Pressure Injury Prevention Practices in the Intensive Care Unit: Real-world Data Captured by a Wearable Patient Sensor

Background. Compliance with turning protocols in the intensive care unit (ICU) is low; however, little is known about the quality of turning, such as turn angle magnitude or depressurization time. Wearable sensors are now available that provide insight into care practices. Objective. This secondary descriptive study describes the turning practices of nurses from 2 ICUs at an academic medical center among consecutive ICU patients. Materials and Methods. A wearable patient sensor was applied to patients on hospital admission. The sensor continuously recorded position data but was not visible to staff. A qualified turn was one that reached > 20° angle and was held for 1 minute after turning. The institution's clinical research repository provided clinical data. Results. A total of 555 patients were analyzed over a 5-month period (September 2015–January 2016); 44 870 hours of monitoring data (x– = 73 hours ± 97/patient) and 27 566 individual turns were recorded. Compliant time was recorded as 54%, with 39% of observed turns reaching the minimum angle threshold and 38% of patients remaining in place for > 15 minutes (depressurization). Turn magnitude was similar for medical and surgical patients. Factors associated with lower compliant time included male sex, high body mass index, and low Braden score. Patients were supine for 72% of the observed time. Conclusions. The investigators found dynamically measured turning frequency, turn magnitude, and tissue depressurization time to be suboptimal. This study highlights the need to reinforce best practices related to preventive turning and to consider staff and patient factors when developing individualized turn protocols.

Nurses have historically relied upon the routine repositioning of patients every 2 or 4 hours in acute care settings to prevent hospital-acquired pressure injuries (HAPIs). However, compliance with turning protocols has been shown to be low.^1,2 A 2014 Cochrane review³ found that although patient turning has sound theoretical rationale and is an integral component of pressure injury prevention, there is a lack of high-quality data determining optimal repositioning frequency, despite the publication of more than 17 000 manuscripts since the 1940s; prior studies are both low in volume and quality.³ Notwithstanding, recent clinical guidelines⁴ recommend individualized turning protocols based on a patient’s unique care needs. This includes modifying turning protocols based on the type of supportive surface used and adjusting the frequency and magnitude (angle) of repositioning.

A major limitation to all prior research studies has been the inability to accurately measure care delivery objectively and at scale. The Leaf Patient Monitoring System (Leaf Healthcare, Inc, Pleasanton, CA) is a noninvasive wearable sensor that continuously records and quantifies turning practices, including data regarding turning frequency, duration of depressurization time, and exact turning angle. Using the wearable pressure sensor, clinicians are able to record the continuous turning practices of nurses throughout entire hospital units with an unlimited number of patients.

Herein, the investigators describe the turning practices of nurses from 2 intensive care units (ICUs) at an academic medical center. By understanding the nurses’ behaviors associated with preventive turning practices, health care providers can develop better educational tools, improve clinical practice, and lay the foundation for generating personalized care plans to prevent HAPIs.

This is a secondary descriptive study of turning data obtained from a pragmatic randomized controlled trial (LS-HAPI Study).^5,6 Patients were enrolled from 1 of 2 adult ICUs at a single academic medical center located in Northern California; ICU A is a 25-bed unit specializing in the care of post cardiothoracic surgical patients, and ICU B is a 33-bed unit specializing in the care of critically ill medical, surgical, neurological, and trauma patients.

Adult patients admitted to ICU A or B and allocated to the nonintervention arm of the LS-HAPI study were analyzed. This group closely reflects standard of care and is representative of current nursing practices related to preventive turning in an ICU setting. The study participants generally have severe or life-threatening injury or disease warranting admission to an ICU. The study population encompasses patients with significant surgical diseases and processes (eg, trauma, general surgery, vascular surgery, cardiac surgery, transplant) and/or severe medical diseases and processes (eg, sepsis, acute respiratory distress syndrome, multiple organ failure). Study exclusion criteria consisted of patients < 18 years of age, patients with an issue preventing effective sensor adhesion (ie, a sternal dressing) or known adhesive sensitivity, extreme frailty/acuity as determined by clinicians precluding study participation, or patients exercising their right of refusal.

