The Association of Shift-Level Nurse Staffing With ... - Semantic Scholar

4 downloads 15402 Views 184KB Size Report
Feb 2, 2011 - linking nurse staffing and patient outcomes, the relationship of staffing to patient ... Service (Dr McCarthy), Madigan Army Medical Center, Tacoma,. Washington .... person and by conference call to verify data entry and coding ...
JONA Volume 41, Number 2, pp 64-70 Copyright B 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins

THE JOURNAL OF NURSING ADMINISTRATION

The Association of Shift-Level Nurse Staffing With Adverse Patient Events Patricia A. Patrician, PhD, RN, FAAN Lori Loan, PhD, RNC Mary McCarthy, PhD, RN Moshe Fridman, PhD

Nancy Donaldson, DNSc, RN, FAAN Mona Bingham, PhD, RN Laura R. Brosch, PhD, RN

Objective: The objective of this study was to demonstrate the association between nurse staffing and adverse events at the shift level. Background: Despite a growing body of research linking nurse staffing and patient outcomes, the relationship of staffing to patient falls and medication errors remains equivocal, possibly due to dependence on aggregated data. Methods: Thirteen military hospitals participated in creating a longitudinal nursing outcomes database to monitor nurse staffing, patient falls and

medication errors, and other outcomes. Unit types were analyzed separately to stratify patient and nurse staffing characteristics. Bayesian hierarchical logistic regression modeling was used to examine associations between staffing and adverse events. Results: RN skill mix, total nursing care hours, and experience, measured by a proxy variable, were associated with shift-level adverse events. Conclusions: Consideration must be given to nurse staffing and experience levels on every shift.

Authors’ Affiliations: Associate Professor and Donna Brown Banton Endowed Professor (Dr Patrician), University of Alabama at Birmingham, School of Nursing; Nursing Research Consultant to the Army Surgeon General (Dr Loan), Chief, Nursing Research Service (Dr McCarthy), Madigan Army Medical Center, Tacoma, Washington; Statistician (Dr Fridman), AMF Consulting, Inc, Los Angeles, California; Clinical Professor, Director, Center for Research and Innovation in Patient Care, and Associate Dean for Practice (Dr Donaldson), University of California, San Francisco, School of Nursing; Chief, Nursing Research Service (Dr Bingham), Brooke Army Medical Center, Ft Sam Houston, Texas; Director, Office of Research Protections (ORP) and ORP Human Research Protection Office (Dr Brosch), Headquarters, US Army Medical Research and Materiel Command, Fort Detrick, Frederick, Maryland. Corresponding author: Dr Patrician, University of Alabama at Birmingham, NB324, 1530 3rd Ave S, Birmingham, AL 35294-1210 ([email protected]). Funding: This project was funded by the TriService Nursing Research Program, Uniformed Services University of the Health Sciences (grant N03-P07); however, the information or content and conclusions do not necessarily represent the official position or policy of, nor should any official endorsement be inferred by, the TriService Nursing Research Program, Uniformed Services University of the Health Sciences, the Department of Defense, or the US Government. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jonajournal.com). DOI: 10.1097/NNA.0b013e31820594bf

64

Although the preponderance of research linking nurse staffing to patient safety is compelling,1-3 there are limitations in the available evidence.3-5 A comprehensive review of studies published before 2004 showed that ample evidence exists of an inverse relationship between nurse staffing and both failure to rescue and inpatient mortality.2 Authors of the review cite Bimprobable[ associations between nurse staffing and 15 other patient outcomes, including pressure ulcers, patient falls, and treatment errors,2(p329) and mixed evidence for relationships between nurse staffing and illnesses such as pneumonia and urinary tract infections. Despite equivocal evidence, nurse-sensitive quality measures are increasingly important in healthcare public reporting and pay-for-performance. Furthermore, regulatory interventions such as California’s landmark 1999 legislation mandating minimum nurse staffing ratios have occurred in the absence of empirical data to prescribe staffing levels that ensure patient safety.2-4,6 The ambiguity in the staffing-outcomes literature may be a result of varied sources of available data, different levels of analysis, and other methodological

