Since the Institute of Medicine (IOM) estimated in 1999 the number of preventable hospital acquired conditions (HACs; IOM, 2000), several national quality improvement initiatives have been implemented. One example is the Centers for Medicaid and Medicare Services (CMS) non-payment policy whereby hospitals no longer receive payment for certain HACs. This policy started in 2008 as part of the CMS long-standing Value-Based purchasing initiative and was later authorized by Congress under the Affordable Care Act (ACA; CMS 2015a). Another initiative is CMS funding in 2011 of 26 Hospital Engagement Networks (HENs) with approximately 3,700 hospitals. The goal of HENs was to identify solutions to reduce HACs and disseminate them to other hospitals and healthcare providers (2015b). As a result of the increased focus on HACs, the Agency for Healthcare Research and Quality (AHRQ) reported a decline of 17% in HACs from 2010 to 2014 (AHRQ, December 2015).

Improvement to decrease pressure ulcer rates and fall rates is especially important because these are among the most common, preventable, and costly HACs. However, in 2014 the rate of HACs was 121 per 1000 discharges, clearly demonstrating that more work is required. Improvement to decrease pressure ulcer rates (PURs) and fall rates (FRs) is especially important because these are among the most common, preventable, and costly HACs (ASHRM, 2015). While PURs decreased by 23% from 2010 to 2014, more than 1 million patients still developed a pressure ulcer in 2014. No change was noted for FRs, with 260,000 patients falling every year (AHRQ, 2015).

Although multicomponent care interventions to prevent pressure ulcers and falls are well-established, evidence about which contextual factors influence PURs and FRs is mixed or lacking (Shekelle et al., 2013). It is generally understood that contextual factors are important to consider when examining HACs, but challenges remain about how context is defined and how it affects HACs. Using the broadest definition: context is “everything that is not the intervention” (Ovretveit et al., 2011, pg. 2). Contextual factors that influence PURs and FRs can be conceptualized at three levels: community, hospital, and nursing unit characteristics. This article reports on the study methods, noting results that demonstrated differences across all characteristics. The discussion considers the findings and implications in the context of rural or urban location.

Contextual Factors

Community Characteristics
An especially important community-level contextual factor is hospital geographic location. An especially important community-level contextual factor is hospital geographic location. Both the hospital and the community it serves differ in rural and urban locations. In the United States (US), there are approximately 2,000 rural hospitals, which differ significantly from the nearly 3,600 urban hospitals (American Hospital Association [AHA], 2011). Compared to urban counterparts, rural hospitals have had a more dramatic shift of care to outpatient settings with subsequent lower patient volume to generate reimbursement and meaningful public reporting (American Hospital Association, 2011). Because rural hospitals have these differences, many have public reporting exemptions and cost-based reimbursement (Casey, Moscovice, Klingner, & Prasad, 2013; Holmes, Pink, & Friedman, 2013).

Approximately 19% of the population (59.5 million) resides in 72% of U.S. land area considered rural (Economic Research Service, 2013; United States Department of Agriculture [USDA], 2013). As access to care remains the most frequently identified rural health priority, rural hospitals are often the only available community healthcare source (Bolin et al., 2015; IOM, 2005; Jaffe, 2015). Several community factors influence the ability to prevent HACs in rural hospitals. For example, compared to urban areas, rural communities have higher rates of poverty and a higher percent of people with chronic diseases and disability (Baernholdt, 2012; Jaffe, 2015). Higher poverty levels in rural areas may influence both formal (financial support) and informal (volunteers) community resources available for rural hospitals (AHA, 2011). Hospitals support broader social community needs, (e.g., meal delivery services, community health education), and thus rural facilities have fewer financial resources toward organizational high quality care initiatives and safety prevention efforts. Finally, the higher proportion of residents age 65 and older and higher poverty levels in rural areas may contribute to populations with more comorbidities and higher severity of illness (Jaffe, 2015). Combined, these factors suggest that the rural community context has important implications for prevention of HACs in rural hospitals.

