Safety and Quality of Pediatric Care in Freestanding Children’s and General Hospitals

We used a large survey of registered nurses in 4 states in 2015–2016 to create measures of nurse-reported patient safety and quality in acute care hospitals. The overall response rate of the RN4CAST-US nurse survey was 26%. The nurse responses resulted in data on >95% of hospitals in our sampling frame; and a nonrespondent survey revealed no statistically significant differences between nurses who respondent to the main survey and nonrespondents, which suggests no evidence of nonresponse bias in the main sample. The methodology for surveying nurses has been described elsewhere.¹⁴  In brief, state licensure lists were used to contact a 30% random sample of registered nurses in California, Florida, New Jersey, and Pennsylvania at their home addresses. In the mailed survey, nurses were asked to provide demographic information, their primary organization of employment, and their assessments of patient safety and care quality. Nurses who reported working in a hospital were also asked to identify their primary unit. In this study, we restrict our analysis to registered nurses working on general pediatric units, PICUs, or NICUs in an acute care hospital, which yielded a final sample of 2094 nurses.

Nurse respondents also reported the names of their hospitals, which allowed us to link the nurse respondents with hospital data from the American Hospital Association annual survey and the Children’s Hospital Association data. These data sets were used to classify our 330 study hospitals into 2 types: (1) FCHs and (2) GHs. FCHs were defined by using American Hospital Association service codes and Children’s Hospital Association designations. In some cases, children’s hospitals were embedded in a larger health system. We contacted these hospitals individually and classified them as an FCH if they had a separate chief executive officer and chief nursing officer from the parent hospital or health system. All other hospitals employing pediatric nurses in our sample were characterized as GHs with pediatric units.

The American Nurses Credentialing Center database was used to identify hospitals that were Magnet recognized during 2015, 2016, or 2017. Our final sample included 2094 nurses in 330 hospitals. Among the 330 hospitals, 21 hospitals were FCHs and 309 were GHs. Among both types of hospitals, 74 were Magnet recognized.

Our outcomes of interest were nurse-reported measures of patient safety and quality based on validated measures from the Agency for Healthcare Research and Quality Hospital Survey on Patient Safety Culture.¹⁵  Nurses were asked to provide an overall safety grade for their hospital. Choices ranged from “A” for excellent to “F” for failing. We evaluated the overall safety grade by the percentage of nurses who gave their hospital an acceptable grade for safety, as indicated by either an A or B grade. The remaining 7 safety items were measured on a 5-point Likert scale from strongly agree to strongly disagree and included a neutral option. The responses of “agree” and “strongly agree” correspond to desirable outcomes for the following statements: staff discuss ways of preventing errors from happening again, staff receive feedback on changes put into place on the basis of event reports, actions of management reveal that safety is a priority, staff feel free to question actions or decisions of those in authority, patient information is not lost during shift or provider changes, things do not fall between the cracks, and mistakes are not held against them.

Nurse-reported quality measures included the quality of nursing care (measured on a 4-point scale: excellent, good, fair, and poor) and whether the nurse would definitely recommend his or her hospital to family and friends (measured on a 4-point scale: definitely yes, probably yes, probably no, and definitely no).

To account for potentially confounding relationships between safety and quality measures and the hospital type (ie, FCH versus GH), we adjusted for nurse and hospital characteristics. Nurse characteristics included age, sex, years of experience as a registered nurse, highest level of educational attainment in nursing, whether the nurse had a national nursing certification in a specialty practice area, and the type of unit the nurse worked on. Hospital characteristics included size (based on the number of beds), teaching status (based on the resident-to-bed ratio), and technology capabilities (based on whether the hospital could perform a major organ transplant and/or open-heart surgery).

To compare hospital characteristics and safety and quality assessments between FCHs and GHs, χ² tests were used. Multivariate logistic regression models were used to estimate the odds of a pediatric nurse reporting favorably on measures of quality and safety given that he or she worked in an FCH. Models were estimated sequentially, first as the unadjusted effect of FCHs on quality and safety and then accounting for hospital-level and individual nurse characteristics. The final model included all hospital and nurse covariates and hospital Magnet status. P < 0 was considered statistically significant for all analyses. Regression models were used to account for clustering of nurses within hospitals by using Huber-White sandwich estimators.¹⁶  Stata 15.1 was used for all computations (Stata Corp, College Station, TX). Institutional Review Board approval was granted by the University of Pennsylvania.

