| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
Alison M. Trinkoff, Carles Muntaner, and Rong Le are with the Department of Family and Community Health, and Meg Johantgen is with the Department of Organizational Systems and Adult Health, University of Maryland School of Nursing, Baltimore, Md. Carlos Muntaner is also with the Center for Addiction and Mental Health, University of Toronto, Toronto, Ontario, and the Institute of Work and Health, Toronto.
Correspondence: Requests for reprints should be sent to Alison M. Trinkoff, ScD, University of Maryland School of Nursing, 655 W Lombard St, Rm 625, Baltimore, MD 21201 (email: trinkoff{at}son.umaryland.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. We examined the relationship between nursing home staffing levels and worker injury rates in 445 nursing homes in 3 states.
Methods. We obtained First Reports of Injury and workers’ compensation data from 3 states (Ohio, West Virginia, and Maryland) for the year 2000. We then linked these data to Medicare’s Online Survey, Certification and Reporting system to obtain nursing home staffing details and organizational descriptors. We used ordinary least squares and log-transformed regression models to examine the association between worker injury rate and nursing home staffing and organizational characteristics.
Results. Total nursing hours per resident day were significantly associated with worker injury rates in nursing homes after we adjusted for organizational characteristics and state dummy variables (P=.0004).
Conclusions. Our findings suggest that nursing home staffing levels have an important impact on worker health. These findings were supported for multiple facilities across different states; therefore, policies and resources that increase staffing levels in nursing homes are warranted.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Nursing home employees working in direct-care facilities perform many physically taxing activities, such as lifting heavy loads, working in awkward postures, and transferring residents.6,10–14 Additionally, manipulating the technology that supports patient care is physically straining. The increased worker injury rates likely result from increased exposure to hazardous conditions and diminished recovery time between exposures.15
Worker injuries in health care institutions associated with staffing levels and skill mix have been previously examined. Because health care institutions have been required to perform more efficiently, the resultant changes are lower staffing levels and higher patient loads, both of which have been shown to increase worker injury. In a study of 12 hospitals in the Minneapolis–St Paul, Minnesota, area that used data from 1990 to 1994, Shogren and Calkins16 found that when registered nurse (RN) positions were decreased by 9%, work-related illnesses and injuries among nurses increased by 65%. A review of the impact of staffing on health care by the Institute of Medicine noted that there is empirical evidence that shows back injuries among nurses are associated with staffing levels.17 Although the extent of worker injuries among resident care staff in nursing homes has been documented,18–20 there have been few studies about the association between injuries and staffing.
The occurrence of these injuries has important implications for staff retention. Owen and Garg21 found that 20% of nurses who reported they had back pain said they had made at least 1 job change in order to decrease the number of nursing home residents that had to be lifted and transferred. Turnover among unlicensed personnel was even higher,22 with 23% annual turnover reported among nursing assistants in one facility.23 In a statewide survey of nursing assistants, 30% reported they planned to quit their jobs.24
We used an ecological design that was based on administrative data to examine the association between staffing rates and worker injuries. To do this, we analyzed the association between staffing variables (total nursing care hours per resident day) and adverse worker outcomes (reported worker injuries) at the institutional level. Analyses were also adjusted for resident acuity, profit status, nursing home size, and availability of nurse aide training.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Data Sources and Measures
We obtained staffing and organizational descriptors from the OSCAR database. Our variables included number of beds, special services, RN and other personnel staffing, type of nursing home ownership, and resident acuity. These data are routinely collected by the Centers for Medicare and Medicaid Services (CMS) to support the survey and certification function and to monitor deficiencies and quality of care in US nursing homes that receive Medicare or Medicaid funds. Because the OSCAR data are continually updated by overwriting the previous data, we purchased historical data and documentation from the Cowles Research Group.25 OSCAR data for the 3 states in our sample were extracted from this large database.