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

challenges, including model specification, risk adjustment, and control variables.4,5,7 Two frequently used existing administrative sources for staffing data are the American Hospital Association (AHA) and hospital payroll data; however, both have limits. The AHA data do not differentiate inpatient from outpatient nurses; hospital payroll data do not differentiate direct care hours from those unrelated to direct care, for example, meetings. Furthermore, data aggregated over time (eg, quarters or years) or place (eg, unit or hospital) do not reflect daily variability or, more importantly, shift variability; aggregation Bsmoothes out[ high and low spikes in staffing. Although research using aggregated staffing has shown significant associations with mortality and some complications,2,3 this is more a reflection of chronically understaffed units and hospitals4; lower staffing levels on some shifts will disappear when aggregated to the unit level. Likewise, the effects of lower-staffed units will be masked when staffing is aggregated to the hospital level. These limitations may confound the ability to attribute discrete events to specific nurse staffing indicators in place when the event occurs. To address these methodological concerns, we conducted a longitudinal, multihospital study to analyze the relationship of nurse staffing and temporally distinct adverse patient eventsVthose events that happen at a specific time. A basic assumption of this study was that adverse patient events occur in the absence of sufficient staffing on that shift. Because staffing is generally scheduled by shift and adjusted each shift based on workload, the shift was selected as the unit of analysis to investigate associations between staffing and adverse patient events, ie, falls, falls with injury, and medication administration errors (MAEs).

graphic proximity to predesignated study hub sites. See Table 1 for hospital and unit characteristics. A data set of 115,062 consecutive shifts was generated from 2003 through 2006.

Methods

Covariates Correlates of adverse events or staffing include patient census, patient acuity, hospital size, and temporal covariates, that is, shift time, day, and

Sample The data source for this analysis, the Military Nursing Outcomes Database (MilNOD), evolved from a multisite demonstration project conducted to test the feasibility and usefulness of replicating the Collaborative Alliance for Nursing Outcomes (CALNOC) database development in the military health system.8 One of the largest in the nation, the military health system serves more than 9 million active and retired beneficiaries and their families, a population considered to be a microcosm of the larger American society.9 Differences in nurses and patients between military and civilian hospitals are addressed elsewhere.8 The hospital sample (N = 13) was selected based on military affiliation and geo-

Staffing Measures Three staffing measures were used. Total nursing care hours per patient per shift was defined as the sum of hours worked by all nursing personnel during the shift divided by the number of patients present at the beginning of the shift. Skill mix was defined as the proportion of hours worked by each skill level of staff (RN, LPN, and unlicensed provider) during a shift. Both total nursing care hours and skill mix have been widely used in other staffing studies, although with different units of collection, that is, total nursing hours per patient day.1,10 Shift interval was standardized to an 8-hour period. Staff category mix was defined as the proportion of hours worked by each of 4 staff member categories during a shift (ie, active military, Department of Defense [DoD] civilian, contractor, and military reservist) as the military employs 4 distinct categories of staff. Adverse Event Measures A fall was defined as a patient’s unplanned descent to the floor10 and further described as with or without injury, depending on whether an injury was sustained, as documented on the incident report, at the time the fall occurred. MAE was defined as a deviation from the physician’s documented order committed by a nurse.11 These were differentiated from errors due to pharmacy or physicians. The source of the adverse event data was institutional incident reports, which were collected, categorized, and analyzed monthly by a trained on-site research assistant.

Table 1. Sample Characteristics: Unit Breakdown

Unit Type Medical-surgical units Step down Critical care Total

Large Hospital (9100 Beds, n = 7) 22 8 12 46

Small Hospital (G50 Beds, n = 6) Total 9 0 6 11

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

31 8 18 57

65

year. Patient census was recorded each shift. Patient acuity was obtained from a standard acuity system used by the military.12 As standard practice, day-shift nurses tabulated each patient’s acuity rating on a scale from 1 = minimal care to 6 = critical care, based on a list of nursing care activities and a time factor value associated with each. Average acuity for each unit on the day shift was used. Hospital size was defined as small (G50 occupied beds) and large (9100 occupied beds). There were no medium-size facilities. Data Collection After institutional review board approval from each hospital, research team members introduced the study to the major stakeholders at each participating hospital. An intensive educational program, including variable definitions and familiarization with the MilNOD Codebook and trial data collection, ensured staff proficiency with the data collection processes. Because of the wide variety of shift designations in the participating units, staff hours were standardized to an 8-hour shift category. Staffing measures were captured for each: day (7:00 AM to 2:59 PM), evening (3:00 PM to 10:59 PM), and night (11:00 PM to 6:59 AM). At the end of every shift, a designated unit staff member entered hours worked by each provider type and staff category into a standardized Microsoft Access database housed on a unit computer. For 12-hour shifts, work hours were split into the 2 periods that encompassed the shift, typically with 4 hours on one shift and 8 on the other. Data entry personnel were instructed not to count nursing hours spent away from the unit (eg, in a class or as borrowed manpower Bfloated[ to another unit). Similarly, hours worked by nursing personnel on loan to a specific unit or by those providing consultation on that unit, such as a wound care nurse, were counted as direct care hours worked on that unit. For adverse patient events, the institutional incident reports were reviewed by trained on-site nurses, and data including unit, date, time, and patient harm were extracted and merged with the shift of occurrence. A 3-month lag time ensured that all incident reports traversed the system and were available for review. Data Quality Evaluation of reliability and validity of the prospectively collected data was built into the early phase of the project, and results are described in detail elsewhere.8 Study staff was available in person and by conference call to verify data entry