The final sample included 2094 pediatric nurses in 330 hospitals. Among the 330 hospitals, 21 were FCHs, whereas the remaining 309 hospitals were GHs with pediatric units (Table 1). There were no differences in hospital size between the 2 hospital types; however, FCHs were more likely to be major teaching institutions (33.3% vs 9.7%; P = .001) and less likely to have high technology capabilities (42.9% vs 52.8%; P = .041) compared with GHs, although our measure of technology is hospital-wide and not specific to pediatrics. Roughly 22% of hospitals in our sample were Magnet recognized. FCHs were significantly more likely to be Magnet hospitals than GHs (47.6% vs 20.7%; P = .004).

Overall, pediatric nurses did not report favorably about the safety and quality of care in their hospitals (Table 2). Although 70.3% of nurses reported an overall safety grade of A or B, only 45.6% of nurses rated the nursing care quality as excellent, and less than half (49.2%) of nurses would definitely recommend their hospital to family and friends. In terms of the culture of patient safety, one-quarter of nurses (25.6%) felt that mistakes were not held against them, and approximately one-third of nurses reported that things did not fall between the cracks (35.7%) or that patient information was not lost during shift or provider changes (38.8%). The majority of nurses discussed ways of preventing errors from happening (70.7%), but fewer (58.9%) felt that the actions of management revealed that safety was a priority. Nurses in FCHs reported more favorably than, or no differently from, nurses in GHs on the safety and quality measures studied.

The bivariate results reported in Table 2 reveal the differences in safety and quality assessments for nurses in FCHs and GHs. The odds ratios (ORs) shown in Table 3 are from multivariate logistic regression models before and after adjusting for hospital-level and individual nurse–level covariates. After we accounted for hospital- and nurse-level factors, nurses in FCHs were still significantly more likely to report favorably compared with nurses in GHs on the following measures: actions of management reveal that safety is a priority (OR 1.46; 95% confidence interval [CI] 1.02–2.07), staff receive feedback on changes put into place on the basis of event reports (OR 1.58; 95% CI 1.06–2.35), and nurse definitely recommends hospital to family and friends (OR 1.53; 95% CI 1.03–2.27) (Table 3, model 2).

After accounting for the hospital’s Magnet status, in the final model, relationships between FCHs and favorable safety and quality assessments were rendered insignificant, suggesting that Magnet status and/or the organizational features common to Magnet hospitals explain, in large part, the observed better quality and safety of care in FCHs. Nurses in Magnet hospitals were more likely to report favorable overall safety (OR 1.62; 95% CI 1.21–2.15) and were more likely to say that actions of management reveal that safety is a priority (OR 1.35; 95% CI 1.00–1.83), that patient information is not lost during shift changes (OR 1.33; 95% CI 1.07–1.65), and that things do not fall between the cracks (OR 1.30; 95% CI 1.05–1.60). With respect to the quality measures, nurses in Magnet hospitals were more likely to rate the nursing care in their hospital as excellent (OR 1.56; 95% CI 1.22–1.99) and were more likely to definitely recommend their hospital to family and friends (OR 1.42; 95% CI 1.02–1.98).

The study hypothesis, that quality and safety indicators are more favorable in FCHs compared with GHs, was confirmed. This finding is consistent with previous research that revealed that FCHs have better nursing resources known to be associated with better patient outcomes and care quality.¹⁷  However, it is difficult and often impossible to replicate structural characteristics of institutions or health systems like FCHs, and replicating structure does not always yield expected outcomes.¹⁸  Thus, we were interested in learning about modifiable features of better-performing pediatric hospitals that could potentially be translated into improvements in pediatric acute care settings more broadly.

We explored the extent to which potentially modifiable and replicable features of nursing, whether in freestanding or general hospitals, were associated with better quality and safety indicators. To that end, we found that nearly half of the FCHs in our study were Magnet hospitals, which is markedly higher than the 20% of GHs that were Magnet hospitals.