Hospital Characteristics
Hospital characteristics pertinent to rural/urban comparisons of HACs include number of beds; Critical Access Hospital (CAH) status; and accreditation by The Joint Commission (TJC) (Baernholdt, Malpass, Hinton, Yan, & Bratt, 2014; Newhouse, Morlock, Pronovost, & Sproat, 2011). Most rural hospitals have less than 100 beds. About 1,400 of the 2,000 U.S. rural hospitals are designated a CAH, which is a hospital in a federal program that receives cost-based reimbursement. CAHs must have 24 hour emergency care services available; a maximum of 25 acute care and swing beds (i.e., a bed for either acute or skilled nursing facility care); and maintain acute care average length of stay of 96 hours or less (Casey, Moscovice, Hung, & Barton, 2012).

When comparing hospital acquired conditions in CAHs with other rural and urban hospitals, review of the literature revealed mixed results (Joynt, Harris, Orav, & Jha, 2011; Vartak, Ward, & Vaughn, 2010). For example, hospitals in small and isolated rural areas had worse rates for PURs, but when the hospital size as defined by number of beds was included, urban hospitals with less than 100 beds had worse PURs than their rural counterparts (Vartak et al., 2010). Additionally, Newhouse and colleagues (2011) found that larger rural hospitals were more likely to be accredited by TJC. This accreditation is associated with higher levels of patient safety process implementations, including those that prevents HACs. Mixed findings about rural hospital characteristics and association with HACs may be due to the lack of studies that include community or nursing unit characteristics.

Nursing Unit Characteristics
Nursing unit characteristics associated with HACs include type of unit, staffing, professional nursing practice, and nurse outcomes. Nursing unit characteristics associated with HACs include type of unit, staffing, professional nursing practice, and nurse outcomes. Type of unit has been associated with PURs (Blegen, Goode, Spetz, Vaughn, & Park, 2011). Intensive care units had lower PURs compared to general adult nursing units; this difference was partly explained by differences in staffing.

The most studied nursing unit characteristics are staffing and professional nursing practice (Bae, 2011). When attempting to demonstrate the influence of staffing on patient outcomes, more than one staffing variable should be included. The National Quality Forum (n.d.) has endorsed measures that include nursing care hours per patient day and nursing skill mix (defined as percentage of nursing staff that are Registered Nurses (RNs) or proportion of hours provided by RNs per patient day). Both have been associated with HACs. For example, lower nosocomial infection rates, PURs and FRs have been associated with higher all nursing staff hours per patient day (He, Dunton, & Staggs, 2012; Kalisch, Tschannen, & Lee, 2012). More RN hours per patient day were associated with lower PURs and FRs (He et al., 2012; Lake, Shang, Klaus, & Dunton, 2010; Patrician et al., 2011; Unruh & Zhang, 2012).

When attempting to demonstrate the influence of staffing on patient outcomes, more than one staffing variable should be included. Professional nursing practice includes concepts such as autonomy; philosophy emphasizing quality of clinical care; nurse participation in hospital affairs and status; professional development; supportive managers and administration; collaborative relationships with physicians; supportive relationships with peers; workgroup cohesion; procedural justice; organizational constraints; and physical work environment (Djukic, Kovner, Brewer, Fatehi, & Cline, 2012). Many relationships between professional nursing practice and HACs have been suggested but findings are inconclusive, indicating the relationships are complex (Bae, 2011). Although associations between nurse outcomes (e.g., burnout, job satisfaction, intent-to-leave, turnover rates) and HACs have been studied less frequently, lower nosocomial infection rates were associated with higher nurse burnout (Cimiotti, Aiken, Sloane, & Wu, 2012), higher job satisfaction was associated with lower FRs and hospital acquired PURs (Choi, Bergquist-Beringer, & Staggs, 2013; Choi & Boyle, 2013). Nurse turnover rates were linked to higher unit-acquired PURs (Park, Boyle, Bergquist-Beringer, Staggs, & Dunton, 2014).

When comparing nursing unit characteristics between rural and urban hospitals and within rural hospital nursing units the results are also mixed. Studies found no difference in staffing, professional nursing practice, and nurse outcomes between rural and urban nursing units (Baernholdt & Mark, 2009; Havens, Warshawsky, & Vasey, 2012). Examining rural hospitals only, one study found that compared to larger rural hospitals, rural hospitals with less than 25 beds (including CAHs) were linked to better professional practice scores and fewer RNs prepared with a BSN or higher degree (Newhouse et al., 2011). Another found no difference between rural hospitals (Baernholdt et al., 2014). It appears that rural location does affect educational level of RNs, which in turn has been linked to HACs (Blegen, Goode, Park, Vaughn, & Spetz, 2013).

