Keywords
prone positioning, SARS-CoV-2, acute respiratory distress syndrome, mechanical ventilation, adverse events, quality improvement
This article is included in the Coronavirus (COVID-19) collection.
prone positioning, SARS-CoV-2, acute respiratory distress syndrome, mechanical ventilation, adverse events, quality improvement
Since its emergence in 2019, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) responsible for coronavirus disease 2019 (COVID-19) has placed unprecedented pressure on healthcare systems across the globe. Intensive care units (ICUs) have been particularly challenged, with approximately 5% of all confirmed cases and up to 20% of hospitalised cases requiring ICU admission1.
In patients with SARS-CoV-2, prone ventilation improves lung perfusion, better recruits atelectatic alveoli and reduces dorsal lung compression2. Proning is long recognised therefore as an adjunctive respiratory treatment that can improve oxygenation and significantly reduce mortality3–6. Although SARS-CoV-2 is novel, recent data suggest that the beneficial effects of proning may persist in mechanically ventilated patients with SARS-CoV-2 infection, as well as in awake patients who can intermittently self-prone5,7. Although the studies on mortality benefit are conflicting, prone ventilation is now widely recommended and implemented in adult ICU patients with severe SARS-CoV-28,9.
Despite its effectiveness, proning is associated with several potential complications including accidental extubation, brachial plexus injury, pressure ulcer formation, ocular injury and central venous catheter-related infections10–13. As the pandemic enters its third year, clinicians are now becoming familiar with the potential risks associated with prone positioning, as well as the strategies that may mitigate them e.g., of a proning team14. That said, rarer and/or previously unknown complications of prone positioning continue to come to light given its widespread use across the world15.
The aim of our study was to describe the range of complications encountered during prone ventilation of adult patients with severe SARS-CoV-2, and to identify associated risk factors for these complications.
This was a single centre retrospective observational study carried out in the intensive care unit (ICU) in Tallaght University Hospital (TUH), an academic tertiary referral hospital, in Dublin using records from patients admitted to the ICU between March and June 2020. The sampling frame was all adult patients admitted to the ICU with laboratory-confirmed infection with SARS-CoV-2 who were treated with invasive mechanical ventilation and prone positioning on at least one occasion. Indications for prone positioning in hypoxemic respiratory failure associated with SARS-CoV-2 infection included a PaO2/FiO2 (partial pressure of arterial oxygen/fraction of inspired oxygen or P/F) ratio < 150, an FiO2 > 0.6 and a positive end-expiratory pressure (PEEP) > 14 cm H2O. Patients were excluded from the study if they received non-invasive ventilation only or were not treated in the prone position. Absolute and relative contra-indications to proning used in our ICU are outlined in Table 1.
Proning was carried out by a seven-person, intensivist-led proning team according to a documented protocol and safety checklist (Table 1 and Table 2). Individuals received appropriate training, including video presentations and doctor-led simulation sessions, prior to joining the proning team. Patients were treated with prone ventilation for a planned duration of 16 hours per day (16.00-10.00), after which they were returned to supine position by the proning team. Problems and/or complications related to the proning manoeuvre were documented in real-time in the patient’s healthcare record as per standard clinical practice in our ICU.
Data collection occurred retrospectively using two sources: the intensive care record used by ICU staff and the accompanying hand-written healthcare record used by non-ICU staff. For most patients, the intensive care record was in an entirely electronic format (IntelliSpace Critical Care and Anaesthesia); for a minority, the intensive care record was also hand-written. We collected the following patient data: age, sex (based on self-report), ethnicity, body mass index, co-morbidities, duration of mechanical ventilation, number of proning sessions performed, ICU length of stay and mortality. The Acute Physiology and Chronic Health Evaluation (APACHE) II and Sequential Organ Failure Assessment (SOFA) scores were also recorded.
