EVERY day nurse managers across the nation are challenged to staff nursing units in a manner that minimizes costs yet promotes quality. The pressure to achieve these two dichotomous objectives is not new but is becoming more intense. Emphasis on the cost and quality connection gained national attention with publication of the Institute of Medicine report, To Err is Human, and was intensified by implementation of new Medicare payment guidelines denying payments for hospital acquired conditions.1,2

McCue et al3 report that labor costs are over half of the hospital's operating expenses, and nurses comprise about one-third of hospital personnel. As the largest single group of hospital employees, nurses are often the focus of reductions when finances are constrained. Adding to the financial concern is the pressure to legislate nurse-staffing ratios. Spetz4 argues that additional economic research is needed on nurse staffing and patient care as research to date is inconclusive and for the most part is ignored by nonnursing decision-makers.

The 2004 Institute of Medicine report, Keeping Patients Safe: Transforming the Work Environment of Nurses, concluded that the work environment of nurses contains threats to the safety of patients. Final recommendations included employing nurse staffing practices that identify needed nurse staffing for each patient unit each shift.5 The reality is that these staffing practices are often staffing plans that indicate the number and mix of nurses based only on patient census. Critical factors such as acuity and age of patients, required intensity of nursing services, and experience and qualifications of staff are not considered in the plans.

Although reducing the numbers or mix of nursing staff is a strategy to control costs, such action is questionable if quality care is also a priority. Experienced nurse managers possess insights into the relationship between the numbers and characteristics of nursing staff and patient outcomes. The experienced manager can intuitively predict the outcomes of staffing levels or skill mix; however, this decision-making process has disadvantages. It takes time to develop expertise, and decisions based on intuition are difficult to justify.

The use of technology innovations can promote effective use of evidence and facilitate improved nurse staffing plans that incorporate aspects of all relevant factors.6 Douglas7 argues that while the evidence exists in relation to nurse staffing, it is not operational. Data to support evidence-based decision-making must be available wherever the decision is made. In addition to just having the data, nurse leaders need objective, accurate decision-making tools linking the economics of nurse staffing with clinical quality.

ESTABLISHING THE LINK BETWEEN NURSE STAFFING AND OUTCOMES

Nurse staffing and patient outcomes

Health care environments are complex with varying cultures and different needs and expectations of both the caregivers and the patients. Such complexity makes establishing definitive causal linkages difficult. However, a number of studies over the past decade have clearly established relationships between nurse staffing and patient outcomes.8-23 Landmark studies by Needleman et al20 and Aiken et al9 provided the initial research. Needleman and colleagues examined state level data from 799 hospitals in 11 states. Using regression analysis and controlling for such factors as risk of adverse outcomes and differences in nursing care needed, results indicated that a higher number of hours of care provided by registered nurses (RNs) were associated with shorter lengths of stay and lower levels of negative patient outcomes such as urinary tract infections, pneumonia, failure to rescue.20 Aiken and colleagues9 examined hospital-level data from Pennsylvania and linked those data to nurse self-reports of the number of assigned patients on their last shift. They found that patient mortality was related to nurse staffing; every additional patient assigned to a nurse would be expected to increase the odds of mortality by 7%. The researchers found a similar relationship between nurse staffing and failure to rescue. For each additional patient, failure to rescue would be expected to increase 7%.9

Follow-up studies have continued to build the case for a relationship between higher nurse staffing levels and positive patient outcomes. In 2007, Kane et al17 published a meta-analysis of 96 studies for the Agency for Health care Research and Quality on nurse staffing and outcomes. The team found that risk adjusted mortality, failure to rescue, hospital-acquired pneumonia, and other adverse events were associated with higher numbers of patients assigned to RNs. They reported that the effect was strongest for patients in intensive care unit (ICU) and those who had surgery. Similar results were found for RN full time equivalent (FTE) per patient day. Indeed, for every additional RN FTE, the risk of mortality was reduced by 9% in ICU patients and 16% in surgical patients. The effect of licensed practical nurses was inconsistent across the studies. Clarke and Donaldson13 reviewed nurse staffing research from 2002 to 2007. They found a similar pattern as described by Kane et al and stated the "critical mass of studies established that nurse staffing is one of a number of variables worthy of attention in safety practice and research" (p.13).