Care Process Outcomes

Studies on care process outcomes are rare, let alone studies that include both HACs and care process outcomes. Working within one hospital setting, Titler, Shever, Kanak, Picone, and Qin (2011) found that both the total number of unique care processes (e.g., restraints and neurological monitoring) and staffing (skill mix) were important independent variables for FRs. Another study examined missed nursing care, which is defined as care processes not done. Missed nursing care has been shown to mediate the relationship between staffing (nurse staff per patient day) and FRs (Kalisch et al., 2012). Finally, Quigley, Hahm, and colleagues (2016) found that implementing custom fall prevention programs tailored to a specific nursing unit gap analysis improved multicomponent fall prevention program efforts on units, but no consistent change in FRs were noted. However, participating units already had fall rates below the national average.

Our study compared rural and urban hospital and nursing unit characteristics, care processes, and HACs, specifically PURs and FRs. No study was found that included community characteristics. Rural and urban hospitals are different and serve different communities. Whether rural and urban nursing units also have differences in other contextual factors, care processes and HACs are important to examine to plan targeted, hospital level interventions to prevent HACs. Our study compared rural and urban hospital and nursing unit characteristics, care processes, and HACs, specifically PURs and FRs.

Study Methods

Sample
Several studies have confirmed the validity and reliability of NDNQI® data. We extracted variables of interest using the National Database of Nursing Quality Indicators (NDNQI®) 2009 data, which contains data for more than 1,500 hospitals with nursing unit level data. The NDNQI® collects monthly nurse staffing and patient outcomes data and annual RN survey data from eligible units within participating hospitals. Data can then be aggregated to quarterly or annual rates. Unit-based nurses participate in the annual RN survey, and each unit selects the month for survey participation. Several studies have confirmed the validity and reliability of NDNQI® data (Garrard, Boyle, Simon, Dunton, & Gajewski, 2016; Klaus, Dunton, Gajewski, & Potter, 2013; Simon, Klaus, Gajewski, & Dunton, 2013; Waugh & Bergquist-Beringer, 2016).

For this study, we included units with at least five RN survey respondents. Quarterly outcome data were matched to RN survey participation, so that PURs and FRs data from the same quarter as the RN survey data were included. If PURs and FRs data were missing, then data from the first available quarter after the RN survey were used. A total of 598 hospitals (4,238 units) were included in the study. All variables were aggregated to the nursing unit level except for hospital characteristics. Since not all nursing units measured the practice environment using the Practice Environment Scale (PES), comparisons for that concept have a sample size of 2,103 nursing units (2,011 urban and 92 rural).

Variables

Geographic designation. Geographic location was defined according to U.S. Census Bureau (2013). Urban included metropolitan (a core urban area of 50,000 or more population) and micropolitan (an urban core population of at least 10,000, but less than 50,000) statistical areas. All other areas were designated rural.

Hospital characteristics. Our study included hospital characteristics by size. Hospital size was divided into three categories: less than 100 beds; 100-199 beds; and 200 or more beds.

Nursing unit characteristics. Nurse staffing was measured in two ways: nursing care hours per patient day (i.e., total number of hours worked by RNs, Licensed Practical Nurses and Licensed Vocational Nurses and unlicensed assistive personnel per patient day) and skill-mix (i.e., percent of total nursing hours provided by RNs). Education was measured as percent of RNs who held a degree of BSN or higher. Experience was measured as percentage of RNs with less than two years and percentage of RNs with more than ten years in practice in practice.

The practice environment was measured using the 31 item PES consisting of five subscales and a composite score, as follows:

1. Nurse Participation in Hospital Affairs (9 items), or opportunities for staff nurses to participate in hospital and nursing committees and hospital policy decisions;
2. Nursing Foundation for Quality of Care (9 items), or the hospital quality system and nurses continuing education programs for development;
3. Nurse Manager Ability, Leadership, and Support (5 items), or nurse manager support of nurses in practice;
4. Staffing and Resource Adequacy (4 items), or whether units have enough nursing staff to provide quality patient care; and
5. Collegial Nurse-Physician Relations (3 items), or working relationships between physicians and nurses.