Complications were identified through clinical review of the patient’s contemporaneous healthcare records, beginning from time of first proning until discharge from ICU (alive or dead). Once identified, complications were categorised as immediate (i.e., occurring at time of proning) or delayed (arising after a single or repeated proning manoeuvre).
We recorded the number of hours for which a patient was prone on a daily basis and the total number of proning sessions provided for each patient. We measured the total number of complications associated with prone positioning and then subcategorised these according to the organ system affected. For skin/mucosal surface ulcers, we noted the location of each lesion and adjudicated whether it was likely to have resulted from prone positioning. Most ulcers did fit this pattern, such as ulcers on the lips, chin, bridge of the nose, ear, neck, chest wall, and anterior abdominal wall. Each neuromuscular complication was fully assessed clinically, with supportive neurophysiological investigations if required, to adjudicate whether it was more likely to have occurred as a result of ICU-acquired weakness. We therefore included neuromuscular complications that fit the pattern for injuries associated with prone positioning, including paraesthesia, isolated foot drop, pain, numbness, rotator cuff injury, deep peroneal nerve injury and shoulder subluxation.
Ethical approval to carry out this study was granted by the SJH/TUH Joint Research Ethics Committee on April 2nd 2020 (SJH/TUH JREC Reference Number 2020-04 Chairman’s Action (01)) and data collection began immediately after this. Written patient consent was waived by the research ethics committee given that this was a fully anonymised retrospective chart review that did not require any active patient participation or intervention.
Demographic and clinical characteristics are reported as counts and percentages or means and standard deviations (SD). Medians and interquartile ranges (IQR) were utilised for non-parametric data. We used Chi-squared tests to assess for significant differences with regard to categorical variables. The Mann-Whitney U test was used to assess for significant differences with regard to non-parametric continuous variables. A Spearman’s correlation coefficient was used to assess for a significant association between the total number of proning complications and patient demographics including body mass index (BMI), ICU illness severity scores including APACHE-II score and SOFA score, proning outcomes including the total number of proning sessions and patient outcomes including ICU length of stay. Our primary outcome was to assess potential risk factors for a significant association with the total number of proning complications experienced by a patient. Our primary analysis therefore was a multivariable Poisson regression model used to evaluate whether predictor variables were independently associated with a significantly increased total number of complications related to proning. Continuous variables were considered for the model if they resulted in a Spearman correlation coefficient > 0.5 with respect to the total number of proning complications. Categorical variables were considered for the model if Mann-Whitney U testing indicated a statistically significant difference in median number of total proning complications between the groups. The number of hospital days before ICU admission, BMI, race and diabetes mellitus were all considered as predictor variables. Age, APACHE II score and SOFA score were added to the model based on prior clinical knowledge. Relative parsimony of models was determined using the Akaike’s Information Criterion (AIC). The dispersion parameter was taken to be 1.
As secondary exploratory outcomes, we assessed for significant correlations between the number of proning sessions, the number of ulcers and the number of neurological complications, respectively, and patient demographics, ICU illness severity scores, proning and patient outcomes using a Spearman’s correlation coefficient for continuous variables and Mann Whitney U testing for categorical variables, respectively. Statistical analysis was carried out using R Project for Statistical Computing (RRID:SCR_001905) software Version 4.1.1.
A total of 17 patients were eligible for inclusion in our study following admission to the ICU with acute hypoxemic respiratory failure associated with laboratory-confirmed SARS-CoV-2 infection, requiring invasive mechanical ventilation and prone positioning on at least one occasion in our ICU, between March and June 2020 (Table 3). A total of 15 other patients were excluded from the study during the same time period as they did not require proning. Therefore, our rate of proning was 53.1% (17/32 patients) for acute hypoxemic respiratory failure associated with laboratory-confirmed SARS-CoV-2 infection. The majority of the patients who were proned were male with a median age of 51 years. The most common comorbidities were hypertension, diabetes and cardiovascular disease. The median time from hospital to ICU admission was short at one day with a relatively low median APACHE II illness severity score of 11.