Staffing and financial outcomes

The linkage between nurse staffing and financial outcomes has been less well established. McCue et al3 explored whether changes in nurse staffing and quality of care impacted financial performance (operating costs and profit margin). Using an econometric model and longitudinal data from 422 hospitals whose data were contained in the Federal government's Healthcare Cost and Utilization Project (HCUP), they found that increasing the nursing staff was associated with greater operating costs. However, the researchers found no significant impact of nurse staffing on profit margins. In other words, even though operating expenses grew with increased staffing levels, hospitals did not have a reduction in profits. The authors propose that this finding may be due to turnover costs and overtime use that is greater in hospitals with low staffing.

Other researchers have approached the study of nurse staffing and costs in a different manner: examining costs through adverse event avoidance and reductions in length of stay.12,24-27 Cho et al12 provided some of the first evidence of the connection between staffing, quality of care, and financial outcomes. They conducted multilevel analysis of over 120,000 surgical patients from 232 California hospitals to determine predictive relationships between nurse staffing and seven adverse events, length of stay, and costs. They found significant relationships between staffing and 2 adverse events: pneumonia and pressure ulcers. Cho and colleagues also found that all 7 adverse events in the study were associated with longer hospital stays and with higher costs. For example, patients who developed pneumonia were hospitalized for two additional days, had a 5% increased chance of death and had costs of $22,000 to $28,000 higher than patients without pneumonia.

Findings from a study by Pappas26 support the value of cost avoidance through a reduction in adverse events. She studied 3200 patient cases that had 1 of 3 selected diagnostic-related groups to develop a reliable methodology for costing adverse events from administrative data. Pappas26 selected five adverse events for inclusion in her study: medication error, fall, urinary tract infection, pneumonia, and pressure ulcer. She also examined the direct cost per case, which was available in the accounting system at the two hospitals where the study was conducted. Pappas26 found that adverse events, depending on type, cost $300 to $2400 per patient day; however, the cost of an increase of 1 hour of RN care per patient day cost $659.

Rothberg et al27 conducted a cost-effectiveness analysis of nurse staffing and patient mortality. Using data from the Bureau of Labor Statistics, medical literature, and data from 2 large hospital studies, they compared the labor costs of a patient-to-nurse ratio of 8:1 and 4:1. Rothberg et al27 found that labor costs were higher with 4:1 ratios than 8:1; however, more deaths occurred with the 8:1 ratio. They estimated that labor costs were $64,000 per life saved when dropping the number of patients per nurse from 7:1 to 6:1. When compared to other "life-saving" interventions (such as thrombolytics for myocardial infarction at $182,000), nurse staffing is a cost-effective intervention that should be part of the financial strategic planning for acute care hospitals.27

Using a macro-economic approach, Dall et al24 examined the value of three nurse staffing levels (6.4, 7.8, and 9.1 hours per patient day(HPPD)) on medical costs, national productivity, and lives saved. They found positive impacts of nurse staffing on patient outcomes: as staffing increases, nosocomial infections decrease and inpatient length of stay decreases. The authors stated that at an average staffing level (7.8 HPPD), the benefit per RN is a savings of $60,000 in medical costs. These savings result from preventing nurse-sensitive adverse events and from reduced length of stay of patients who do not experience an adverse event.24

Clearly, the findings from research link nurse staffing to patient and organizational outcomes and indicate that nurse staffing is a valuable life-saving intervention in the treatment of hospitalized patients. Just as leaders in any acute care hospital purchase medical equipment to provide the latest technology for physicians to use with their patients, hospital boards must strategically fund the most appropriate nurse staffing levels for patients and provide the latest technology advances so that staffing decisions can be made with the most accurate information.

DEVELOPING AN EVIDENCED BASED STAFFING TOOL

Data derived from information systems in the form of scorecards and dashboards have increasingly been used in health care to improve performance. The emphasis on metrics dominates performance improvement activities and a variety of scorecards and dashboards have been incorporated into monetary reward systems.28 Frith et al29 examined various sources and characteristics of data that can be used to assist nurse leaders in decision-making. They stress that data from multiple sources with common definitions and established linkages among data should be presented in a format that is easily used for decision-making. This premise defines the work of a unique practice and academic collaboration established in 2007.

An informal discussion following a graduate student conference led to a dynamic partnership between Catholic Health Initiatives (CHI), a 78-acute care hospital system and the University of Alabama in Huntsville (UAHuntsville), a research intensive university. The focus of the collaborative partnership is to make current, relevant data available to nurse managers at the unit level and integrate evidence on staffing with patient and organizational outcomes. The multiple databases of the large hospital system provide a rich source of data that can be used to expand research studies related to nurse staffing and patient outcomes. The ultimate goal of the collaboration is the development of a nursing services dashboard based on empirical evidence linking critical patient and organizational outcomes with nurse staffing to facilitate evidence-based staffing decisions by nurse managers.