Subscales 1 and 2 reflect the environment at the hospital level, whereas subscales 3-5 reflect the environment at the nursing unit level (Lake, 2002). All 31 items were rated from strongly disagree to strongly agree (1 to 4).

Two nurse outcomes were collected: intent to leave and job satisfaction. Intent to leave was measured as percentage of RNs who planned to leave their current position within the next year. Job satisfaction was measured as the degree to which people like their work. This seven item Likert-type scale was scored from strongly agree to strongly disagree (1 to 6) where higher score represent higher satisfaction.

Care process outcomes. Our study included three care process outcomes: percentage of patients at risk for pressure ulcer which were all patients on the unit whether they had developed a pressure ulcer or not; percentage of patients who fell who were assessed for fall risk, and percentage of patients who fell who were at risk.

Hospital acquired conditions (HACs). We included three HACs. Pressure ulcer rates were measured in two ways: percentage of assessed patients who had a hospital acquired pressure ulcer or a unit acquired pressure ulcer. Pressure ulcers rates were collected as a prevalence rate once a quarter. Whether a pressure ulcer was hospital or unit acquired was determined by chart review. Fall rates included all patients who fell in a quarter and this was calculated as the number of falls per 1,000 patient days. Repeat falls are included in the fall rate. A fall is defined in NDNQI® data as an unplanned descent to the floor with or without injury to the patient. The fall can be unassisted or assisted, as when a staff member attempts to minimize the impact of the fall. (NDNQI®, 2010)

Analyses
We performed statistical tests to examine rural and urban differences for all variables considered. We used Chi-square tests for the categorical variables (i.e., hospital size, unit type), and two-sample t tests for six variables measuring practice environment and job satisfaction. Since all other variables were either in percentages or rates, we used Wilcoxon rank-sum tests, which are the robust test for non-normal data.

Results

Several findings were statistically significant. There were proportionally more hospitals with fewer number of beds in the rural sample (p<.005) (see Table). Among nursing unit characteristics, unit type had similar proportions across the three types of units in rural and urban hospitals. Compared to urban nursing units, rural units had significantly higher nursing care hours per patient day (10.9 hours versus 11.6 hours); lower skill mix (.73 versus .69); and lower percentage of RNs with a bachelors or higher degree (52.4% versus 33.3%).

Table. Comparison of Urban and Rural Nursing Units

***p<.005; ** p<.05; *p<.1;

^a For practice environment results N=2103 with 2011(95.6%) urban and 92(4.4%) rural nursing units respectively.

Overall, compared to their urban colleagues, nurses in rural nursing units rated their practice environment lower with significant differences for the composite score and three subscales: nursing foundations for quality of care (3.08 versus 3.04); nurse manager ability, leadership, and support of nurses (2.93 versus 2.85); and collegial nurse-physician relations (2.95 versus 2.86). Compared to urban nurses, rural nurses were significantly less satisfied with their jobs (3.72 versus 3.66), but rural nurses reported significantly less intent to leave their current positions compared to their urban colleagues (22.9 versus 19.8).

For care processes, percentages of patients deemed at risk for pressure ulcers were similar in rural and urban nursing units (20.4 versus 19.1). Patients who fell in rural units were significantly more likely to have had a risk assessment (83.5 urban versus 94.0 rural) and also significantly more likely to have been deemed at risk for falls (82.9% versus 64.9%). Patients in rural nursing units were significantly less likely to develop a hospital acquired pressure ulcer (4.7% versus 3.3%).