The median number of proning sessions per patient was four with a median time of 17 hours (Table 4). The most common complications noted were skin ulcers in 15/17 (88.2%) patients and neurological complications including pain, paraesthesia, numbness, foot drop and limb weakness in 12/17 (70.6%) patients. Nasogastric tube blockage or displacement was also commonly noted in 6/17 (35.3%) cases.
Proning outcome | Summary statistics |
---|---|
Number of proning sessions, median (IQR) | 4 (3-6) |
Duration in hours of each proning session, median (IQR) | 17 (16-18) |
Total number of complications per patient, median (IQR) | 5 (3-7) |
Number of patients with ulcers, n/total (%)* | 15/17 (88.2%) |
Total number of ulcers per patient, median (IQR) | 3 (1-4) |
Number of patients with neurological complications, n/total (%) | 12/17 (70.6%) |
Total number of neurological complications per patient, n/total (%)** | 2 (1-4) |
Nasogastric tube blockage or displacement, n/total (%) | 6/17 (35.3%) |
ICU mortality was 17.7% (3/17 patients) with a further 17.7% (3/17 patients) dying in hospital post ICU discharge (Table 5). Patients had a relatively long ICU and hospital length of stay with a median duration of ventilation of 18 days. All included patients required vasopressors during their ICU admission.
There was a strong correlation between both the number of neurological complications (r = 0.76; p = 0.00; Table 6 and Table 7) and the number of ulcers (r = 0.7; p = 0.001) with the total number of proning complications. There was also a strong correlation between both hospital length of stay (r = 0.63; p = 0.001) and ICU length of stay (r = 0.64; p = 0.002) and the total number of proning complications. There was a moderate correlation between the total number of proning complications and the number of proning sessions (r = 0.4; p = 0.02). There was no significant correlation between total number of proning complications and age, BMI, frailty score, APACHE score and SOFA score.
Baseline patient characteristic | Total number of proning complications, median (IQR) | P-value |
---|---|---|
Sex | ||
Male | 6 (3-8) | 0.4 |
Female | 4 (2.8-5.3) | |
Race* | ||
Asian | 8 (7-11) | 0.04* |
Black | 6 (4-7) | |
White | 3 (2-5) | |
Comorbidities | ||
Diabetes mellitus | 11 (10-12) | 0.01 |
No diabetes mellitus | 5 (2-6) | |
Hypertension | 6 (4-8) | 0.72 |
No hypertension | 5 (3-7) | |
Cardiovascular disease | 3 (2-5) | 0.38 |
No cardiovascular disease | 6 (4-8) | |
Baseline patient characteristic | No. of proning sessions required, median (IQR) | P-value |
Sex | ||
Male | 4 (4-7) | 0.4 |
Female | 4 (3-5) | |
Race | ||
Asian | 4 (4-13) | 0.4 |
Black | 7 (5-9) | |
White | 4 (3-4) | |
Comorbidities | ||
Diabetes mellitus | 4 (4-13) | 0.4 |
No diabetes mellitus | 4 (4-6) | |
Hypertension | 4 (2-4) | 0.35 |
No hypertension | 4 (4-7) | |
Cardiovascular disease | 4 (3-9) | 1.00 |
No cardiovascular disease | 4 (3-6) | |
Baseline patient characteristic | No. of ulcers, median, IQR | P-value |
Sex | ||
Male | 3 (2-5) | 0.32 |
Female | 2 (1-3) | |
Race* | ||
Asian | 5 (3-5) | 0.13 |
Black | 2 (2-3) | |
White | 2 (1-3) | |
Comorbidities | ||
Diabetes mellitus | 5 (4-6) | 0.06 |
No diabetes mellitus | 2 (1-3) | |
Hypertension | 3 (3-3) | 0.36 |
No hypertension | 2 (1-5) | |
Cardiovascular disease | 4 (4-5) | 0.33 |
No cardiovascular disease | 3 (1-4) | |
Baseline patient characteristic | No. of neurological complications, median (IQR) | P-value |
Sex | ||
Male | 3 (0-4) | 0.67 |
Female | 2 (2-2) | |
Race* | ||
Asian | 3 (2-7) | 0.24 |
Black | 4 (2-4) | |
White | 2 (0-3) | |
Comorbidities | ||
Diabetes mellitus | 7 (5-8) | 0.09 |
No diabetes mellitus | 2 (0-4) | |
Hypertension | 3 (2-4) | 0.56 |
No hypertension | 2 (1-4) | |
Cardiovascular disease | 1 (1-2) | 0.29 |
No cardiovascular disease | 3 (1-4) |
Age, the number of hospital days before ICU admission, BMI, race, diabetes, APACHE II score and SOFA score were all considered as variables in a multivariable Poisson regression model for prediction of total number of proning complications (Table 8). Our final model included the variables that best achieved a balance between model fit and model complexity using the AIC, which in this case resulted in diabetes as a univariate predictor for total number of proning complications; indicating that the presence of diabetes increased the total number of proning complications per patient by about 0.9. This model achieved the lowest AIC at 87.4 and the highest adjusted coefficient of determination at 0.45.
