Hospital-Level Availability of Prone Positioning in Massachusetts ICUs

To the Editor:

Prone positioning (PP), a cost-effective therapy (1) with a mortality benefit in moderate–severe acute respiratory distress syndrome (ARDS) (2), is strongly recommended in guidelines for severe ARDS (3) but has been poorly adopted (4–6). Although clinician-level barriers to implementation of PP have been explored (4), institutional barriers may supersede physicians’ beliefs regarding the effectiveness of PP. We sought to evaluate the institutional availability of PP.

Methods

We conducted a survey of all ICUs at acute-care hospitals in Massachusetts, April–July 2019. We e-mailed surveys to ICU nurse and physician leadership; if the surveys remained incomplete after four follow-ups, we completed the survey by phone. For hospitals that provided more than one response, one survey was randomly selected from hospitals with duplicate responses (n = 6) or the survey with the more complete responses (n = 3) was selected. The survey asked, “Does your ICU have the ability to prone patients”? (“yes,” “no,” or “case-by-case”). Hospitals that responded “yes” were termed “prone-ready.” Follow-up questions inquired about institutional protocols/guidelines describing indications/instructions for PP, nurse training in PP, and reasons for not using PP. Respondents were also presented with a list of adjunctive treatments for ARDS and asked to select treatments used at their institution. We collected descriptive data for each hospital (number of ICU and total hospital beds, profit status, teaching status, case mix index (7), and Centers for Medicare and Medicaid Services star rating [8]) using publicly available information (7–9).

Using the chi-square test and ANOVA, we compared hospital characteristics on the basis of the hospitals’ ability to perform PP. Statistical testing was two-tailed, with α = 0.05 using SAS 9.4 (SAS Institute). The Beth Israel Deaconess Medical Center Institutional Review Board deemed the study exempt from review.

Results

Among 60 acute-care hospitals in Massachusetts with ICUs, 54 responded to the survey (90% response rate; six nonresponders were nonacademic hospitals, five of which had <250 total beds). Twenty-four respondents (44.4%) were “prone-ready”; 15 (27.8%) could provide PP on a case-by-case basis, and 15 (27.8%) could not provide PP. Prone-ready hospitals accounted for 358 ICU beds out of a total of 600 ICU beds in the state (59.7%); case-by-case hospitals and PP-unavailable hospitals accounted for 71 (11.8%) and 122 ICU beds (20.3%), respectively. Prone-ready hospitals were more likely to be larger teaching hospitals with a more severe case mix index (Table 1). Twenty-seven hospitals (37.0%) had a PP protocol/guideline [20 (83.3%) prone-ready, 6 (40.0%) case-by-case, and 1 (6.7%) PP-unavailable]. Thirty-four hospitals (63.0%) reported that some or all nurses had received training in PP [24 (100%) prone-ready, 9 (60.0%) case-by-case, and 1 (6.7%) PP-unavailable].

Table 1.

Characteristics Associated with Hospitals That Are Prone-Ready

Hospital Characteristics	Prone-Ready (n = 24)	Prone Positioning Available on a Case-by-Case Basis (n = 15)	Prone Positioning Unavailable (n = 15)	P Value
Teaching hospital, n (%)	9 (37.5)	4 (26.7)	0 (0)	0.03
Total medical or mixed medical/surgical ICU beds, mean (SD)	14.9 (9.9)	8.1 (5.1)	4.7 (2.3)	0.002
Hospital size, n (%)				0.02
Small (<250)	11 (45.8)	11 (73.3)	13 (86.7)
Medium (250–424)	6 (25.0)	4 (26.7)	2 (13.3)
Large (≥425)	7 (29.2)	0 (0.0)	0 (0.0)
Nonprofit status, n (%)	22 (91.7)	11 (73.3)	11 (73.3)	0.23
High public payer hospital, n (%)	10 (41.7)	11 (73.3)	5 (33.3)	0.39
Case mix index*, mean (SD)	1.1 (0.29)	0.9 (0.1)	0.9 (0.2)	0.01
Centers for Medicare and Medicaid Services star rating, mean (SD)	3.0 (1.3)	3.1 (1.0)	2.8 (1.1)	0.71

Prone-ready hospitals are more likely to be teaching hospitals, have more medical or mixed medical/surgical ICU beds, have more total hospital beds, and have patients with a higher case mix index.

^*3M All Patient Refined grouper, version 30, reported through the Massachusetts Center for Health Information and Analysis.