Discussion

Findings and Implications
...a cost saving strategy should include learning how to prevent and treat pressure ulcers from organizations with lower rates. Not surprisingly, we found differences between rural and urban hospitals and nursing unit characteristics, care processes, and HACs. We found no differences in fall rates between rural and urban nursing units, which confirms previous results (Ma & Park, 2015). We also found no difference between rural and urban nursing units’ acquired pressure ulcers rates, but the hospital acquired pressure ulcer rate was almost 30% lower in rural hospitals. Considering our study found that percentages of patients deemed at risk for pressure ulcers were similar in rural and urban units, rural hospitals may have done a better job preventing hospital acquired pressure ulcers. Given that pressure ulcer incidence rates vary considerably in acute care settings with a range between 0.4 to 38 %, and the cost of treating a single full-thickness pressure ulcer is as high as $70,000 (American Society for Healthcare Risk Management [ASHRM], 2015), a cost saving strategy should include learning how to prevent and treat pressure ulcers from organizations with lower rates.

Falls are associated with increased lengths-of-stay, increased utilization of healthcare resources, and poorer health outcomes (The National Quality Measures Clearinghouse [NQMC], 2012). Our fall rates were within the range of 1.3 to 8.9 falls/1,000 patient days reported in observational studies (NQMC, 2012). Compared to urban nursing units, about 10% more patients who fell in rural units had been assessed for fall risk prior to a fall. However, in rural nursing units, 18% more patients who fell were deemed at risk. Why more patients who fell in rural units had been identified as being at fall risk and still fell is unknown. Perhaps the rural nursing units did not have the resources to prevent falls. As with PURs, much can be gained from studying institutions with lower FRs.

Falls are associated with increased lengths-of-stay, increased utilization of healthcare resources, and poorer health outcomes. Rural nurses rated their practice environment lower than their urban colleagues. This is different from previous studies that found no differences in ratings of the practice environment among rural and urban nurses (Baernholdt & Mark, 2009; Havens et al., 2012). Rural nurses’ ratings of practice environment were similar to previous studies (Cline et al., 2014). In both rural and urban settings, ratings of practice environment were high. PES values above 2.5 indicate high agreement that a good practice environment is present and mean scores above 2.5 for 4-5 subscales indicate a favorable practice environment (Lake & Friese, 2006). The highest ranked practice environments are among nurses in Magnet® hospitals. The nurses in our study did not rank any of the subscales or the composite score as high as nurses in Magnet® hospitals (Lake & Friese, 2006; Ma & Park, 2015). However, in our study the practice environment was favorable in both rural and urban nursing units.

Similar to practice environment, rural nurses reported lower job satisfaction than urban nurses, but rural nurses had lower rates of intent to leave in the next year. The lower job satisfaction scores for rural nurses are different from previous findings where urban and rural nurses scored similarly (Baernholdt & Mark, 2009) or rural nurses scored higher (Bratt, Baernholdt, & Pruszynski, 2014). The combination of lower job satisfaction and lower rates of intent to leave suggest it may be harder for rural nurses to find a comparable position in their community. On the other hand, once a rural nurse leaves, hospitals often have difficulties finding a replacement. Younger rural residents migrate to urban areas or rural nurses increasingly choose to commute for jobs in urban areas (Dotson, Dave, & Cazier, 2012; Skillman, Palazzo, Hart, & Butterfield, 2007). All nurse managers need to be mindful of the best strategies to recruit and retain nurses, but this may be particularly true for nurse managers in rural areas. Strategies may include recognizing the uniqueness of rural nursing practice and using that to attract nurses who seek the greater autonomy and higher levels of interconnectedness between nurses’ professional and personal lives found in rural areas (Baernholdt, Jennings, Merwin, & Thornlow, 2010; Dotson et al., 2012).

The highest ranked practice environments are among nurses in Magnet® hospitals. Another way to increase job satisfaction and retention of rural nurses is to support continuing education opportunities, including RN-BSN programs and nurse residency programs, despite logistical challenges of fewer resources and greater distance from educational programs (Bratt et al., 2014). Supporting continuing education to enable more RNs to earn a BSN degree could also increase quality and safety. As the IOM noted (2010), the growing complexity of care requires a more educated workforce and some evidence has suggested that higher education of nurses is associated with better patient outcomes. While the last five years has demonstrated an increase in RN-BSN programs and enrollment, the increase in BSN prepared nurses rose only by 2%, from 49% in 2010 to 51% in 2014 (Altman, Butler, & Shern, 2016). Clearly, more support for BSN level education is warranted.

