%)1751 (37.2%)<0.001 No561 (41.6%)2399 (71.3%)2960 (62.8%)High-flow nasal canula oxygen therapy Yes1316 (97.7%)1674 (49.8%)2990 (63.5%)<0.001 No31 (2.3%)1690 (50.2%)1721 (36.5%)Extracorporeal membrane oxygenation Yes33 (2.4%)53 (1.6%)86 (1.8%)0.057 No1314 (97.6%)3311 (98.4%)4625 (98.2%)Renal replacement therapy Yes66 (4.9%)231 (6.9%)297 (6.3%)0.015 No1281 (95.1%)3133 (93.1%)4414 (93.7%)OutcomesDays of awake proning Median [IQR]2.00 [1.00, 4.00]NANANADay of ICU admission to first awake prone Median [IQR]0 [0, 0]NANANATime from ICU admission to intubation (days) Median [IQR]3.00 [1.00, 6.00]1.00 [0, 1.00]1.00 [0, 2.00]<0.001Days of mechanical ventilation (days)b Median [IQR]9.00 [5.00, 16.0]8.00 [3.91, 14.0]8.00 [4.00, 14.0]0.076Length of ICU stay (days)a Median [IQR]5.84 [3.01, 11.7]4.47 [1.96, 10.7]4.96 [2.25, 10.9]0.047Length of stay in hospital (days)a Median [IQR]14.0 [8.89, 22.3]12.9 [7.44, 22.4]13.2 [7.89, 22.4]0.426Patient status Discharged home935 (69.4%)2092 (62.2%)3027 (64.3%)<0.001 Hospital ward6 (0.4%)47 (1.4%)53 (1.1%) Mortality - Hospital Ward31 (2.3%)137 (4.1%)168 (3.6%) Mortality - ICU189 (14.0%)491 (14.6%)680 (14.4%) Palliative care1 (0.1%)10 (0.3%)11 (0.2%) Transfer to other hospital/facility58 (4.3%)196 (5.8%)254 (5.4%) Transfer to rehabilitation105 (7.8%)303 (9.0%)408 (8.7%) Unknown22 (1.6%)88 (2.6%)110 (2.3%)APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation.aLength of ICU and hospital stay were censored at 90 days.bDuration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. Table 1A comparison of the baseline characteristics, interventions and outcomes of all COVID-19 patients who underwent awake prone positioning versus those who did not.Table 1APP (N = 1347)No APP (N = 3364)Overall (N = 4711)P-valueBaseline characteristicsAge at admission Mean (SD)56.6 (15.5)58.3 (16.8)57.8 (16.4)0.002Age group at admission <60704 (52.3%)1605 (47.7%)2309 (49.0%)<0.001 60 - 69345 (25.6%)785 (23.3%)1130 (24.0%) 70 - 79231 (17.1%)679 (20.2%)910 (19.3%) >8067 (5.0%)295 (8.8%)362 (7.7%)Sex Male873 (64.8%)2083 (61.9%)2956 (62.7%)0.072 Female474 (35.2%)1279 (38.0%)1753 (37.2%)Any prone positioning Yes1347 (100%)518 (15.4%)1865 (39.6%)<0.001Received COVID-19 vaccine Yes475 (35.3%)1569 (46.6%)2044 (43.4%)<0.001 No833 (61.8%)1358 (40.4%)2191 (46.5%)Pregnant Yes19 (1.4%)91 (2.7%)110 (2.3%)0.016 No402 (29.8%)1012 (30.1%)1414 (30.0%)APACHE-II on day 1 Mean (SD)13.1 (5.20)15.9 (7.18)14.9 (6.69)<0.001Number of Comorbidities 0369 (27.4%)1207 (35.9%)1576 (33.5%)<0.001 1397 (29.5%)674 (20.0%)1071 (22.7%) 2280 (20.8%)555 (16.5%)835 (17.7%) ≥3301 (22.3%)928 (27.6%)1229 (26.1%)BMI at admission Median [IQR]30.8 [26.6, 35.8]30.1 [25.6, 36.1]30.4 [26.0, 36.0]0.798BMI groupNormal weight156 (11.6%)451 (13.4%)607 (12.9%)<0.001Obese - Class I299 (22.2%)520 (15.5%)819 (17.4%)Obese - Class II148 (11.0%)297 (8.8%)445 (9.4%)Obese - Class III160 (11.9%)389 (11.6%)549 (11.7%)Overweight345 (25.6%)640 (19.0%)985 (20.9%)Underweight2 (0.1%)58 (1.7%)60 (1.3%)InterventionsTocilizumab Yes163 (12.1%)289 (8.6%)452 (9.6%)<0.001 No1184 (87.9%)3075 (91.4%)4259 (90.4%)Neutralizing antibodies Yes49 (3.6%)115 (3.4%)164 (3.5%)0.130 No189 (14.0%)325 (9.7%)514 (10.9%)Remdesivir Yes141 (10.5%)104 (3.1%)245 (5.2%)<0.001 No1206 (89.5%)3260 (96.9%)4466 (94.8%)Invasive mechanical ventilation at any time during ICU admission Yes376 (27.9%)1175 (34.9%)1551 (32.9%)<0.001 No971 (72.1%)2189 (65.1%)3160 (67.1%)Invasive mechanical ventilation after day 1 Yes343 (25.5%)680 (20.2%)1023 (21.7%)<0.001 No33 (2.4%)495 (14.7%)528 (11.2%)Noninvasive mechanical ventilation Yes786 (58.4%)965 (28.7%)1751 (37.2%)<0.001 No561 (41.6%)2399 (71.3%)2960 (62.8%)High-flow nasal canula oxygen therapy Yes1316 (97.7%)1674 (49.8%)2990 (63.5%)<0.001 No31 (2.3%)1690 (50.2%)1721 (36.5%)Extracorporeal membrane oxygenation Yes33 (2.4%)53 (1.6%)86 (1.8%)0.057 No1314 (97.6%)3311 (98.4%)4625 (98.2%)Renal replacement therapy Yes66 (4.9%)231 (6.9%)297 (6.3%)0.015 No1281 (95.1%)3133 (93.1%)4414 (93.7%)OutcomesDays of awake proning Median [IQR]2.00 [1.00, 4.00]NANANADay of ICU admission to first awake prone Median [IQR]0 [0, 0]NANANATime from ICU admission to intubation (days) Median [IQR]3.00 [1.00, 6.00]1.00 [0, 1.00]1.00 [0, 2.00]<0.001Days of mechanical ventilation (days)b Median [IQR]9.00 [5.00, 16.0]8.00 [3.91, 14.0]8.00 [4.00, 14.0]0.076Length of ICU stay (days)a Median [IQR]5.84 [3.01, 11.7]4.47 [1.96, 10.7]4.96 [2.25, 10.9]0.047Length of stay in hospital (days)a Median [IQR]14.0 [8.89, 22.3]12.9 [7.44, 22.4]13.2 [7.89, 22.4]0.426Patient status Discharged home935 (69.4%)2092 (62.2%)3027 (64.3%)<0.001 Hospital ward6 (0.4%)47 (1.4%)53 (1.1%) Mortality - Hospital Ward31 (2.3%)137 (4.1%)168 (3.6%) Mortality - ICU189 (14.0%)491 (14.6%)680 (14.4%) Palliative care1 (0.1%)10 (0.3%)11 (0.2%) Transfer to other hospital/facility58 (4.3%)196 (5.8%)254 (5.4%) Transfer to rehabilitation105 (7.8%)303 (9.0%)408 (8.7%) Unknown22 (1.6%)88 (2.6%)110 (2.3%)APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation.aLength of ICU and hospital stay were censored at 90 days.bDuration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. A comparison of the baseline characteristics, interventions and outcomes of all COVID-19 patients who underwent awake prone positioning versus those who did not. APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation. Length of ICU and hospital stay were censored at 90 days. Duration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. Frequency of awake prone positioning use and change over time With the increase in the number of APP publications, the use of APP in Australian ICUs increased significantly (Fig. 1). During the first wave of the COVID-19 pandemic in Australia in early 2020, the use of APP was infrequent. However, from August 2021 to December 2021, approximately one-third of patients requiring ICU admission received APP. From January 2022, the proportion of patients undergoing APP declined, with <5% of patients undergoing APP in June 2022 at the end of the study period. The proportion of patients receiving APP per month is presented in the Supplementary Appendix Fig. 1.Fig. 1The proportion of Australian ICU admissions with patients who received at least one episode of APP over the COVID-19 pandemic and the cumulative number of peer-review publications on APP. Includes all adult patients (age≥16) who were admitted to ICU during the period of 2021-06-01 to 2022-05-31 (indicated by dotted lines). Proven incidental patients were excluded.Fig. 1 The proportion of Australian ICU admissions with patients who received at least one episode of APP over the COVID-19 pandemic and the cumulative number of peer-review publications on APP. Includes all adult patients (age≥16) who were admitted to ICU during the period of 2021-06-01 to 2022-05-31 (indicated by dotted lines). Proven incidental patients were excluded. Associations between awake prone positioning and patient outcomes A comparison of patients who did and did not receive APP is provided in Table 1. Rates of missing data are provided