Twenty-three respondents (42.6%) indicated they did not use PP in the past year, accounting for 119 (19.8%) of ICU beds in the state. Common reasons for not using PP were physician or nurse discomfort (n = 14), lack of nurse training (n = 13), and/or lack of proper equipment (n = 9; in seven free-text comments, lack of “a rotating bed” was noted).

Forty-two respondents (77.8%) indicated they had transferred patients with ARDS to another facility in the past year; however, hospital transfer was not associated with PP availability (43% of transferring hospitals were prone-ready; P = 0.49). Use of adjunctive treatments for ARDS before transfer was common, regardless of PP availability (Figure 1).

Figure 1.

Treatments initiated for acute respiratory distress syndrome (ARDS) before referral to another hospital. The 42 hospitals that reported transferring patients with ARDS to other hospitals were asked to select treatments that were provided before transfer. The numbers of hospitals that selected each option are graphed, stratified by self-reported ability to prone patients. The number of hospitals that were unable to prone but nevertheless were able to offer a treatment before transfer is shown. PEEP = positive end-expiratory pressure.

Discussion

Although guidelines recommend PP for patients with severe ARDS, prior studies have shown underuse of PP. We found that most hospitals in Massachusetts were either unable, or not completely able, to routinely offer PP. Reasons identified by ICU leadership for hospital-level lack of PP availability included multiple modifiable factors, such as lack of training and misconceptions about equipment requirements. Our results suggest that institutional-level barriers to implementation of PP are a promising initial target to improve implementation.

Our findings also provide context to prior studies of PP adoption that focused on rates in ICUs participating in trial networks (3, 5) and/or within teaching centers (6). For example, Duan and colleagues evaluated only ICUs that offered PP, and reported that 10% of the appropriate patients received PP (6). Our finding of low real-world availability of PP suggests that the prevalence of PP use among eligible patients is likely even lower than previous estimates.

We found that other adjunctive treatments for ARDS were often used at centers where PP was unavailable. Notably, some adjunctive interventions are based on weaker evidence, are more expensive, or require levels of monitoring similar to those required for PP, suggesting that resource limitations or staffing constraints alone do not explain the lack of PP adoption at the institutional level. Some respondents reported a common misconception that “lack of equipment” was a barrier to instituting PP. However, PP does not require specialized equipment beyond basic cushioning to support the face, chest, and pelvis (10, 11).

Our study has limitations. First, we surveyed hospitals in one state, and Massachusetts is notable for having many geographically close tertiary-care and/or teaching hospitals, making generalizability to other states unclear. However, teaching hospitals were more likely to be prone-ready; therefore, our findings may overestimate the proportion of prone-ready hospitals in states with fewer teaching hospitals. Second, the surveys may have been affected by social desirability bias and nonresponse bias. In this instance, where the “socially desirable” response would be to claim an ability to perform PP, our findings would overestimate the number of prone-ready hospitals. Third, we were unable to quantify the number of eligible patients affected by a hospital’s inability to perform PP. However, more than 20% of the state’s ICU beds were at prone-unavailable hospitals, and it is probable that some patients did not receive PP when they were eligible. Future studies should investigate the real-world prevalence of PP use among eligible patients. Fourth, it is possible that discomfort with PP is justifiable. The safety and effectiveness of PP have been demonstrated in studies of large tertiary-care hospitals, but the risks of PP in settings with low ARDS case volumes are unclear and should be investigated. If safety of PP at centers with low ARDS case volumes cannot be demonstrated, strategies to manage patients with severe ARDS at such centers will need to be established. Fifth, although we reasoned that restrictions on PP availability (e.g., due to overnight staffing constraints) reflected institutional limitations on the use of PP, it is possible that hospitals that reported PP on a case-by-case basis could be classified as prone-ready. However, hospitals that were prone-ready were more than twice as likely as case-by-case hospitals to have a PP protocol and to have performed PP. Categorizing hospitals that reported PP on a case-by-case basis as prone-ready would increase the proportion of institutions with available PP to 72%, well below institutional rates documented in prior PP-implementation studies. Our finding that many institutions are unable to offer PP raises multiple questions for future investigation, including a deeper exploration of provider-level perceived barriers to PP and the interactions between institutional-level and provider-level barriers.

Although prior work has cited insufficient clinical recognition of patient eligibility for PP as a major barrier to PP uptake (5), barriers at the clinician level can only be overcome after structural barriers have been addressed. Our finding of low uptake of an evidence-based intervention with a mortality benefit at an institutional level (sometimes due to misconceptions) raises multiple questions for future investigation, and suggests that attempts to implement PP among eligible patients will need to include consideration of hospital-level barriers.