All nurse managers need to be mindful of the best strategies to recruit and retain nurses, but this may be particularly true for nurse managers in rural areas. It is noteworthy that, despite the rural nurses’ lower ratings of practice environment and job satisfaction, the care process outcomes and HACs rates were not worse and were sometimes even better compared to urban rates. While we only compared rates and not associations, the contradictions between the rural nurses lower ratings and still better performance warrants further examination. Previous studies have described rural nurses’ deep connection to their communities, colleagues, and patients as an important factor for providing high quality care in spite of any obstacles (Baernholdt et al., 2010; Dotson et al., 2012). While this is commendable and rural nurses should be praised for their performance, they should also be provided resources known to improve work environment and satisfaction, such as continuing education opportunities as described above (Bratt et al., 2014).

Limitations
...the growing complexity of care requires a more educated workforce and some evidence has suggested that higher education of nurses is associated with better patient outcomes. Our study had limitations. We used a convenience sample of hospitals and nursing units which voluntarily subscribe to NDNQI® with the goal of improving nursing characteristics, care processes, and outcomes. These hospitals differ from other hospitals. For example for staffing in 2004, NDNQI® hospitals had nearly two hours higher RN hours per patient day than U.S. general hospitals (Lake et al., 2010). While the data is of high quality with robust procedures in place to minimize differences in collection, analyses, and aggregation, it is possible that missing data skewed our results. However, since NDNQI® data are unrelated to coding for Medicare reimbursement, the data do not carry the same risk of bias as other administrative databases (Waters et al., 2015). Additionally, we are reporting raw rates that are not adjusted for patient characteristics. Other data not included are nursing shift level variables which, in at least one study, found an association between staffing on a shift and falls (Patrician et al., 2011).

Finally, as with any research examining rural-urban comparisons, our results reflect the definition that was available in the dataset which was rural/urban delineation at the county level. Albeit a commonly utilized federal definition in research comparing rural and urban areas, it is not as granular as using Rural-Urban Commuting Areas (RUCAs) zip-code approximation which may have produced different results (Cromartie & Bucholtz, 2008). Nevertheless our study did produce differences between rural and urban settings in nursing unit characteristics and care process outcomes that were associated with HACs in previous studies.

Conclusion

With the largest and growing component of healthcare cost being inpatient hospital care (31% of the U.S. healthcare budget) the need to decrease HACs and lower cost is of utmost importance (Mayne, Girod, & Weltz, 2013). The most frequent and costly HACs are falls and pressure ulcers. Total costs for treatment of pressure ulcers in the U.S. is estimated at $11 billion annually (ASHRM, 2015). While other studies have found that tailored prevention fall programs will increase uptake of specific program components and processes (Quigley, Barnett et al., 2016), the differences in our study between urban and rural nursing unit pressure ulcer rates, staffing, education, practice environment, and nurse outcomes suggests that further study leading to improvement strategies may need to be tailored differently in urban and rural hospitals.

Acknowledgement: This research was supported by the Agency for Health Care Research and Quality R01 HS023147.

Authors

Marianne Baernholdt, PhD, MPH, RN, FAAN
Email: mbaernholdt@vcu.edut

Marianne Baernholdt is Director of the Langston Center for Quality, Safety, and Innovation, and the Nursing Alumni Endowed Distinguished Professor at Virginia Commonwealth University School of Nursing.

Ivora D. Hinton, PhD
Email: idh2r@eservices.virginia.edu

Ivora D. Hinton is Coordinator of Data Analyses and Interpretations at University of Virginia School of Nursing.

Guofen Yan, PhD
Email: gy4g@eservices.virginia.edu

Guofen Yan is an Associate Professor at University of Virginia, Department of Public Health Sciences.

Wenjun Xin, MS
Email: wx4q@eservices.virginia.edu

Wenjun Xin is a biostatistician at University of Virginia, Department of Public Health Sciences.

Emily Cramer, PhD
Email: ecramer2@kumc.edu

Emily Cramer is a Research Assistant Professor at University of Kansas School of Nursing.

Nancy Dunton, PhD, FAAN
Email: ndunton@kumc.edu

Nancy Dunton is a Research Professor at University of Kansas School of Nursing.