in the Supplement. In both groups, patients were predominantly male (64.8% vs. 61.9%, p = 0.072) and the median BMI was similar (30.8 vs. 30.1, p = 0.798). Compared with patients who did not undergo APP, patients receiving awake positioning were younger (median age 58 vs. 60 years, p = 0.002) and were less likely to have received at least one dose of a COVID-19 vaccine (35.3 vs. 46.6, <0.001). Of note, there were 19 pregnant females who received APP. Overall, patients who received APP were less likely to receive invasive mechanical ventilation (27.9 %v 34.0%, p < 0.001), but more likely to receive noninvasive ventilation (58.4% vs. 28.7%, p < 0.001) and high-flow nasal oxygen (97.7% vs. 49.8%, p < 0.01). Patients who did not receive APP were more unwell, with a higher median APACHE-II score on admission (15.0 vs. 13, p < 0.001) and more likely to have 3 or more comorbidities (27.6% vs. 22.3, p < 0.001) (Table 1). In patients who underwent APP, the median number of days of APP was 2 days (IQR 1–4 days). For patients who required initiation of invasive ventilation after day 1 of their ICU admission, compared to those who did not have APP, APP patients had a longer time to intubation (3 vs. 1 days, p < 0.001) and ICU LOS (15.2 vs. 11.8 days, p < 0.001) but a similar duration of mechanical ventilation (9 vs. 8 days, p = 0.170) (Table 2). Patients who had APP and were intubated after day 1 of ICU admission were less likely to be discharged home (36.2% vs. 46.3%, p < 0.001), and had a higher crude ICU mortality (35.6% vs. 22.1%, p < 0.001).Table 2A comparison of the baseline characteristics, interventions and outcomes of all patients who were not ventilated on the first day of ICU admission but had at least one episode of invasive ventilation during ICU admission underwent awake prone positioning versus those who did not.Table 2APP (N = 343)No APP (N = 680)Overall (N = 1023)P-valueAge at admission Mean (SD)58.7 (13.2)56.3 (14.8)57.1 (14.3)0.0089Age group at admission <60157 (45.8%)366 (53.8%)523 (51.1%)0.0412 60 - 69110 (32.1%)175 (25.7%)285 (27.9%) 70 - 7971 (20.7%)122 (17.9%)193 (18.9%) >805 (1.5%)17 (2.5%)22 (2.2%)Sex Male225 (65.6%)443 (65.1%)668 (65.3%)0.9658 Female118 (34.4%)236 (34.7%)354 (34.6%)Prone positioning Yes343 (100%)296 (43.5%)639 (62.5%)<0.001Received COVID-19 vaccine Yes113 (32.9%)246 (36.2%)359 (35.1%)0.0506Pregnant Yes4 (1.2%)14 (2.1%)18 (1.8%)0.3969APACHE-II on day 1 Mean (SD)14.2 (5.13)17.4 (7.23)16.3 (6.75)<0.001Number of Comorbidities ≥387 (25.4%)173 (25.4%)260 (25.4%)0.5812 087 (25.4%)196 (28.8%)283 (27.7%) 195 (27.7%)166 (24.4%)261 (25.5%) 274 (21.6%)145 (21.3%)219 (21.4%)BMI at admission Median [IQR]30.8 [26.6, 35.8]31.5 [26.1, 37.5]31.1 [26.4, 36.8]0.276BMI group Normal weight43 (12.5%)93 (13.7%)136 (13.3%)<0.001 Obese - Class I86 (25.1%)126 (18.5%)212 (20.7%) Obese - Class II48 (14.0%)76 (11.2%)124 (12.1%) Obese - Class III40 (11.7%)106 (15.6%)146 (14.3%) Overweight96 (28.0%)131 (19.3%)227 (22.2%) Underweight0 (0%)17 (2.5%)17 (1.7%)Tocilizumab Yes42 (12.2%)104 (15.3%)146 (14.3%)0.2219Neutralizing antibodies Yes18 (5.2%)15 (2.2%)33 (3.2%)0.0187Remdesivir Yes21 (6.1%)25 (3.7%)46 (4.5%)0.1047Noninvasive mechanical ventilation Yes264 (77.0%)291 (42.8%)555 (54.3%)<0.001High-flow nasal canula oxygen therapy Yes340 (99.1%)454 (66.8%)794 (77.6%)<0.001Extracorporeal membrane oxygenation Yes31 (9.0%)25 (3.7%)56 (5.5%)<0.001Renal replacement therapy Yes45 (13.1%)110 (16.2%)155 (15.2%)0.2321 No298 (86.9%)570 (83.8%)868 (84.8%)Re-admitted to ICU Yes2 (0.6%)3 (0.4%)5 (0.5%)1Days of awake proning Median [IQR]2.00 [1.00, 4.00]NANANADay of ICU admission to first awake prone Median [IQR]0 [0, 0]NANANATime from ICU admission to intubation Median [IQR]3.00 [2.00, 7.00]1.00 [1.00, 2.00]1.00 [1.00, 3.00]<0.001Days of mechanical ventilation (days)b Median [IQR]9.00 [6.00, 16.0]8.00 [4.00, 14.0]9.00 [4.00, 15.0]0.170Length of ICU stay (days)a Median [IQR]15.2 [9.60, 24.2]11.8 [6.30, 20.7]13.2 [7.47, 22.3]<0.001Length of stay in hospital (days)a Median [IQR]22.3 [15.5, 33.5]20.0 [12.1, 31.3]20.7 [13.3, 32.2]0.029Patient status Discharged home124 (36.2%)315 (46.3%)439 (42.9%)<0.001 Hospital ward4 (1.2%)11 (1.6%)15 (1.5%) Mortality - Hospital Ward4 (1.2%)20 (2.9%)24 (2.3%) Mortality - ICU122 (35.6%)150 (22.1%)272 (26.6%) Transfer to other hospital/facility22 (6.4%)62 (9.1%)84 (8.2%) Transfer to rehabilitation48 (14.0%)101 (14.9%)149 (14.6%) Unknown19 (5.5%)20 (2.9%)39 (3.8%) Palliative care0 (0%)1 (0.1%)1 (0.1%)APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation.aLength of ICU and hospital stay were censored at 90 days.bDuration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. Table 2A comparison of the baseline characteristics, interventions and outcomes of all patients who were not ventilated on the first day of ICU admission but had at least one episode of invasive ventilation during ICU admission underwent awake prone positioning versus those who did not.Table 2APP (N = 343)No APP (N = 680)Overall (N = 1023)P-valueAge at admission Mean (SD)58.7 (13.2)56.3 (14.8)57.1 (14.3)0.0089Age group at admission <60157 (45.8%)366 (53.8%)523 (51.1%)0.0412 60 - 69110 (32.1%)175 (25.7%)285 (27.9%) 70 - 7971 (20.7%)122 (17.9%)193 (18.9%) >805 (1.5%)17 (2.5%)22 (2.2%)Sex Male225 (65.6%)443 (65.1%)668 (65.3%)0.9658 Female118 (34.4%)236 (34.7%)354 (34.6%)Prone positioning Yes343 (100%)296 (43.5%)639 (62.5%)<0.001Received COVID-19 vaccine Yes113 (32.9%)246 (36.2%)359 (35.1%)0.0506Pregnant Yes4 (1.2%)14 (2.1%)18 (1.8%)0.3969APACHE-II on day 1 Mean (SD)14.2 (5.13)17.4 (7.23)16.3 (6.75)<0.001Number of Comorbidities ≥387 (25.4%)173 (25.4%)260 (25.4%)0.5812 087 (25.4%)196 (28.8%)283 (27.7%) 195 (27.7%)166 (24.4%)261 (25.5%) 274 (21.6%)145 (21.3%)219 (21.4%)BMI at admission Median [IQR]30.8 [26.6, 35.8]31.5 [26.1, 37.5]31.1 [26.4, 36.8]0.276BMI group Normal weight43 (12.5%)93 (13.7%)136 (13.3%)<0.001 Obese - Class I86 (25.1%)126 (18.5%)212 (20.7%) Obese - Class II48 (14.0%)76 (11.2%)124 (12.1%) Obese - Class III40 (11.7%)106 (15.6%)146 (14.3%) Overweight96 (28.0%)131 (19.3%)227 (22.2%) Underweight0 (0%)17 (2.5%)17 (1.7%)Tocilizumab Yes42 (12.2%)104 (15.3%)146 (14.3%)0.2219Neutralizing antibodies Yes18 (5.2%)15 (2.2%)33 (3.2%)0.0187Remdesivir Yes21 (6.1%)25 (3.7%)46 (4.5%)0.1047Noninvasive mechanical ventilation Yes264 (77.0%)291 (42.8%)555 (54.3%)<0.001High-flow nasal canula oxygen therapy Yes340 (99.1%)454 (66.8%)794 (77.6%)<0.001Extracorporeal membrane oxygenation Yes31 (9.0%)25 (3.7%)56 (5.5%)<0.001Renal replacement therapy Yes45 (13.1%)110 (16.2%)155 (15.2%)0.2321 No298 (86.9%)570 (83.8%)868 (84.8%)Re-admitted to ICU Yes2 (0.6%)3 (0.4%)5 (0.5%)1Days of awake proning Median [IQR]2.00 [1.00, 4.00]NANANADay of ICU admission to first awake prone Median [IQR]0 [0, 0]NANANATime from ICU admission to intubation Median [IQR]3.00 [2.00, 7.00]1.00 [1.00, 2.00]1.00 [1.00, 3.00]<0.001Days of mechanical ventilation (days)b Median [IQR]9.00 [6.00, 16.0]8.00 [4.00, 14.0]9.00 [4.00, 15.0]0.170Length of ICU stay (days)a Median [IQR]15.2 [9.60, 24.2]11.8 [6.30, 20.7]13.2 [7.47, 22.3]<0.001Length of stay in hospital (days)a Median [IQR]22.3 [15.5, 33.5]20.0 [12.1, 31.3]20.7 [13.3, 32.2]0.029Patient status Discharged home124 (36.2%)315 (46.3%)439 (42.9%)<0.001 Hospital ward4 (1.2%)11 (1.6%)15 (1.5%) Mortality - Hospital Ward4 (1.2%)20 (2.9%)24 (2.3%) Mortality - ICU122 (35.6%)150 (22.1%)272 (26.6%) Transfer to other hospital/facility22 (6.4%)62 (9.1%)84 (8.2%) Transfer to rehabilitation48 (14.0%)101 (14.9%)149 (14.6%) Unknown19 (5.5%)20 (2.9%)39 (3.8%) Palliative care0 (0%)1 (0.1%)1 (0.1%)APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation.aLength of ICU and hospital stay were censored at 90 days.bDuration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. A comparison of the baseline characteristics, interventions and outcomes of all patients who were not ventilated on the first day of ICU admission but had at least one episode of invasive ventilation during ICU admission underwent awake prone positioning versus those who did not. APP, awake prone positioning; ICU, intensive care units; BMI, body mass index; APACHE-II, Acute Physiology and Chronic Health Evaluation. Length of ICU and hospital stay were censored at 90 days. Duration of ventilation was censored at 90 days; Median and IQR were calculated based on patients who had at least one episode of IMV. Overall, APP in the ICU was not associated with any significant difference in risk-adjusted in-hospital mortality (HR = 0.94, 95% CI = 0.8–1.1). However, patients with advanced age, higher APACHE-II scores, the requirement for CRRT, and an increased number of comorbidities, were at greater risk of death (p < 0.001) (Fig. 2A).Fig. 2Cox proportional hazards model of (A) in-hospital mortality for all patients; (B) in-hospital mortality of patients who underwent IMV after day 1; (C) IMV of patients who had at least one episode of IMV after day 1. Missing data were handled with multiple imputation.Fig. 2 Cox proportional hazards model of (A) in-hospital mortality for all patients; (B) in-hospital mortality of patients who underwent IMV after day 1; (C) IMV of patients who had at least one episode of IMV after day 1. Missing data were handled with multiple imputation. In patients who required invasive ventilation after their first day in the ICU, those who did not undergo APP were more likely to be intubated earlier (HR 2.18, CI 1.9–2.5, p < 0.001) (Fig. 2C) and were less likely to die in hospital (HR 0.70, CI 0.54–0.90, p = 0.005 (Fig. 2B). These findings were similar on complete-case analysis (see Supplement Fig. 2). Discussion Key findings Our study describes the rapid uptake of APP into clinical practice in Australian ICUs over late 2021 and early 2022. The rate of APP peaked during the ‘delta’ wave of the pandemic, where 39.9% underwent APP. Following this, the rates of APP declined, so that <5% underwent APP towards the end of the study period. APP was always initiated on the first day of ICU admission and was associated with increased use of both noninvasive ventilation and high-flow nasal oxygen. A median of 2 days of APP was provided. APP patients were less likely to receive invasive mechanical ventilation overall; however, in those that required mechanical ventilation after day 1 in the ICU, they were intubated later and had higher risk-adjusted mortality. Relationship to previous studies Our findings are in keeping with the multiple systematic reviews and meta-analyses published in 2022,[20], [21], [22] which demonstrate a reduced need for intubation in those