We used First Report of Injury (FROI) databases for 2 of the states, Ohio and West Virginia, to measure worker injuries. This is believed to be the best source of injury reports because the process of filing workers’ compensation claims has many systematic biases that can lead to suboptimal ascertainment of injury.26 On the other hand, because workers’ compensation claims tend to be filed for only the most severe injuries,27 we felt it was also important to include worker injury data from these claims to examine the study question. Therefore, we also used workers’ compensation claims data from Maryland to calculate worker injuries.
FROI and workers’ compensation data were obtained from state agencies. The FROI data are comparable to Occupational Safety and Health Administration OSHA-200 log data, but they are obtainable at the state level for some states. Although injury data were obtained for individual workers, we aggregated injury data to the organizational level for the analyses. All reported injuries were included, regardless of type, although the overwhelming majority of injuries were musculoskeletal in origin (predominantly back injuries).
Worker injury rates by skilled nursing facility were calculated with formulas for injury incidence from the Bureau of Labor Statistics’ Occupational Safety and Health Definitions.28 To produce an overall rate, we aggregated the total number of nonfatal injuries among RNs, licensed practical nurses (LPNs), and aides for each facility and divided the aggregate by the sum of the full-time equivalents (FTE) for these 3 employee categories. Multiplying the rates by 100 allows reporting per 100 FTE.
Staffing variables were created with coding rules designed by Harrington et al.29 FTE data were reported for a 14-day period, and we used the coding rules to convert staffing data to staffing hours per resident day by taking the total nursing staff FTEs reported for a 2-week period and multiplying by 70 work hours for the period. We divided the total staffing hours by the total number of residents and then by 14 days in the reporting period. In accordance with Harrington et al.,29 we included all full-time, part-time, and contract positions for RNs; directors of nursing were excluded. We included all LPNs and licensed vocational nurses, and, for nursing aide staffing, we included all certified nursing assistants, nursing assistants in training, and medication aides.
File Construction
We applied exclusion criteria to remove nursing homes from the database based on the recommendations by Harrington et al29: (1) too small—those with fewer than 15 beds, (2) hospital based, (3) no RN hours—having 60 or more beds but no RN hours, (4) extra RN hours—more than 12 RN hours per resident day, (5) few nursing staff hours—less than 0.5 total nursing hours per resident day, and (6) excess nursing hours—more than 12 total nursing hours per resident day. We excluded skilled nursing facilities that had excess nursing hours to remove those facilities that function as acute care step-down facilities and therefore do not reflect the staffing patterns for long-term care providers. For example, in West Virginia, after we applied each of the exclusion criteria the original sample of 133 facilities decreased to 129 after deleting small facilities, to 103 after deleting hospital-based facilities, and to 102 after deleting facilities with excess nursing staff hours; only 77% of the original facilities remained.
Worker injury data required considerable cleanup. We culled the data received from the states to extract injuries that occurred in nursing homes during 2000. For the databases that included a Standard Industrial Code (http://www.osha.gov/pls/imis/sic_manual.html), nursing homes were identified with an 805 code (skilled nursing, intermediate nursing, and nursing and personal care facilities). Upon review of these codes, we determined that some facilities were not nursing homes (assisted living, temporary staffing agency, system corporate office) and deleted them. Through further analysis and recoding, we retained only those records that represented injuries to RNs, LPNs, and aides. To facilitate analysis and linkage of the databases, we assigned the CMS 6-digit provider number for each facility to each worker injury record. Because facility names in the injury database were written as text with abbreviations, common names, and corporation names, direct linkage to the OSCAR name was not always possible. In such cases, we used the CMS Nursing Home Compare database and other sources to match the nursing home with its address.