Upon arrival to a study unit, all patients meeting the study’s inclusion criteria had a wearable sensor applied to their torso with the sensor’s adhesive backing (Figure 1). The unit’s administrative assistant paired the sensor to the individual patient using a unique sensor identification number and a patient identifier that was automatically populated by the institution’s Admission-Discharge-Transfer (ADT) HL-7 data stream. Position-related data were captured every 10 seconds by the sensor and relayed over a secure mesh network to a structured query language database for quantification. Detailed information on the parent study is available.^5,6

Minimum thresholds for turning were established based on best available evidence and expert opinion.^7,8 These thresholds were defined as: turning at a minimum of every 2 hours (per organizational policy); a minimum magnitude of turn of 20°; and at least 15 minutes of tissue depressurization, which was a dynamic target. For example, if a patient stayed on his/her newly turned side for half of the minimum expected depressurization time (eg, 7.5 minutes vs. 15 minutes), then the time-to-next turn was proportionally adjusted (ie, turn time would be reduced by 50%, such that a turn would be required within 1 hour instead of 2). This was performed continuously to achieve the goal of at least 15 minutes of cumulative tissue depressurization time every 2 hours.

The patient monitoring system is designed to promote optimal turning practices by providing nurses with real-time patient positioning data. In an effort to report on standard care practices and to not influence patient care, nurses caring for patients included in this analysis (the nonintervention group [control group]) did not receive feedback data from the wearable patient sensor regarding the turning quality, frequency, or duration of depressurization, nor did they receive any visual advisories from the wireless monitoring system to promote optimal turning practices (as was the case in the study treatment group where the monitoring system’s feedback loop was closed).⁶ Nurses caring for patients in the control group relied upon traditional reminders to coordinate patient turning efforts. This was not standardized across units but may include manual documentation of patient position, use of paper turning clocks posted at the bedside, or nursing judgment.

Analysis

The wearable sensor measures and records the position of a patient every 10 seconds. To minimize signal noise that may occur with frequent patient movement, 3 body regions were defined: left, back, and right sides (Figure 2). For a turn to register and be suitable for inclusion in the analysis, a patient had to have been in a prior body region for at least 1 minute, move to a new body region, and then stay in this new body region for at least 1 minute (ie, move from the left side to the back, or the right side to the left side). Changes in position were then calculated to determine the degree of position change (laterally and vertically) as well as the duration of time spent in each position. Spontaneous patient movements are captured by the sensor and were treated as a patient turn if minimal turning thresholds were met. Sensor position measurements were tagged with the date and time to enable the assessment of specific changes over time.

To compute the overall compliance with preventive care measures, the following formula was used:

A final compliance score was calculated for each patient, resulting in a percentile score representing the overall proportion of time the patient’s care was within recommended care guidelines, ie, repositioning at least every 2 hours.

For analysis, standard descriptive measures of central tendency were used to describe the various components of patient turning. An analysis of variance (ANOVA) was used with Bonferroni correction for post hoc tests for the differences in parametric data across multiple groups, such as differences in degree of turning with age, nursing shift, day of week, and Braden score. Chi-square tests were used for nonparametric data. To enable interpretation, some variables were restructured into stratified groups. For example, the Braden risk score was restructured into 3 groups according to level of risk, and differences between groups were evaluated as low (19–23), medium (13–18), and high risk (≤12). A significance criterion of .05 was used for all tests. Statistical analyses were completed using IBM SPSS Version 20 (IBM Corp, Armonk, NY).

A total of 555 patients were enrolled over a 5-month period (September 2015–January 2016). Mean age was 60 ± 19 years (range, 18–97 years), and there were more men than women (54% vs. 45%, respectively), with racial groups reflecting the served community (54% White, 16% Asian, 6% Black, 2% Hawaiian/Pacific Islander, <1% American Indian or Alaska Native, and 20% not declared).