There was a strong positive correlation between the number of proning sessions and both ICU length of stay (r = 0.74; p = 0.0003) and hospital length of stay (r = 0.52; p = 0.006). There was also a strong positive correlation between the number of proning sessions and the number of ventilator days (r = 0.77; p = 0.00). There was a moderate positive correlation between the number of proning sessions and total number of proning complications (r = 0.4; p = 0.02). There was no relationship between the number of proning sessions and age, frailty score, APACHE II score, SOFA score, number of ulcers and number of neurological complications during the ICU stay.
There was a strong positive correlation between the SOFA score and the number of ulcers (r = 0.75; p = 0.005; Table 9). There was also a strong positive correlation between the total number of proning complications and the number of ulcers (r = 0.7; p = 0.001). There was additionally a strong positive correlation between frailty score (r = 0.59; p = 0.02), hospital length of stay (r = 0.67; p = 0.005), ICU length of stay (r = 0.63; p = 0.01), ventilator days (r = 0.68; p = 0.002) and the number of ulcers. There was no significant correlation between the number of ulcers and age, BMI, time from hospital to ICU admission, APACHE II score, the number of proning sessions or neurological complications.
There was a strong negative correlation between both time from hospital to ICU admission (r = -0.6; p = 0.02; Table 10) and APACHE II score (r = -0.69; p = 0.008) with the number of neurological complications. There was a strong positive correlation between the total number of proning complications and the number of neurological complications (r = 0.76; p = 0.00). There was no significant correlation between the number of neurological complications and age, BMI, frailty score, hospital/ICU length of stay, the number of proning sessions, the number of ulcers or ventilator days.
In this single-centre retrospective observational study of complications associated with prone positioning during mechanical ventilation for acute hypoxemic respiratory failure due to infection with SARS-CoV-2, we found that 88.2% (15/17) patients suffered complications associated with prone positioning during their ICU stay. The most common complications noted were skin ulcers in 15/17 (88.2%) patients and neurological complications including pain, paraesthesia, numbness, foot drop and limb weakness in 12/17 (70.6%) patients. Nasogastric tube blockage or displacement was also commonly noted in 6/17 (35.3%) cases. In a multivariable Poisson regression model used to evaluate predictor variables which were independently associated with a significantly increased total number of complications related to proning, only diabetes mellitus was independently associated with an increased total number of proning complications.