patients undergoing APP. However, these studies did not demonstrate a benefit on mortality or ICU and hospital length of stay. There remains a paucity of data on the duration and intensity of APP episodes that offers the greatest physiological benefit. Study implications The description of a significant change in clinical practice across Australian ICUs over a short period is a unique feature of this study. We hypothesise that the rapid implementation of APP is likely to be due to a combination of factors, including the growing evidence base for the procedure, periods of high ICU demand where measures aimed at reducing ICU admission and invasive ventilation were paramount, and the willingness of clinical staff to adopt and adapt novel interventions during the COVID-19 pandemic, when new treatments were scarce. APP is also an appealing intervention. It is low cost, potentially low risk and when the patient can self-prone, is easily implemented at the bedside without the need for significant ICU resources such as staffing. In the early phases of the pandemic, there were high levels of interest in novel interventions to treat COVID-19. Many new therapies and interventions were implemented with varying degrees of evidence to support them. Information was disseminated in a variety of new ways, including social media and the use of ‘preprints’. In the case of APP, the evidence base has evolved over time. Initial publications, which suggested benefit, were often small single centre observational studies focusing on physiological outcomes. The later publication of RCTs and meta-analyses with mixed results suggested a potential benefit, but high-quality evidence remains lacking. Indeed, our results support the assertion that the use of APP needs greater investigation. Here, although APP overall was associated with a lower requirement for invasive ventilation, in those where this was eventually required, in-hospital mortality was greater. There may be multiple explanations for this finding, including the potential deleterious effects of delayed intubation (where patient self-induced lung injury worsens the underlying pulmonary insult), or simply the progression of the underlying disease, where “failure to thrive” despite APP selects a high-risk phenotype. Unfortunately, our data are insufficient to delineate this. Additionally, the practice of APP is not described here, and the number and duration of APP episodes is likely to have varied between patients and individual sites. This highlights the need for well-developed selection criteria, the ability to readily identify nonresponders and a standardised approach to APP. For example, Ibarra et al. described predictors for treatment success, which included an RR < 25 at enrolment and an APP duration >8 h per day.23 The role of early adopters is vitally important to understand both the benefit and harm of new therapies prior to widespread implementation. Critical care registries and well-conducted observational studies can often provide timely information on evolving therapies and interventions. This emerging data should be continually reviewed, and the publication of both positive and negative results equally encouraged. We must exercise some caution before widely adopting a new therapy, with consideration of either early adoption or de-adoption of therapy as supported by evidence.24 Strengths and limitations This is the first study to provide a longitudinal description of APP practice across 78 Australian ICUs during the COVID-19 