Methods
Setting and Study Design. This was an observational cohort study to assess the effect of the self-proning intervention on the independent COVID-19 positive patient. It was conducted at Yale New Haven Health Bridgeport Hospital, a 425-bed Level II Trauma center, and approved by the Bridgeport Hospital Institutional Review Board (IRB). Patients admitted to various locations throughout the hospital were considered for this study, including medical and medical-surgical units, the intermediate or Progressive Care Unit (PCU), Intensive Care Units (ICUs), and the ED. Education involved in-servicing 366 hospital staff members to include nurse leadership, nurse management, registered nurses (RNs), physicians, patient care technicians (PCTs), and business associates (BAs). They were informed on the purpose of self-proning and its effects on the lungs in the independent COVID-19 positive patient who did not require assistance with the intervention, how to qualify and educate the patient, and the data collection process.
Patient Population . The study population included 93 out of 98 recruited COVID-19 positive patients admitted to the hospital between April 2020 and May 2021. The population was further subdivided into nonintubated and intubated patients for data collection and comparison purposes. Self-proning data was collected on the intubated patients prior to undergoing mechanical ventilation. The study participants utilized a nasal cannula (N/C) or non-rebreather (NRB) for supplemental oxygenation and were instructed to perform a daily intervention to self-prone over a 4-hour interval. Select COVID-19 positive patients on high flow oxygen (HFO) were also included in the study in a discretionary manner if they met inclusion criteria. Patients in the recruited population who reported an alternative preferred body position and did not engage in proning were excluded from the study.
Materials. In-services on the self-proning protocol were performed for the staff. They were provided with documentation and ready-made packets for use on each patient who met the inclusion criteria. Each packet had a preformed ”Self-Proning” label on it with a fill-in space for patient initials and contained the following (Figure 1A-1D): 1) an algorithm with a diagram decision tool for self-proning, 2) a protocol with methods and illustrations for the self-proning procedure designed as a handout for the patient written in English and Spanish, 3) a handout illustration comparing differences in lung secretion placement in the supine and prone positions in the COVID-19 patient in an easy-to-view format, and 4) a documentation log with columns for date, start and stop times, body position, total time, and comments to be taped to the outside of the patient’s room and filled in by RNs and PCTs per body repositioning event.
Figure 1A. Contents of the Self-Proning Packet. Algorithm decision tool for self-proning.
Figure 1B. Contents of the Self-Proning Packet. Patient protocol for self-proning (English and Spanish versions were made available).
Figure 1C. Contents of the Self-Proning Packet. Illustration of the lungs comparing differences in secretion placement in the supine and prone positions in the COVID-19 patient.
Figure 1D. Contents of the Self-Proning Packet. Documentation data log.
Inclusion and exclusion criteria: Recruitment for qualified subjects was based on inclusion and exclusion criteria and determined by the medical and research staff. We included patients who met the following criteria: COVID-19 positive and admitted to the hospital, 18 years of age and older, male or female, utilizing a N/C or NRB, ability to self-reposition, mentation intact, unobstructed airway. We excluded patients based on the following criteria: COVID-19 negative, < 18 years old, not utilizing an N/C or NRB, not admitted to the hospital, immediate need for intubation, altered mental status (AMS) or agitation, trauma (spinal, thoracic, facial, or surgical), abdominal surgery, inability to supine immediately, pregnancy, pressure wounds or ulcers.
Data Collection. Vital signs were obtained by RNs or PCTs from a telemonitor, pulse oximeter, or a vitals machine using standard equipment and entered on a preformed documentation log kept outside of the patient’s room. The information included the date, vital signs pre-proning, start time for any of four body positions (prone, right side recumbent, sitting up in the Fowler’s position, left side recumbent), a second set of vital signs taken no sooner than 10 minutes after each position change, stop time for the body position, total time in the position, and a signature and comments column to denote variances in the body positioning process, oxygenation, disposition, and patient preferences (Figure 1D). The Epic electronic medical record system was used to collect and confirm patient chart data, including demographics, supplemental oxygenation, vital signs, medical history, and disposition.
Outcomes. The primary outcome in this study for the COVID-19 positive patient on supplemental oxygen was the change in SpO2 at least 10 minutes after self-proning. The secondary outcome was the intubation rate for patients who practiced self-proning.