The results of our study are largely in keeping with a growing body of evidence for potential harm associated with prone positioning while mechanically ventilated. Adverse events related to prone positioning have become particularly relevant during the SARS-CoV-2 pandemic given widespread use of this adjunctive treatment for severe acute hypoxemic respiratory failure, often in centres that may not have previously provided training or patient care with prone positioning. In their cross-sectional study of SARS-CoV-2 patients requiring prone positioning while mechanically ventilated, Binda et al., found a high frequency of prone-related pressure ulcers, bleeding and medical device displacement during prone positioning in a population of predominantly middle-aged obese male patients10. The likelihood of pressure-ulcer development was independently associated with the duration of prone positioning. A recent scoping review by Gonsález-Seguel et al., identified > 40 adverse events reported in prone positioning ARDS studies, subdivided according to the occurrence of adverse events during prone positioning vs. during a prone manoeuvre15. These included additional adverse events that have not yet been reported by systematic reviews, e.g., graded pressure ulcers and peripheral nerve injuries16–18. Of note, nearly half of the eligible studies in the scoping review did not report adverse events, which may indicate lack of recognition of this important aspect of proning management.
Similarly, the American Thoracic Society/European Society of Intensive Care Medicine/Society of Critical Care Medicine guidelines have recommended against implementing prone positioning for mechanically ventilated patients with P/F ratios > 100 because the authors had “lower confidence in the balance between desirable compared with undesirable outcomes”19. The studies that support these guidelines were, however, published prior to the SARS-CoV-2 pandemic and may not reflect the more widespread use of prone positioning now; including centres that may not have utilised this previously. Nevertheless, these meta-analyses did concur with our findings that the most common complications associated with prone positioning during mechanical ventilation were pressure ulcers and medical device blockage/displacement; notably transient endotracheal tube obstruction4,20,21 There was, however, no increased risk of unplanned extubation, pneumothorax or unplanned central catheter removal associated with prone ventilation in these meta-analyses.
Our study confirms this growing body of international evidence that prone positioning during mechanical ventilation for SARS-CoV-2 infection is associated with increased risk of complications that are not insubstantial in nature and that occur more frequently than previously reported in trials such as the PROSEVA trial that took place prior to the onset of the pandemic3. However, as outlined by Concha et al., in their recent scientific letter, many of these complications, particularly skin ulcerations and nasogastric tube obstruction, can be self-limiting in nature and resolve once the patient improves sufficiently to return to supine ventilation22. We also found a significant association between the total number of proning sessions and the total number of proning complications, the number of ulcers and the number of neurological complications experienced by patients while prone; which indicates that prolonged use of prone ventilation may be harmful in this regard. However, this increased incidence may be biased by prolonged survival in patients ventilated in the prone position; exposing them to the associated risk of complications for a longer period of time. Indeed, the mortality in our study was relatively low at 17.7% with a relatively high median number of ventilator days of 18; indicating that patients with this survival advantage are likely to be ventilated for longer with a concomitant longer exposure to prone positioning over this time period.
There are various ways in which the risks associated with prone positioning can be mitigated in the era of SARS-CoV-2. Use of the “swimming position” to help prevent risk of brachial plexus injury while in the prone position has been recommended in multiple recent studies11,23. Use of the reverse Trendelenburg position has also been reported as a strategy to reduce facial and eye oedema/injuries as well as ventilator-associated pneumonia, vomiting and severe oxygen desaturation, particularly when combined with alternating face rotation and repositioning every two hours24. Adequate staff education and training is essential for appropriate up-skilling, particularly for staff who have been re-deployed, to ensure that this treatment can be provided safely for those who need it25. Support with protective coverings for pressure areas is key, along with endotracheal tube monitoring when turning patients (as outlined in our institutional proning protocol in Table 1 and Table 2)26–28. Use of a pre-manoeuvre safety checklist, including the right equipment, monitoring, number of staff and a team leader, can also help to mitigate these risks as well23. For prevention of pressure ulcers specifically, McEvoy et al., highlight the effectiveness of a multi-faceted care bundle, which reduced incidence in their single centre intervention study by 25%29. As outlined by Gattinoni et al., depth and duration of sedation should be guided by standardised sedation algorithms and not by body position, in order to reduce further risk of complications associated with proning30.