Data Analysis
Statistical analyses were performed with SAS, version 8.2 (SAS Institute Inc, Cary, NC). We used descriptive statistics to examine the association between organizational characteristics and facility by state. We used multivariate regression to identify the independent effect of these organizational characteristics—particularly staffing—on worker injury rate. Because linked-facility sample sizes for West Virginia and Maryland were small, we combined nursing home data from all 3 states into 1 file (n = 445 linked facilities) to eliminate concerns about adequacy of power for these analyses. We included state dummy variables in regression models because of systematic differences across the states. Before we analyzed the association between worker injury rate and nursing home characteristics (acuity index, total residents, percentage of Medicaid, location, profit status, aide training, and nursing hours per resident day), we screened the data for normality, missing values, outliers, and multicollinearity. Acuity was measured with the Acuindex, which was developed as part of the work on the CMS Minimum Data Set. The Acuindex takes into account the proportion of residents with activities of daily living dependencies and the proportion requiring special treatments (e.g., suctioning, parenteral feeding). Because this measure reflects resident care burden, it also could influence worker injury rates across facilities. Therefore, we included acuity in our analysis to control for variation in case mix. We defined facilities with aide training as those facilities with an approved Nurse Aide Training and Competency Evaluation Program. Among predictors, the percentage of Medicaid residents was highly correlated with the acuity index in all 3 states. The percentage of Medicaid was then dropped from further multivariate analysis. Additionally, because injury rates were highly skewed, we modeled the log of total injuries.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
). Ohio had the most nursing homes (n=778) of the 3 states sampled, followed by Maryland (n=196) and West Virginia (n= 102). Staffing levels were less distinct across the states, with all 3 states having total nursing hours per resident day (sum of RN, LPN, and nursing aide staffing) that ranged from an average of 3.3 to 3.5 hours. However, West Virginia had a far lower average number of RN hours per resident day (0.35) compared with Maryland (0.54) and Ohio (0.57), which indicated a lower skill mix. Our analysis of the overall nursing home characteristics by state showed that while acuity was similar across states, total residents, percentage of Medicaid residents, profit status, aide training, and location differed. Maryland facilities had more residents on average, a lower proportion of for-profit facilities, and the lowest percentage of Medicaid residents. In West Virginia, two thirds of the facilities were in rural areas, whereas in Ohio and Maryland, at least three fourths were in urban areas. The proportion of facilities with aide training was twice as high in West Virginia compared with Ohio and Maryland.
|
, there were no large differences in the average staffing levels for the linked sample compared with the OSCAR sample. The organizational characteristics also were similar. We then compared the staffing means by state for the total OSCAR sample with the means by state for the linked sample (Tables 1
and 3). Once again, there was little or no change in average staffing (nursing hours per resident day) by state, before and after linkage. As a result, we proceeded with the linked sample for our analysis.
|
|
shows injury rates for the 3 states. We found that overall injury rates in Maryland were the lowest, which was expected because the injury rates were calculated from workers’ compensation claims rather than from FROI data. However, the almost total lack of claims filed by nurse aides in Maryland was unexpected. We also found other differences in rates by state, which likely reflected differences in injury reporting and coding.
The results of the ordinary least squares regression showed that total nurse hours per resident day was significantly associated with worker injuries after we adjusted for acuity, profit status, aide training, total residents, and state (P =.0004). Our analysis showed that 25% of the variance in worker injury was explained by the model (Table 4). For each additional hour increase in nursing care, injuries were predicted to decrease by 2.4 per 100 FTEs. The number of total residents also had a significant negative effect: as size increased, worker injuries decreased. To examine this further, we stratified nursing homes by number of residents and found that injury rates were lower in homes where there were more residents, although staffing did not vary. Because of the apparent underreporting of injury rates among nurse aides in Maryland, we reran the regression models and excluded Maryland. The results were the same (data not shown).
|
). Each additional hour of nursing care decreased the injury rate by nearly 16%. Thus, for every unit increase in staffing (total hours of nursing care), worker injury rates decreased by 2 per 100 FTEs. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
The consistency of the association between staffing and injury across states and facilities is noteworthy and supports the credibility of the findings, although there are limitations to our study. The ecological design did not allow us to make inferences about individual workers.32 Missing data from the injury databases hindered our ability to link across databases, and the presence of missing data in certain fields (e.g., occupation) also reduced the completeness of the data analysis. Despite these limitations, comparison of descriptors from nursing homes in the original OSCAR sampling frame with those in the linked frame showed surprisingly few differences.