The collaboration is anchored by a large interdisciplinary team. Contributing members from CHI include executive nurse leaders in the CHI hospitals and CHI system leaders and data managers from business intelligence, finance, human resources, performance management, strategy, quality and risk. Nurse researchers, statisticians, business operations faculty, and computer modeling and simulation experts are the contributing team members from the university. This article describes the challenges and implementation issues in developing a system dashboard that informs nursing leaders about the expected organizational and patient outcomes of their staffing decisions.

Phase one

The first phase of the collaboration was research on nurse staffing and patient outcomes. As discussed earlier, much of the previous research on staffing and patient outcomes was conducted at the hospital level using state or national level databases. The intent of this collaborative research was to build on the strengths of previous research while minimizing identified limitations.13,30-32 The practice and academic collaboration afforded an opportunity to explore the relationship of staffing on patient and hospital outcomes using a multilevel approach as suggested by Cho30 and Lake.33,34 Three levels of data were available-patient, unit, and hospital data-from over 70 hospitals located across the nation. Since previous research often focused on surgical units, this research would target general medical or medical/surgical units.8,12,35-37 In addition, the financial impact of nurse staffing decisions would be included.

Prior to beginning the research, input was solicited from chief nurse executives in the CHI hospitals. These leaders were queried about the factors that they viewed impacted patient outcomes. From these discussions and the review of literature, patient, unit, and nurse variables were identified. These included patient acuity, adverse patient outcomes, patient satisfaction, numbers of staff, numbers of registered nurses, characteristics of nursing staff (education, experience), and the nurse-patient ratio.

The next step was to identify appropriate CHI databases that contained key data elements critical to the research process. This was accomplished through face-to-face and web meetings with representatives from across the system. The endeavor was a challenge because data were stored in a variety of system databases (financial, human resources, operations, and quality) collected and refreshed at different times (yearly, quarterly, and monthly), distributed for analysis and action at different times, and incorporated different terminology for common items such as the names of patient units.

In spite of the disparate data sources, a pilot research proposal was developed and conducted.15 A quantitative, cross-sectional research design was used to investigate the predictive relationship between nurse staffing and patient outcomes in acute care hospitals. Data of more than 34,000 patients from 11 medical-surgical units in 4 hospitals over a 2-year period were obtained from CHI system databases. The data were analyzed using hierarchical linear modeling. Findings of the pilot study confirmed previous research that staffing levels are predictive of patient outcomes.15 The percentage of RN staff in the skill mix was predictive of adverse patient outcomes. For every 1% increase in RNs in the skill mix, a 3.38% decrease in adverse patient events was noted. Both RN and Licensed Practical Nurse (LPN) hours per equivalent patient day and increases for both in the skill mix were predictive of patient length of stay (LOS), although the impact of RN hours was greater than LPN hours. For example, an increase of 1 RN hour per patient day was predictive of a 16.54% decrease in LOS for patients with a median complication index, and 1-hour increase in LPN hours per patient day predicted a 5.67% decrease in LOS. The findings of the pilot study were reported to the CHI executive nursing and multidisciplinary leaders and served to confirm the value of the project and to stimulate commitment.

Phase two

The focus in the current phase is two-fold. First, the research on nurse staffing and patient outcomes continues. Although the pilot study results are significant, the goal is to expand the study to include more patients and units to further establish the impact of nurse staffing on patient outcomes and verify the relationships identified in the pilot study. Additional variables related to nurse characteristics and financial outcomes are included in this phase of the study.

The development of the nursing services dashboard is the new focus in phase 2. The purpose of the dashboard is to facilitate evidence-based staffing decision-making by displaying information about the hospital and the unit in an organized fashion. For this phase, experts with expertise in computer software design and implementation from the UA Huntsville Center for Modeling, Simulation, and Analysis have been added to the team.

Data to be displayed in the dashboard include census, patient days, case mix index, complication index, average mortality rate, staff worked hours, nurse/patient ratios, rates of hospital acquired conditions, and financial indicators. Dashboard data will be updated regularly so that nurse managers have critical and timely information for decision-making. The vision for this project is that once the dashboard is fully functional, the nurse manager will have access to current operational and financial data. Data that were available days, weeks, or months after the fact will be available in current time. The innovative predictive feature of the dashboard will link staffing decisions with patient outcomes, the potential for patient harm and the financial implications of the decisions.