Compliance with turning

A total of 44 870 hours of monitoring data were analyzed. Mean monitoring time was 73 ± 97 hours per patient (range, 4–890 hours). More patients were enrolled from ICU B than ICU A (66% vs. 34%, respectively) and admitted to a surgical service than a medical service (63% vs. 37%, respectively).

Overall compliance with turning was 54%. There were statistical differences in compliance across nursing shifts; night shift (11pm–7am) had significantly lower compliance with turning protocols (x– = 46%) than morning shift (7am–3pm; x– = 56%) and afternoon shift (3pm–11pm; x– = 56%) (95% confidence interval [CI], 7–13%; x– = 10%; F = 50.1; degree of freedom [df] = 775; P < .005; Figure 3). Using a 2-way ANOVA, a main effect of shift was significantly associated with compliance (F = 49.5; df = 2; P < .005), while day of the week and an interaction term of both main effects were not significant (P = .59; P = .96, respectively).

Quality of turning

Overall. In total, 27 566 patient turns were recorded and analyzed. The average turn magnitude was 24° ± 29°. Less than half of all turns were greater than the 20° pre-established turning angle threshold (10 623/27 566; 38.5%). Despite differences in compliance by shift, there were no meaningful clinical differences in the turning magnitude between the different shifts (afternoon, 25° ± 29°; night, 25°±29°; morning, 22° ± 28°). The distributions of all turns performed in the study are shown in Figure 4.

Intensive care unit and service team. The proportion of patients admitted to either a surgical or medical treatment team varied between ICU units, with ICU A admitting a greater proportion of surgical patients than ICU B (90% vs. 51%). There were no meaningful differences in the average turn angle between medical or surgical patient populations (25° ± 29° vs. 23° ± 28°; P > .05). A significant difference in mean magnitude of turning existed between the 2 ICUs (ICU A, 21° ± 28° vs. ICU B, 26° ± 29°; P < .0005).

Patient characteristics

Sex and age. Compliance differed significantly by sex (men 49% vs. women 57%; P < .005) and by body mass index (BMI) (P < .0005; Figure 5), with men with a high BMI (>30) having the lowest compliance with the turn protocol (43%). However, there were no differences in the average turning magnitude for men and women (24° ± 28°; 24° ± 29°, P > .05). To determine whether age influences turning behaviors, patients were stratified into 4 groups: 18–40 years, 41–60 years, 61–80 years, and >80 years. Patients 80 years or older had the highest proportion of compliant care (62%; F = 3.812; df = 553; P = .01), estimated to be up to 18% more than patients 41–60 years and 17% more than patients 61–80 years.

Braden Risk Score. Patients were stratified into 3 groups based on their Braden risk score on admission to the ICU (≤12, 13–18, ≥19). Compliance with turning differed significantly between Braden groups (F = 8.5; df = 548; P < .005), with low-risk patients (Braden score ≥ 19) experiencing greater compliance (66% compliant time) than high-risk patients (Braden score ≤ 12) (55% compliant time; mean difference 95% CI, 4%-18%); this difference remained on day 3 of ICU admission. High-risk patients had significantly larger mean turning angles than lower-risk patients (21° vs. 12°; F = 490; df = 27281; P < .005), possibly due to increased activity in lower-risk patients.

Depressurization time. Of the 27 566 eligible turns, only 37.8% of patients (n = 10 421) remained in their new position for at least 15 minutes (mean time, 43 minutes). Men spent a significantly longer time in a new body position than women (48 vs. 39 minutes; P < .0001). There were no significant differences in depressurization time based on age group or nursing shift (P > .05). The distribution of body position between the left, back, and right sides was unequal with patients spending the majority of time in a supine position (71.6%), followed by their right (16.3%) and left sides (12.1%).