The strength of our study includes a full comprehensive description of the baseline characteristics, proning complications and ICU outcomes for patients requiring prone positioning during mechanical ventilation for severe hypoxemic respiratory failure during the first wave of the SARS-CoV-2 pandemic, a time during which prone positioning was much more widely adopted as a therapeutic manoeuvre than previously. We have described the burden of complications associated with prone ventilation and risk factors associated with an increased frequency of complications in this regard. This work draws attention to an increasingly recognised quality and safety aspect of proning care that can lead to considerable future morbidity and healthcare resource utilisation if not anticipated and carefully managed when it does occur.
Limitations to our study include the fact that this is a single centre study that was retrospective in nature with a relatively small sample size. This relatively small sample size did however include all patients who were admitted to our ICU with SARS-CoV-2 pneumonitis requiring mechanical ventilation and proning during the first wave of the pandemic so we fully captured all of this cohort with no loss to follow-up, but may have limited our power to detect significant associations between patient risk factors and the number of proning complications experienced. Given our small sample size, we were not able to disaggregate our primary outcome according to sex, nor perform a detailed analysis of the effect of sex on our results. We would like to be able to address this limitation in the future with a larger sample size. We also do not have long-term data about the duration of proning complications and any associated morbidity or healthcare costs. We acknowledge that proning management and familiarity with institutional proning protocols may have improved between the first wave of the SARS-CoV-2 pandemic and later waves but believe that the depiction of complications associated with prone ventilation in this paper is valid in describing an increasingly important problem associated with this potentially life-saving treatment. This was also a time during which prone positioning during mechanical ventilation was used far more frequently than previously so the learning curve was high and the potential for adverse events was marked, as we have shown. It is important to continue to learn lessons from this to standardise prone positioning and to increase the evidence base for its quality and safety in future patient management.
In conclusion, in this single centre retrospective observational study of complications associated with prone positioning during mechanical ventilation for acute hypoxemic respiratory failure due to infection with SARS-CoV-2, we found that 88.2% (15/17) patients suffered complications associated with prone positioning during their ICU stay. The most common complications noted were skin ulcers, neurological complications and nasogastric tube blockage or displacement. In a multivariable Poisson regression model, only diabetes mellitus was independently associated with an increased total number of proning complications. Adequate staff education and training is essential to ensure that this treatment can be provided safely for those who need it.
Our study data is not publicly available in a data repository due to restrictions required by our Institutional Review Board given the presence of potential patient identifiers in a single centre study. However, full consideration will be given to any requests made to the corresponding author for data sharing of a fully anonymised dataset for the purpose of validation or extended research in this regard.
Is the work clearly and accurately presented and does it cite the current literature?
Yes
Is the study design appropriate and is the work technically sound?
Partly
Are sufficient details of methods and analysis provided to allow replication by others?
Yes
If applicable, is the statistical analysis and its interpretation appropriate?
I cannot comment. A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?
No source data required
Are the conclusions drawn adequately supported by the results?
Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: critical care medecin
Alongside their report, reviewers assign a status to the article:
Invited Reviewers | |
---|---|
1 | |
Version 1 08 Aug 23 |
read |
Provide sufficient details of any financial or non-financial competing interests to enable users to assess whether your comments might lead a reasonable person to question your impartiality. Consider the following examples, but note that this is not an exhaustive list:
Sign up for content alerts and receive a weekly or monthly email with all newly published articles
Register with HRB Open Research
Already registered? Sign in
Submission to HRB Open Research is open to all HRB grantholders or people working on a HRB-funded/co-funded grant on or since 1 January 2017. Sign up for information about developments, publishing and publications from HRB Open Research.
We'll keep you updated on any major new updates to HRB Open Research
The email address should be the one you originally registered with F1000.
You registered with F1000 via Google, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Google account password, please click here.
You registered with F1000 via Facebook, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Facebook account password, please click here.
If your email address is registered with us, we will email you instructions to reset your password.
If you think you should have received this email but it has not arrived, please check your spam filters and/or contact for further assistance.
Comments on this article Comments (0)