Improving Hospital Infant Safe Sleep Compliance by Using Safety Prevention Bundle Methodology

Penn State Children’s Hospital is a tertiary care, academic, pediatric hospital in Southcentral Pennsylvania, with 114 inpatient beds, including a 42-bed NICU. There are 2100 deliveries per year, and the NICU has ∼600 admissions per year. This QI project was aimed to improve safe infant sleep practices in the following units: NICU, general inpatient, hematology and oncology, peri-anesthesia, PICU, pediatric intermediate care, and maternity and NBN.

This project was approved as exempt by the Penn State College of Medicine Institutional Review Board.

The hospital safe sleep committee (a team of physicians, registered nurses, occupational therapists, and child life specialists) began meeting in July 2016 to update the existing safe sleep policy to reflect coverage beyond the NICU and the maternity ward and to create elements of the HAC bundle, which are detailed in the measures section below. Members of the safe sleep committee audited a convenience sample of admitted infants (aged <1 year) each month from March 2017 through October 2019 (no data collected June 2017), ensuring representation from each unit and from day and evening shifts. Audit methodology (direct observation of compliance with HAC bundle elements) was the same across units and shifts. The goal was to audit at least 30 infants per month on the basis of the minimum subgroup size common guideline for p-charts.²⁴ 

Interventions were chosen by team members during multidisciplinary team meetings. Four key time points and/or interventions across the study period were (1) nursing education on the HAC bundle, and the updated policy was completed in August 2017; (2) the HAC bundle and policy were both implemented on September 15, 2017; (3) nursing councils began to share audit data in March 2018; and (4) peer-to-peer bundle checklist reviews during registered nurse shift handoff (by using a reference card listing the bundle elements), and electronic medical record (EMR) addition to create a nursing task to provide parent safe sleep education (triggered automatically by admission order for a patient aged <1 year), were implemented in March 2019. A key component of both the bundle and policy included health care providers’ role modeling safe sleep practices. Ongoing education was provided to various entities throughout the organization, including faculty, volunteers, and nursing staff.

Although there was constant iterative learning from monthly audit reports, other notable education interventions during the study period included the following: (1) September 2017: education to parents through updating newborn video instruction and increased exposure to appropriate safe sleep modeling; (2) March 2018: safe sleep committee members visited nursing unit councils to share audit data; and (3) February 2019: subject matter expert training by guest speakers from the University of Pennsylvania as part of a Pennsylvania Department of Health grant to improve safe sleep in Pennsylvania (Pennsylvania Safe Sleep Program) and distribution of new parent education brochures with clear pictures of safe and unsafe sleep practices.

In addition to overall compliance (compliance with all measures combined), there were 3 outcome measures that were tracked over the study period: infant in supine position, head of bed flat, and sleeping space free of extra items. We also included the process measure of caregivers having received safe sleep education, which was recorded in the EMR by the nurses after completion. Medical exceptions to supine and flat sleep position included airway compromise (including Pierre-Robin sequence), congenital malformation, life-threatening gastrointestinal reflux, palliative care, respiratory distress, and other, which were specified by the ordering provider. Only patients with medical exceptions were excluded from data reporting.

All data are categorical and therefore statistical control charts (p-charts) were used to measure progress on the above measures. Charts were created with QI-Charts version 2.0.23 (Scoville Associates, Austin, TX, 2009). Rules used for identifying special cause were (1) a single point outside control limits, (2) a run of ≥8 points in a row above or below the centerline, and (3) 6 consecutive points increasing or decreasing.²⁴ 

Eighteen percent of eligible patients were audited (average of 53 audits per month and an average 293 eligible admissions per month). Overall compliance with all 4 measures was low in the beginning of our quality initiative (9%), and there was a centerline shift to 36% after implementation of the updated policy and HAC bundle (Fig 1). Further special-cause variation was noted in the context of further QI efforts in January and July 2019, with centerline shifts to 53% and 72%, respectively. The infant in supine position measure (Fig 2) stayed relatively constant over the entire study period, with no centerline shift (81%). Keeping the head of bed flat (Fig 3) had a starting centerline of 62%, with special-cause variation in April 2018 coincident with sharing of audit data, leading to a centerline shift to 80% and a further centerline shift in October 2018 to 93%. Maintaining a sleep space free of extra items began with a centerline of 52% and shifted to 81% in April 2018, concurrent with sharing of audit data (Fig 4). The process measure of caregivers receiving safe sleep education increased from 10% to 66% to 84% during the study period (Fig 5), with the first special-cause variation occurring when nursing education was rolled out in August 2017 and the second special-cause variation occurring in February 2019, which was not immediately preceded by a major intervention. Although there was variation among individual units, the trends were all similar, and thus we present aggregate hospital data only.