66

and coding practices. The adverse event data were categorized and evaluated for agreement by 2 study team members. Data entry errors were continuously assessed with quarterly reports, and the study units were contacted for unusually high or low adverse occurrence numbers. Outliers were reconciled with unit managers. Missing data percentages were included in the quarterly reports. Data Analysis In addition to ongoing data assessments, at the outset of analysis the entire data set was evaluated for out-of-range and missing data. Out-of-range elements that could not be reconciled were recoded as missing. Last value carried forward was used to impute missing census (7% of shifts), staffing values (2% of shifts), and patient acuity (35% of days) information. This method was chosen based on observations that census and patient acuity values were positively autocorrelated; that is, they show considerable overlap from day to day, as well as from shifts within a day. Shifts with missing outcomes were excluded from analysis. Because multiple adverse events per shift were extremely rare (eg, only 0.08% of shifts had 2 or more MAEs), all outcomes were dichotomized as 0 indicating the absence of, or 1 the presence of, an adverse patient event on a shift. The probability of each adverse event was modeled using Bayesian hierarchical logistic regression.13 This modeling framework accounts for the nesting of shifts within days within units. For comparability and simplicity of presentation, we chose to fit and report an identical model specification to all outcomes and across all unit types. Because patient care characteristics in small hospitals differ from those in larger ones (eg, level of specialization, as well as organizational and structural factors), we adjusted for hospital size. Unit types were analyzed separately. Only statistically significant odds ratios are reported. (See Tables for full results, Supplemental Digital Content 1, which displays the hierarchical logistic regression modeling results for falls, http:// links.lww.com/JONA/A39; Supplemental Digital Content 2, which shows the hierarchical logistic regression modeling results for falls with injury, http://links.lww.com/JONA/A40; and Supplemental Digital Content 3, which shows the hierarchical logistic regression modeling results for medication errors, http://links.lww.com/JONA/A41).

Results Table 2 provides the covariates summarized by unit type. Figure 1 shows the observed rate of each

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Figure 1. Observed rates of outcomes by unit type.a Abbreviation: MAE, medication administration error. aRates are calculated based on percentages of shifts with the event occurrence. Of 99,412 shifts with complete data, 974 had a fall. Of 99,338 shifts with complete data, 211 falls occurred that resulted in injury. Of 97,655 shifts with medication administration error data, 1,395 had a documented medication error.

outcome by unit type. MAEs occurred more frequently than either type of falls. Falls and Falls With Injury Table 3 shows the results of analyses for falls, falls with injury, and MAEs. A greater proportion of RNs relative to unlicensed assistive personnel (the comparison category) was significantly associated with fewer falls in medical-surgical and critical care units but not in step-down units. Fewer falls were associated with a higher percentage of DoD civilian nurses working on a shift. Higher nursing care hours per patient per shift were significantly associated with a decreased likelihood of both falls and falls with injury. Increased acuity was associated with increased falls only on medical-surgical units. A higher patient census was significantly related to more falls in both step-down and medical-surgical units. Falls without injury were more frequent on night shift; day of the week was not associated with falls. Medication Administration Errors As with falls, a higher percentage of RNs compared with unlicensed assistive personnel was significantly associated with fewer MAEs in medicalsurgical and critical care units. A higher proportion of DoD civilian nursing staff was associated with fewer MAEs, particularly in step-down and critical care units.

A higher number of total nursing care hours per shift were significantly associated with fewer MAEs occurring in medical-surgical and critical care units. Night shift had significantly fewer MAEs. Daily census and acuity were positively associated with MAEs.