This is the first study to use a patient sensor to objectively quantify nursing care related to the prevention of HAPIs in patients admitted to an ICU. Through analyzing nearly 45 000 hours of monitoring data and more than 27 000 patient turns, the investigators found turn frequency, turn magnitude, depressurization time, and position distribution to be suboptimal. Furthermore, they identified biases in care delivery based on a patient’s sex and BMI. These findings extend the understanding of turning care practices in the ICU in several important ways.

Much has been written about pressure injuries (previously referred to as pressure ulcers). A search in PubMed reveals approximately 17 000 citations dating back to the 1940s, with nearly 1200 published clinical trials. Over the past 2 years alone (2015-2016), more than 1400 related articles have been published and almost 50 clinical trials. Despite extensive literature, preventive methods for HAPIs have essentially remained unchanged for 150 years, with the routine repositioning of patients every 2 or 4 hours depending on the clinical setting. Repositioning of patients has become an important part of clinical care in the ICU; however, little is known about the actual delivery of this care.

Studies attempting to measure turning practices in the ICU have been limited due to the need to manually observe clinical practices, significantly limiting the scale of the studies conducted. Manual research methods are time-consuming, limit clinical trial size, may miss care episodes, and introduce significant observer and reporting biases. For example, Krishnagopalan et al² conducted a prospective observational study of 74 patients, performing manual observations every 15 minutes for a total observation time of 566 hours. They found only 3% of patients received regular repositioning every 2 hours and 49% of patients remaining unchanged for longer than 2 hours. Similarly, Goldhill et al⁹ used a small number of hourly observations on 393 patients from 48 ICUs in the United Kingdom. They found the median time for repositioning was 4 hours, with only 42% of patients receiving a second   hourly repositioning.

Unlike the studies by Krishnagopalan et al² and Goldhill et al⁹, technology such as the small patient sensor allows for larger sample sizes, is less likely to introduce observer bias, and is considerably more accurate. A further methodological advantage above sample size is the resolution of available data. This is the first study of turning practices that accurately, quantitatively, and continuously measures patients across 2 ICUs for their entire duration of admission. Without the need to manually observe clinical care, the resolution of data in the present study (measurements every 10 seconds) is significantly greater (~45 000 hours) than that of prior studies^2,9 that have relied on repeated manual observations within small cohorts.

The study results herein unveil important new information about care delivery related to patient turning. For the prevention of HAPI, the quality of turning is as important, if not more important, as the turning frequency. This is the first study to report that the magnitude of the majority of patient turns performed in the ICU are less than the recommended⁸ 30° angle threshold. In addition, once turned, less than half of patients achieve at least 15 minutes of tissue depressurization time. If the goal of preventative turning is to give pressurized tissue time to recover, it stands to reason that efforts should be made to ensure patients remain in their new position long enough to allow for appropriate tissue depressurization. Studies of wheelchair-bound patients estimate that tissue reperfusion is complete after pressure-relieving maneuvers of 5 minutes.^10,11 However, adequate reperfusion time in acutely ill patients admitted to the ICU bears further study. In the meantime, the inability to reliably measure these parameters is a significant issue when considering both turn magnitude and depressurization time in the ICU. It is important to note that neither this study nor the primary study attempted to influence the turn magnitude or the depressurization time of patient turns by using foam wedges or fluid positioners; however, the investigators strongly encourage the use of these support devices to help reach and maintain relief positions for an adequate amount of time.¹²

Patient acuity and hemodynamic instability are often cited as reasons why patients in the ICU do not receive recommended preventative turning, although recent data show acuity should not be a barrier to turning, as the benefits of patient turning outweigh the perceived risks.^13,14 The support for this position is growing. In their consensus statement, Brindle et al¹⁵ concluded there is evidence that prolonged bed rest and immobility cause negative clinical effects, with intolerance developing due to sustained body position and gravitational forces.¹⁶ Further, it has been shown that failure to turn patients in the ICU due to perceived instability contributes to a higher incidence of HAPIs in this patient population.¹³ A go slow approach to patient turning is recommended, with supportive therapies being used as necessary (ie, use of vasopressors to support blood pressure).