Our study reveals that it is possible to use the HAC safety prevention bundle model and apply it to safe sleep to improve adherence to the bundle of AAP guidelines for all hospitalized infants. Safe sleep recommendations should be followed by both staff and family every time an infant who does not have a medical exception enters a children’s hospital. Consistent modeling of appropriate practices will reinforce safe sleep standards for families once they return home, with the potential to save lives.¹¹  Evidence-based safety prevention bundles have been revealed to be effective ways to help mitigate harmful events, such as HACs, that can develop in a medical setting.^25–28 

HAC bundles are traditionally used for conditions that have the potential to cause harm when a patient intersects with the health care system. These conditions are more common in, but not exclusive to, the hospital setting. Typical HAC conditions lend themselves to be counted in a binary fashion, such as, “Was this infection due to an indwelling catheter or central line?” Although the risk of unsafe sleep is not often thought of as a risk due to a hospital-related condition, the process of implementing safety bundle methodology was useful for increasing adherence with the different AAP-defined measures. In our study, we use audits and run charts to measure failure to adhere to the prevention bundle, as opposed to the traditional HAC (eg, CAUTI), which measures audits but also actual clinical events that occur because of failure to adhere to the prevention bundle. Having clearly defined bundle elements, instructions on how to perform these elements, and knowledge that they were being measured helped improve our compliance. In addition, adding peer-to-peer bundle checklist reviews with a ready reference of the bundle helped to improve staff knowledge and engagement with safe sleep practices. The above measures enabled us to strive to the audit goal of >80% adherence revealed to decrease traditionally measured HACs.^9,10 

Applying the principle of an HAC bundle to safe sleep presents some challenges because there are many components to a safe sleep environment.²  In a hospital setting with a wide variety of patients, from typical newborns to infants with complex medical needs, developing scoring criteria to capture all of these elements was challenging because there are potentially many exceptions to the AAP policy because of the medical conditions of a patient. By measuring some of the individual safe sleep elements, we were able to get a clearer picture of compliance and opportunities for improvement than would have been possible by simply calculating overall compliance.

The measurement parameter infant in a supine position was well established at the start of our quality initiative and was maintained at a high rate of compliance throughout the QI initiative (Fig 2). This is the one variable we measured that did not reveal improvement. This may be expected given the focus of the 1994 AAP Back to Sleep campaign was to place all infants to sleep on their back. This message continues to be taught today as part of the safe sleep message; it is reassuring to see our high rate of compliance. However, we should continue to strive to achieve 100% compliance to this measure.

Keeping the head of the bed flat appeared to be the metric that is most challenging for many hospitalized patients. At our institution, it was common to slightly elevate the head of newborn bassinets before this intervention. This antiquated practice was found to be supported by nurses and medical providers despite evidence and recommendations to the contrary.²⁹  There is no evidence that infants have more reflux if the head of the bed is flat, and the AAP policy clearly states that the sleeping surface should be flat.²  Through our education efforts, we were able to reveal improvement in this metric (Fig 3).

At baseline, adherence to sleeping space free of extra items was lower than the infant in a supine position measure and was an area for which ongoing education and reminders were needed (Fig 4). Avoiding loose bedding in the sleep environment is especially important for infants outside of the NBN because 1 study revealed that infants aged ≥4 months more often succumb to SUID from rolling into objects, whereas younger infants die more often from bed-sharing.³⁰ 

Providing and documenting parent education was identified to be the bundle element with greatest opportunity for improvement early in the project (Fig 5). Staff roles in providing consistent and standard education and modeling to families was crucial to the success of this project. Shared monthly compliance data reports, provision of new parent education pamphlets, and creation of a reminder task in the EMR all helped to improve compliance.