Predictive equations based on the staffing and patient outcomes research will enable the manager to develop "what-if" staffing scenarios and receive feedback on patient and financial outcomes associated with each scenario. Managers will be able to forecast, based on evidence from research with CHI's own patient populations, the impact of various staffing patterns on patient outcomes and costs rather than staffing to keep costs down on a shift-by-shift basis. Managers will be able to predict the "true" costs associated with staffing decisions and their impact to the bottom line. Managers will be able to staff to achieve optimal patient and financial outcomes and will be able to make staffing decisions that incorporate findings from research and reduce the risk and costs of an adverse event occurring due to poor staffing.

In these times of financial restraint, the CHI system, like hospitals across the nation, expect managers to maintain budgeted staffing levels-often based on hours per patient day. Experienced nurses know that many factors impact the need for nursing care and know that inadequate staff levels can lead to errors, delays, and missed care. Inadequate staffing levels are also correlated with nursing turnover and poor patient satisfaction. These costs and the negative consequences of poor staffing will increase the cost of care even though budgeted staffing goals are met.

The dashboard will provide the CHI nurse manager with an objective alternative to that innate sense developed through experience. Increasing staffing levels will not occur if budgetary standards are the major consideration, but increasing staff may be the most appropriate decision if staffing levels present identifiable risks and costs to maintaining clinical quality. Use of the dashboard will provide CHI's nurse managers both quality and financial outcomes decision support.

Challenges

The challenges of this initiative have been great, but none has been so overwhelming that the project has been abandoned. The initial challenge was identification of all the data elements needed for both the staffing research and the dashboard. As previously noted, data are collected, analyzed, reported, and stored in various system administrative databases. It was critical to identify data elements and data sources needed for both the research and the dashboard. Once identified, the various data sources had to be configured to feed into an integrated database that could be used for decision support. This effort continues to require many hours of discussion and planning.

Disparate data systems are a common occurrence and not unique to CHI. Most organizations encounter the same challenges identified in this project when seeking to integrate data. Databases are designed and implemented to meet specific functional and departmental needs. The various disciplines and service area personnel often use different terminology and do not fully understand how data need to be collected and reported for maximum usefulness. For example, human resource databases can include inaccurate nurse credentials, and financial reports are often reported at month's end whereas decisions about staffing are made daily. Unfortunately, the lack of understanding of how and when to report data impedes decision-making.

A second challenge was maintaining the interest and enthusiasm of hospital nurse executive leaders. The project required many months of planning and investigation; in the meantime, nurse leaders moved forward, challenged to meet day-to-day quality and financial goals. To generate and maintain nurse executive involvement, the CHI system and UAHuntsville team made presentations at regular meetings of the hospital chief nurse executives. Initially, findings from the pilot study were presented. A second presentation focused on the use of data in a dashboard format to facilitate decision-making. While these presentations reported preliminary work, seeing actual results served to invigorate and motivate the nurse leaders.

The key to success in meeting the challenges is commitment from both organizations; particularly useful is having a designated primary coordinator at each organization. The liaison at CHI has access to key individuals at the corporate level and has the support of the system Chief Nursing Officer. The CHI liaison facilitates the critical discussions with hospital nurse leaders as well as with corporate leaders who are instrumental in bringing the data elements together. The UAHuntsville principal investigator is responsible for coordinating activities at the university.

Substantial financial commitments have been made by both organizations. Internal grants from the university fund critical aspects of the statistical analysis and dashboard development. Support from CHI includes the time and effort of the designated coordinator as well as assistance from various system departments in identifying and forwarding essential data elements.

CONCLUSION

Needleman et al38 propose that the business case and the social case for levels of nurse staffing need to be examined. He asserts that increased staffing results in better patient outcomes (the social case) but increases costs (business case); thus a constant tension is created between the patient caregivers and the fiduciary guardians. Needleman et al38 argue that a change in reimbursement is needed. The assertion is sound and findings from staffing research may eventually result in such a change; but reimbursement rewarding staffing levels that achieve quality outcomes is not an immediate reality. In the meantime, healthcare leaders must work within the current system. The CHI/UAHuntsville collaboration is seeking to do that by providing nurse managers with a decision-support tool designed to facilitate the use of staffing research evidence in combination with nurse, patient, and organizational data.

REFERENCES