Discussion The findings of this study support the assumption that adverse events occur during shifts that are staffed with fewer personnel overall and fewer RNs in particular. Differences in fall rates by unit type were expected because of the nature of patients’ illness-related mobility restrictions in each type of unit. Differences in rates of MAEs were also expected based on the observation that critical care nurses care for fewer patients and thus may be more familiar with the medications for each patient, whereas medical-surgical nurses care for several patients concurrently and may not be as familiar with each patient’s specific medication regimen. Relationships with RN skill mix were most notable in falls with injury; each 10% decrease in RN skill mix was associated with a 36% increase in the likelihood of falls with injury in critical care units and with a 30% increase on medical-surgical units. This association was greater on critical care

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

67

Table 2. Shift-Level Covariates by Unit Type (n = 111,522 Shifts)a Variable Shift census Skill mixb RN (%) LPN (%) UAP (%) Staff categoryb Active military, % DoD civilian, % Contract, % Reserve, % NCHPPS Total RN LPN Patients per RN

Medical-Surgical (n = 57,913 Shifts)

Step-down (n = 18,039 Shifts)

Critical Care (n = 35,570 Shifts)

15.68 (7.18)

10.63 (5.54)

5.82 (2.87)

51 (14) 22 (17) 28 (15)

58 (17) 24 (18) 19 (15)

77 (19) 14 (17) 9 (17)

44 34 19 3

36 39 22 3

41 47 8 5

4.29 2.15 0.87 4.82

(28) (24) (19) (9) (2.84) (1.60) (0.96) (2.37)

5.43 3.16 1.22 3.26

(24) (26) (19) (8) (2.97) (2.09) (1.12) (1.59)

9.42 6.87 1.12 1.46

(32) (31) (14) (11) (6.27) (3.90) (1.70) (0.73)

Abbreviations: UAP, unlicensed assistive personnel; NCHPPS, nursing care hours per patient shift. a Data are presented as mean (SD). b Skill mix and staff category columns may not add up to 100% because of rounding.

units, where high-acuity patients are beginning ambulation after critical illness. The RNs on those units could be more aware of underlying physiological implications of patient-specific medications, extended bed rest, illness, and injury matters that might not be well understood by other levels of nursing personnel. There was a strong relationship between total staffing (nursing care hours per patient shift) and falls with injury. Depending on unit type, a 15% to 51% increase in falls with injury was shown with each

decrease of 1 hour of nursing care per shift. This finding differs from a CALNOC report6 showing that staffing changes from mandated nurse-to-patient ratios were not associated with patient falls. One explanation may be the relatively wide variation in staffing noted across shifts in our study. This variation is likely muted when staffing data are aggregated to a monthly level, the procedure commonly used with other nursing outcomes databases. The percentage of staff on a shift who were DoD civilians had several remarkable associations. A 10%

Table 3. Hierarchical Logistic Regression ORs Falls, Falls With Injury, and MAEa Falls Predictorb Shift level Night shift Skill mix (10% decrease) % RN % LPN Staff category (10% decrease) % Military % Civilian % Contract Total NCHPPS (1-h decrease) Day level Day of week: Monday Census (increase of 3 patients) Acuity (1-SD increase) Unit level: large hospital

MedicalSurgical

Stepdown

Falls With Injury Critical Care

MedicalSurgical

Stepdown

MAE

Critical Care

1.36 1.11 1.08

1.20

1.30 1.14 1.48 1.18 1.15

1.09 1.48 1.17 1.07

1.09 1.33

1.14 1.36

1.14

1.11

1.17 1.13 2.34

1.29

1.36

MedicalSurgical

Stepdown

Critical Care

0.43

0.46

0.41

1.13 1.10

1.50 1.25

1.17

1.67 1.51

1.13

1.47 1.05

0.76 1.57 6.77

7.19

1.07 1.13

1.36

20.16

1.14 1.17

Abbreviations: UAP, unlicensed assistive personnel; NCHPPS, nursing care hours per patient shift. a Only the statistically significant ORs are reported. b Reference group excluded for each categorical variable: day shift for shift time, UAP for provider type, reserve for provider category, Tuesday-Sunday for day of week, and small hospital for hospital size. Only large hospitals in the sample had step-down units.