We encountered some unique barriers in implementing our HAC bundle to units outside of the NBN and NICU. Designing a process for appropriate medical exceptions to the safe sleep bundle was the most significant challenge. Initially, the ordering provider documented one of the following medical exceptions to supine and flat sleep position when appropriate: airway compromise (including Pierre-Robin sequence), congenital malformation, life-threatening gastrointestinal reflux, palliative care, respiratory distress, and other. Anecdotally, we found that life-threatening gastrointestinal reflux became overused for many infants who seemingly had typical infant reflux. Consistent guidelines that defined life-threatening reflux were missing, which had 2 important consequences. First, an exception card was placed in the infant’s bed noting that the child had a life-threatening condition when most of them did not. This notation caused increased parental anxiety, which was expressed to hospital staff. Second, staff were modeling poor safe sleep for the parents when the decision to list an exception was often made arbitrarily. In 2018, the North American Society for Pediatric Gastroenterology, Hepatology and Nutrition released guidelines recommending “not to use positional therapy (ie, head elevation and lateral and prone positioning) to treat symptoms of GERD in sleeping infants.”²⁹  As a result of our experience, we have since removed life-threatening reflux as an option; however, in those rare truly severe circumstances, providers can still list it under other.

Although many medical exceptions listed were related to supine flat positioning, auditors reported that once this particular exception was selected, the remainder of the guidelines were often ignored. For example, if a child has had spinal surgery, keeping them flat on their back is not practical nor advisable, and thus it is reasonable to place a medical exception for supine positioning. However, we often found that even if an exception was made that required prone positioning, the infant’s sleep environment would become unsafe from loose bedding and extra items in the crib. We subsequently clarified our medical exception rule so that it applied only to positioning and that the other components of the bundle, namely, sleep space free of bedding and nonmedical objects, still had to be followed.

A final barrier was resistance from providers and staff to implementation in older infants. Although the AAP policy and our institutional policy all stated that safe sleep guidelines should be continued through 1 year of age, many staff and providers were not adherent once the infant was rolling over independently and certainly once the infant could sit up independently. These concerns required repeated reinforcement.

Our study included a number of limitations. First, we tried to standardize the audit process but found there were some situations for which scoring the compliance on the audit was not clear, primarily occurring in units outside of the NBN. Ultimately, there was a degree of subjectivity as to how an auditor scored the patient’s sleep environment. This problem could be mitigated in the future by periodically conducting group reviews or photograph reviews to increase interobserver reliability. Second, the success of HAC prevention bundles is commonly revealed through reduction of cases, such as fewer CAUTIs and central line–associated bloodstream infections. The outcome from implementing a safe sleep HAC bundle, namely, prevention of unsafe sleep injuries or deaths, would be extremely difficult to evaluate. However, the same evidence-based approach to prevention that is the basis for all HAC methodology (HAC avoidance) was used in this study. Third, the fact that the medical exceptions were provider driven, and did not have standard definitions (eg, what constitutes respiratory distress), may mean that more patients were excluded from our study than needed to be. In future analyses, it would be instructive to include a process measure of the number and type of medical exceptions granted. Finally, this was a single-center study at an academic children’s hospital, and thus our experience may not be generalizable to other centers.

Embedding safe sleep into the hospital QI priorities as one of our Children’s Hospital’s HACs will be helpful for long-term sustainability, as will the EMR audit alerts. Further research would be valuable to understand if the improved in-hospital rates of safe sleep translate to increased compliance in the infant’s usual sleeping space once discharged from the hospital.

Many infant deaths attributed to unsafe sleep practices have preventable risk factors. It is important that any time an infant aged <1 year enters the inpatient hospital system, staff adhere to AAP standards for safe sleep and parents observe and are repeatedly educated about these standards. Using an HAC safety prevention bundle approach, we have shown we can increase the adherence to infant safe sleep guidelines throughout an entire children’s hospital.

We thank the entire Penn State Children’s Hospital Safe Sleep Committee for their tireless efforts to keep all infants in a safe sleep environment, Jessica Beiler, MPH, for her assistance in the institutional review board process, and Meredith Buckley for her assistance in preparation of the figures.

AAP

American Academy of Pediatrics

CAUTI

catheter-associated urinary tract infection

EMR

electronic medical record

HAC

hospital-acquired condition

NBN

newborn nursery

QI

quality improvement

SPS

Solutions for Patient Safety

SUID

sudden unexpected infant death