Our findings reveal wide variation in the quality and safety of inpatient pediatric settings. Only 46% of nurses reported excellent quality of nursing care in their hospitals, suggesting room for improvement among FCHs and GHs alike. Nurses in FCHs had 53% higher odds of definitely recommending their hospital to family and friends, compared with nurses on pediatric wards in GHs. However, after accounting for Magnet status, the relationship between FCH and quality and safety was no longer significant. This suggests that a hospital’s Magnet status, more so than its structural organization as a freestanding hospital, is important to the provision of high-quality and safe patient care. Furthermore, we found that quality and safety indicators were better in Magnet hospitals overall, regardless of whether they were FCHs or GHs. Nurses in Magnet hospitals had 62% higher odds of reporting a good overall safety grade and 42% higher odds of definitely recommending their hospital.

Although we confirmed our hypothesis that FCHs are associated with better quality and safety, we identified Magnet status and the organizational attributes represented by Magnet status, such as supportive work environments and evidence-based staffing levels, as a likely explanation for more favorable care. Attributes of the nurse work environment common among Magnet hospitals can serve as exemplars for guiding improved quality and safety in both FCHs and GHs alike.

The organizational standards that are central to Magnet recognition can serve as a blueprint for improving the nurse work environment and patient care.^12,13,19,20  Attributes common to hospitals with excellent nurse work environments that are relevant to patient safety and quality include a high level of nurse engagement in managerial decision-making and implementation of clinical innovations likely to improve quality and safety. The inclusion of clinical nurses on important hospital committees and investment by management in coaching them to be effective committee members has been shown to be associated with better outcomes.^21,22  When organizations value the embedded human capital of nurses by promoting their clinical autonomy, allowing nurses to identify and solve safety problems and providing the adequate staffing and resources to surveil patients, patients experience better quality of care. In the pediatric setting, researchers have identified relationships between greater amounts of missed nursing care and poorer safety ratings in hospitals with unfavorable work environments, including inadequate nurse staffing.^23,24  Magnet recognition does not require a specific patient-to-nurse staffing ratio but does require that hospitals use evidence to determine a safe nurse staffing level and evaluate its effectiveness over time. It is telling that patient safety interventions rarely include mention of nurse staffing adequacy, although the research literature evaluating effectiveness of safety interventions under real-life circumstances, such as bundled-care interventions to prevent central-line infections, reveal that a high degree of fidelity to protocols is required but often does not happen because of nurse staffing that falls below targeted levels.²⁵ 

Among the limitations of our study is the absence of objective clinical evidence of adverse quality and safety indicators as well as case-mix or severity-of-illness risk adjustments. Clinical measures available on all acute pediatric patients in administrative data are scarce. Preventable inpatient deaths are low in pediatric acute care. Patient satisfaction, another informative measure of adult quality and safety, is not available for pediatric care. We relied on nurse reports of quality and safety, which have been shown in empirical research to be highly predictive of objective clinical patient outcomes in adult²⁶  and pediatric patient populations.^24,27  Our data are from 4 large states that account for approximately one-quarter of all hospital admissions nationally; the composition of hospitals is roughly comparable with hospitals nationally. A strength of the study is that almost all hospitals in the 4 states are included, thus avoiding response bias at the hospital level, which is a major problem in studies of hospital performance. We have previously documented that response bias among nurse nonrespondents to the survey is not a limitation.¹⁴  Finally, costs²⁸  and market conditions²⁹  are among the major factors in a hospital’s decision to apply for Magnet recognition. Although an economic resource analysis was outside the scope of this study, it is worth acknowledging that Magnet hospitals may have greater financial resources than non-Magnet hospitals to invest in other organizational areas, which may also improve patient care quality and safety.

Our findings are consistent with research suggesting that progress in advancing care quality and safety may be hampered by greater confidence than is warranted in the power of structural changes to improve outcomes and by too little attention to patient-centric cultures, adequate investments in clinical human resources, and work environments that enable health professionals to provide excellent care.²⁸  More than 15 years ago, in the landmark report, Keeping Patients Safe: Transforming the Work Environment of Nurses, the Institute of Medicine highlighted the importance of the nurse work environment in promoting patient safety and quality.³⁰  Our findings support this agenda and identify Magnet adoption, or a focus on the core principles of Magnet nurse work environments, as a mechanism for improving pediatric inpatient safety and quality.