In addition to describing previously unmeasured turning characteristics, the investigators also have uncovered biases in care delivery. The data reveal that within this cohort, men received significantly less preventative turning than women, and once turned, men spent an average of 10 minutes longer in their new body position than women. Historically, it was thought that women were at an increased risk for developing HAPIs¹⁷ ; however, a recent examination¹⁸ of more than 3.4 million pressure injuries across the United States between 2008 to 2012 found the opposite, with the incidence of pressure injuries higher for men, regardless of age, than women (men: 2%, n = 325 293/16 366 959 vs. women: 1.6%, n = 351 110/22 645 567) (odds ratio = 1.282; 95% CI, 1.276–1.288; P < .001).¹⁸

With the vast majority of the nursing profession being female,¹⁹ one explanation for gender differences in the delivery of preventive turning may be the difficulty nurses have in manually turning larger bed-bound patients. Nurses in the ICU may be able to reposition female patients independently with little to no assistance; however, due to body mass, female nurses may require additional assistance in repositioning male patients, delaying this activity. The finding that compliance with turning protocols reduces as BMI increases (Figure 5) further supports this theory. Consequently, the data presented here show a larger male patient (BMI > 30) has nearly 20% less compliant turning time than an average-sized woman and spends an average of 23% longer in a given position. These physical limitations in care delivery need to be overcome to eliminate disparities in care. Technology, such as the wearable patient sensor, allows providers to identify care disparities in real time to help ensure delivery of a high-quality prevention program for all patients at risk for pressure injuries.

Findings from this study should be weighed with the following limitations. This study included all patients within the ICU, regardless of risk for pressure injury. It is plausible that spontaneous patient movements were captured by the sensor and influenced compliance; however, a patient would have to be in an originating position for at least 1 minute and self-turn with a sufficient magnitude > 20°. Notwithstanding, physiologically, there should be no difference between self-turns and caregiver-assisted turns as long as the parameters of the movement are identical.

Second, the investigators have established new quantitative parameters for reducing signal noise and measuring and classifying patient turning. It is possible that these thresholds have influenced either a loss or gain in the percentage of observed versus actual compliance time.

Finally, as with all unblinded studies, a potential for observer bias exists. As these data correspond closely to unpublished quality control data collected prior to commencing the RCT, the authors believe any such bias, if present, to be negligible.

Further studies are needed to characterize and validate these behaviors in ICUs as well as confirm the external validity and generalizability of these results.

This is the first study to use a wearable patient sensor to describe preventive turning practices within ICUs. The investigators found that dynamically measured turning frequency, turn magnitude, and tissue depressurization time to be suboptimal. They also uncovered biases in care delivery based on a patient’s sex and BMI. This study highlights the need to reinforce best practices related to preventive turning and to consider staff and patient factors when developing individualized turn protocols. Wearable patient sensors can be used to identify and help mitigate disparities in care delivery. As the first study of its kind, further studies are needed to validate and refine these findings.

The authors acknowledge the leadership of Nancy Lee and Wendy Foad, past and present Vice President of Patient Care Services & Chief Nursing Officer’s; Pravene Nath, Chief Digital Officer; staff of the participating units; Jennifer Brown and the institutions compliance and ethics bodies from Stanford Health Care, Stanford, CA.

Affiliations: Office of Research, Patient Care Services, Stanford Health Care, Stanford, CA; Division of Primary Care and Population Health, Stanford University, Stanford, CA; Leaf Healthcare, Inc, Pleasanton, CA; and Patient Care Services, Stanford Health Care

Correspondence: David Pickham, PhD, RN, 301 Ravenswood Avenue, Suite I238, Menlo Park, CA 94025; dpickham@stanfordhealthcare.org

Disclosure: The primary study was co-funded by Leaf Healthcare, Inc (Pleasanton, CA). Mr. Pihulic and Dr. Larson are employees of Leaf Healthcare, Inc.