pandemic. It is a prospective observational study that captured a majority of critically ill patients with COVID-19 infection in Australia. However, we acknowledge there are several limitations. Some key clinical details, such as assessment of frailty, patient mobility and chest x-ray changes were not available. Additionally, some details of APP, including duration, daily frequency, and physiological response were not recorded. Furthermore, we only have details of APP in relation to the timing of ICU admission as opposed to the phase of illness. It is possible that the role of APP varies over the course of illness. We are also unable to describe patient selection criteria or practice variation across centres, nor do we have information related to adverse events. We believe that the prevalence of APP is likely to be underestimated, as APP on the same day as intubation is unable to be determined. Additionally, limitations of medical treatment (LOMT) for individual patients are unknown and likely introduces confounding bias for intubation rates and mortality, i.e. possibly more patients not suitable for intubation were trialled with APP. However, the group of patients who received APP were younger with fewer comorbidities and so unlikely to have significant LOMT. In addition, a traditional systematic review of all literature published on APP was not conducted, and as such, it is possible that we have under reported the number of published studies on this topic. Finally, as with all studies of this nature, there were missing data. This reflects the dynamic nature of data collection during the COVID-19 pandemic, where ICU research resources were often stretched. Critically, data are likely to be missing at random, and our findings concerning the risk-adjusted association with in-hospital mortality were similar with both complete case analyses and multiple imputation. Conclusions APP was rapidly adopted over the course of the COVID-19 pandemic in Australian ICUs. This was temporarily associated with publication of awake proning clinical research, and practice peaked during a period of high ICU demand. APP was associated with a reduced need for invasive ventilation overall; however, the patients who derive benefit and the optimal dose of APP remain unknown. The rapid translation of medical evidence into practice can lead to the adoption of interventions before high-quality evidence is available and should be undertaken cautiously. Funding SPRINT-SARI Australia is supported by the Department of Health, Commonwealth of Australia (Standing Deed SON60002733). CRediT authorship contribution statement Barry Johnston: Resources, Writing – original draft, Writing – Review & Editing, Visualisation; Hannah Rotherham: Conceptualisation, Resources, Writing – original draft, Writing – Review & Editing, Visualisation; Peinan Zhao: Methodology, Formal Analysis, Data Curation, Visualisation; Aidan Burrell: Conceptualisation, Resources, Writing – Review & Editing, Visualisation; Andrew Udy: Conceptualisation, Resources, Writing – Review & Editing, Visualisation. Research data for this article The datasets generated and/or analysed during the current study are not publicly available due ethics but are available from SPRINT-SARI Australia on reasonable request. Conflict of interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Author Andrew Udy declares a conflict of interest as an Associate Editor of Critical Care and Resuscitation (AU). If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.