Analysis. We determined the change in SpO2 (including positive and negative changes) before and during proning events for patients who participated in the intervention. The data were analyzed utilizing descriptive statistics in EXCEL to determine measures of central tendency, frequency, variability, and quartile ranks. Further analysis of the data was done utilizing inferential statistics to determine confidence intervals, the coefficient of determination (R2 ) through linear regression, a t -test paired two sample for means (assuming unequal variances), and calculated probability (p -value). The independent variable (IV) was the prone position (stomach or side-lying), and the dependent variable (DV) was the oxygen status (SpO2) while in the prone position.
Results
The convenience sample in this observational cohort study included 93 out of 98 recruited patients who were COVID-19 positive and admitted to the hospital in medical or intensive care units. In the nonintubated population (n =78), the median patient age (years) was 59 (interquartile range [IQR] 15, 29-84) and for the intubated population (n =15) it was 65 (IQR 18, 38-84). Length of stay (LOS) in the hospital was 13 days for the nonintubated population with a 1% mortality rate and 22 days for the intubated population with a 73% mortality rate (Table 1). Both populations engaged in stomach, lateral, or a combination of both types of proning. The number of proning events, median time for the event, and SpO2 values were recorded on the documentation log (Figure 2).
Table 1. Sociodemographic and Baseline Clinical Characteristics
Figure 2. Schematic Flowchart. Nonintubated and intubated (post proning) patient populations engaged in self-proning. Abbreviations: SpO2 = oxygen saturation.
The means for all documented stomach and lateral proning events for the nonintubated population are reflected in Figure 3. The mean of the starting SpO2 (pre-proning) for the 48 patients who were engaged in stomach proning was 93% (SD 3.3, IQR 4, 95% confidence interval [CI] 92 – 94, 18). The mean for the change in SpO2 during proning was 95% (SD 2.4, IQR 3, 95% CI 95 – 96, 14). Skewness data for the SpO2 was negative before (-0.5) and during proning (-0.9). A comparison of the before and during proning SpO2 values yielded t (170) = -5.22, p <0.001. The mean for changes in SpO2 before lateral proning in this population was 92% (SD 4.3, IQR 4, 95% CI 92 – 93, 25) and during proning 93% (SD 3.6, IQR 4, 95% CI 93 – 94, 21). Skewness was negative (-1.4) before and during proning (-1.8). Comparison of the changes in SpO2 values before and during proning resulted in t (205) = -1.92, p 0.055 (Table 2).
Figure 3. Nonintubated Patient Population (stomach and lateral). Changes in SpO2 are reflected in each graph before and during the proning event.Upper graph : Stomach proning (93 recorded SpO2 values) mean before proning 93% and during proning 95%. Lower graph : Lateral semi-proning (107 recorded SpO2 values) mean before proning 92% and during proning 93%.
Table 2. Self-Proning Patient Statistical Data
For the intubated population (Figure 4), the mean of the pre-proning SpO2 values for stomach proning (n =15) was 90% (SD 5.0, IQR 7, 95% CI 87 – 93, 17). The mean for the change in SpO2 during proning was 95% (SD 3.3, IQR 4, 95% CI 92 – 97, 11). Skewness data for the SpO2 was negative before (-0.8) and during proning (-0.1). A comparison of the before and during proning SpO2 values yielded t (19) = -2.62, p <0.017. The mean for changes in SpO2 before lateral proning in this population was 92% (SD 4.0, IQR 5, 95% CI 90 – 93, 17) and during proning 95% (SD 2.8, IQR 5, 95% CI 94 – 96, 9). Skewness was positive (0.004) before and during proning (0.1) andt (39) = -3.32, p 0.002 (Table 2).
Figure 4. Intubated Patient Population (stomach and lateral). Changes in SpO2 are reflected in each graph before and during the proning event.Upper graph : Stomach proning (12 recorded SpO2 values) mean before proning 90% and during proning 95%. Lower graph : Lateral semi-proning (23 recorded SpO2 values) mean before proning 92% and during proning 95%.
R2 values (Table 2) representing the percentage of variance for the changes in SpO2 (DV) explained by the prone position (IV) are reflected in linear regression models for the nonintubated (stomach proning R2 <0.001, lateral proning R2 <0.001) and intubated populations (stomach proning R2 0.028, lateral proning R2 0.011) (Figure 5).
Figure 5. Linear Regression Models and R2 . Illustrations of the relationship between proning and the changes in SpO2 for the patient populations. Upper and lower left : Nonintubated population - stomach R2<0.001, lateral R2 <0.001.Upper and lower right : Intubated population - stomachR2 0.028, lateral R20.011.