68

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

decrease in the percentage of civilian staff on a given shift was associated with a 33% to 48% increased likelihood of falls and as much as a 67% increased chance of MAEs. To help explain this relationship, we examined the differences in demographics between military and civilian nurses from a crosssectional annual survey of nurses from our participating hospitals. Noteworthy among the differences was the level of experience. Military nurses had, on average, 9 years less experience than the civilian nursing personnel (5 vs 14 years, respectively; t = j17.88, P G 0.001). This difference was also observed in a separate study using a different military sample.14 This experience level difference reflects a philosophical distinction in career expectations of military and civilian nurses. Military nurses begin their careers in bedside care units and are expected to move into leadership positions and away from direct care. Civilian nurses do not have this expectation, infrequently hold nonYdirectcare positions, and are usually hired with at least 1 year of experience. Our study’s personnel category may be serving as a proxy for experience, which other researchers have found to be associated inversely with MAEs and falls.15 Future studies could investigate how individual nurse assignments and/or the degree of teamwork among staff nurse personnel may impact adverse event occurrences. Alternate explanation for falls and MAEs that may have influenced our findings includes case-mix differences, which we did not control for directly; rather we analyzed unit types separately as a rough control for case mix. Still, even among critical care units in large hospitals, we could expect differences in case mix that were not captured by our acuity measure. In analyzing shift times, 45% fewer MAEs occurred on critical care units at night. This lower incidence may have resulted because fewer medications are administered during night shifts, fewer new or modified medication orders are received on night shifts, and therefore fewer opportunities for error would exist, or there may be fewer interruptions during night shift medication dispensing. The limitations in this study include the use of incident reports for our primary outcome variables, the amount of missing data on acuity, and the lack of adjustment for risk of falling or for risk of MAE. Skepticism and controversy surround the use of institutional incident reports as markers of patient safety because of assumed underreporting from the voluntary nature of reporting, fear of administrative reprisal, and fear of litigation.16 However, the quarterly rates of falls and MAEs reported in this study as ranges per 1,000 patient-days (ie, 1.0-3.3 and 2.6-5.7, respectively) are comparable to others

reported in the literature that used incident reports (eg, falls 3.73 [Dunton et al10], 3.01-3.19 [Donaldson et al6], 2.2 [Blegen and Vaughn17]; and MAEs 6.74 [Blegen and Vaughn17], 5.36-6.22 [Mark and Belyea18]). Because we were investigating whether an adverse patient event occurred on a shift, we did not consider patient risk of falling or doses of medications dispensed. As a result, we were not able to examine the relationship of staffing to prevention of falls in the presence of risk. The incidence of more than 1 adverse event per shift was extremely small; thus, once an adverse event occurs on a shift, it may serve to heighten attention to preventing a second event throughout the rest of the shift. For the rarer outcome of injury falls, odds ratios (ORs) for large versus small hospitals and yearly ORs relative to 2006 rates were unstable because of the very low numbers of events for subsets of the data relevant to those calculations. Therefore, those specific results need to be used with caution. Another limitation may be the generalizability of our results. We studied care in a particular healthcare system that is governed by additional rules, regulations, and accountability, possibly limiting variability between the hospitals themselves. Yet, our study documents substantial variability in staffing and adverse events. Although there are differences in the military healthcare system, its beneficiaries mirror the civilian population according to a recent report.9 The authors found no published case-mix comparisons between military and civilian hospitals nationally, however, the military monitors beneficiary care provided in both military and network hospitals (ie, care purchased by the military within the civilian sector). Network hospitals have a higher average length of stay, attributable to a higher procedural complexity, than military hospitals19; however, the difference was small (ie, 0.66 days in 2009) and perhaps not clinically significant. There are notable strengths to this study. Because the database was designed to inform leadership for ongoing decision making, the data elements were standardized using the measures recommended by the National Quality Forum,20 and plans were incorporated to ensure reliability and validity of the data and its collection processes. This study broke new ground in its methodological granularity and in linking unit-level, shiftspecific staffing to key adverse events that are preventable in hospitals and provides strong evidence for continuing discussions examining the relationship of patient outcomes to nursing care. We hope our results can more specifically argue for the value of continuing the quest to determine

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

69

the right nursing skill mix at the right time and for the right patients. Utilizing prospective databases within hospital systems will be extremely useful for future tracking, research, and analyses.

Implications for Nurse Managers and Executives This study provides evidence that every shift must be adequately staffed with the right numbers and mix of nursing staff qualifications and experience. However, it cannot prescribe that number or mix. In this imperfect science, a better understanding of Bcorrect[ staffing levels can be attained only by the knowledge that comes from ongoing monitoring of staffing and outcomes at the lowest organizational level possible. The findings from this study support a working model for collecting and disseminating reliable, valid, and usable data across hospitals to support patient safety. Promoting the use of high-quality inpatient data can contribute to the development

of evidence-based policies and procedures to monitor the effect of nurse staffing on clinical and service outcomes. This model is currently being modified as an interdisciplinary clinical outcomes database for patient safety metrics in Army medical facilities. Fundamental questions remain, including the ways in which unit-level patient care factors, beyond the structural indicators, interact to alter patient care outcomes and hospital-level performance. Although there is general consensus in the literature that nurse staffing is linked to patient outcomes, little is known about the ways in which variations in characteristics of the nurse workforce, nurse workload, and specific nursing practices interact to affect acute care outcomes.

Acknowledgment The authors acknowledge the MilNOD database manager, Ms Stacy Heiner, BSN, RN, and the entire TriService MilNOD research team.

References 1. Needleman J, Buerhaus P, Meattke S, Stewart M, Zelevinsky K. Nurse-staffing levels and the quality of care in hospitals. N Engl J Med. 2002;346(22):1715-1722. 2. Lang TA, Hodge M, Olson V, Romano PS, Kravitz RL. Nurse-patient ratios: a systematic review on the effects of nurse staffing on patient, nurse employee, and hospital outcomes. J Nurs Adm. 2004;34:326-337. 3. Kane RL, Shamliyan TA, Mueller C, Duval S, Wilt TJ. The association of registered nurse staffing levels and patient outcomes: systematic review and meta-analysis. Med Care. 2007; 45:1195-1204. 4. Clarke S, Donaldson NE. Nurse staffing and patient care quality and safety. In: Hughes R, ed. Patient Safety and Quality: An Evidence Based Handbook for Nurses. Rockville, MD: Agency for Healthcare Research and Quality; 2008: 2-111-2-136. 5. Lake ET, Chueng RB. Are patient falls and pressure ulcers sensitive to nurse staffing? Western J Nurs Res. 2006;28(6):1-24. 6. Donaldson N, Bolton LB, Aydin C, Brown D, Elashoff JD, Sandhu M. Impact of California’s licensed nurseYpatient ratios on unit-level nurse staffing and patient outcomes. Policy Polit Nurs Pract. 2005;6(3):198-210. 7. Clarke S. Registered nurse staffing and patient outcomes in acute care: looking back, pushing forward. Med Care. 2007; 45:1126-1128. 8. Patrician PA, Loan L, McCarthy M, Brosch LR, Davey KS. Towards evidence-based management: creating an informative database of nursing-sensitive indicators. J Nurs Scholarsh. 2010;42(4):358-366. 9. Gimbel RW, Pangaro L, Barbour G. America’s Bundiscovered[ laboratory for health services research. Med Care. 2010;48(8): 751-756. 10. Dunton N, Gajewski B, Taunton RL, Moore J. Nurse

70

11. 12.

13.

14.

15.

16.

17. 18.

19.

20.

staffing and patient falls on acute care hospital units. Nurs Outlook. 2004;52:53-59. Allan EL, Barker KN. Fundamentals of medication error research. Am J Hosp Phar. 1990;47:555-571. Army Medical Expense and Performance Reporting System Program Office (n.d.). Workload management system for nursingVarmy. Available at http://www.ampo.amedd.army. mil/wmsna/index.html. Accessed September 6, 2010. Spiegelhalter DJ, Thomas A, Best NG, Lunn D. WinBUGS: Bayesian Inference Using Gibbs Sampling for Windows, Version 1.4. Cambridge, UK: MRC Biostatistics Unit; 2003. Patrician PA, Shang J, Lake ET. Organizational determinants of work outcomes and quality care ratings among Army Medical Department registered nurses. Res Nurs Health. 2010;33:99-110. Blegen MA, Vaughan TE, Goode CJ. Nurse experience and education: effect on quality of care. J Nurs Adm. 2001;31(1): 33-39. Blegen MA, Vaughn T, Pepper G, Vojir C, Stratton K, Boyd M, Armstrong G. Patient and staff safety: voluntary reporting. Am J Med Qual. 2004;19(2):67-74. Blegen MS, Vaughn T. A multisite study of nurse staffing and patient occurrences. Nurs Econ. 1998;16(4):196-203. Mark BA, Belyea M. Nurse staffing and medication errors: cross-sectional or longitudinal relationships? Res Nurs Health. 2009;32:18-30. Department of Defense. Evaluation of the TRICARE Program: fiscal year 2010 report to Congress. Available at http://www. tricare.mil/downloads/TRICARE20102_28_10v7.pdf. Accessed September 6, 2010. National Quality Forum. National Voluntary Consensus Standards for Nursing-Sensitive Care: An Initial Performance Measure Set. Washington, DC: Author; 2004.

JONA  Vol. 41, No. 2  February 2011

Copyright @ 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.