As expected, state variables were highly significant, which underscores the importance of adjusting for them in a combined model. A minimal number of injuries were reported by nurse aides in Maryland. Although the exact reason for this is unknown, the injured aides in Maryland most likely did not file workers’ compensation claims, probably owing to a lack of awareness; posting the law in the workplace is not required. Also, the injury definitions can reduce the likelihood of filing claims, e.g., back injuries in Maryland must have an acute onset to be claimable. Furthermore, claims in Maryland must be filed and signed by the injured employee—a provider or other party cannot initiate the claim—which may serve as a disincentive to file among those who have insecure jobs.
Profit status and acuity were not significantly associated with worker injury when state, size, and staffing were controlled. On the other hand, Banuszak-Holl and Hines22 found that nursing turnover, a factor correlated with injuries, was higher among for-profit nursing homes, which also tended to have lower staffing ratios.33 This was also true for our sample. The lack of impact of aide training was unexpected, because training has been associated with lower injury rates,17 although we did not take into account the impact of staffing in these studies. Adjusting for differences in resident acuity removed case mix as a potential source of confounding, which was important because nursing homes with more dependent residents may have higher rates of worker injury. It is also possible that such homes have more assistive equipment that reduces injury risk to workers.22 The current approaches to nursing home staffing are often made on the basis of staff-to-resident ratio or hours per resident day, with no accounting for differences in acuity. This is reflected in our data, wherein the acuity index from the OSCAR database was not correlated with staffing (r =0.03). Ongoing research is being conducted to examine the association between acuity and staffing in nursing homes.
The OSCAR data also have limitations. The Centers for Medicare and Medicaid Services performs edit checks on the OSCAR data to identify errors. Straker34 compared 1995 OSCAR data with data from the Ohio Department of Health to examine consistency in several variables, including staffing. Staffing correlations per patient day were 0.61, although self-reports did not typically cover the same period reported as the OSCAR assessment. Another study examined actual payroll and found correlations less than 0.5 between the data reported in both the OSCAR and the payroll,35 although these analyses had strict exclusion criteria.
As for the worker injury data, some injuries will be missed even with the use of FROI data. For example, workers may seek injury care from their regular health provider and fail to mention that the injury is work-related.26 Despite such limitations, FROIs are generally a more complete source of potentially claimable injuries to health care workers than workers’ compensation data.27 Ideally, the hours worked would exclude paid non-work time, although we had no way to remove this from our analysis. However, this time is minimal among nurses, who often skip breaks and lunches and perform uncompensated overtime because of short staffing.36,37 Because injury data from the 3 states were treated similarly in our analysis, these distinctions should not affect the ability to associate injuries with staffing.
Despite our successful attempt at using different worker injury databases from multiple states in this analysis, there should be standardization of both reported data and definitions of worker injury.38 Outcomes data reported at the facility level should be available even when facilities manage injuries at the corporate level to allow for analysis of staffing and related outcomes. The National Quality Forum now recommends that staffing and skill mix be examined as performance measures for evaluating health care quality.39
Our study has shown that the impact of staffing is also important for worker health. By improving staffing levels in nursing homes, both workers and residents will benefit. With the impending shortage of long-term care workers, it is imperative that we promote the health of this essential group of care providers; they will be increasingly needed to care for an aging population.
|
Acknowledgments |
|---|
Human Participant Protection
The project was reviewed by the institutional review board of the University of Maryland and was determined to be exempt from the institutional review board approval process according to DHHS 45 CFR 46.101.b (4).
|
Footnotes |
|---|
Contributors
A.M. Trinkoff originated the study, supervised its implementation, and led the writing. M. Johantgen created the database and directed the data analysis. C. Muntaner assisted with the study, analysis of findings, and article preparation. R. Le assisted with the study and completed the analyses. All authors originated ideas, interpreted findings, and reviewed drafts of the article.
Accepted for publication September 9, 2004.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2. Kinsella K, Velkoff VA. An Aging World: 2001. Available at: http://www.census.gov/prod/2001pubs/p95-01-1.pdf. Accessed July 9, 2002.
3. American Association of Retired Persons (AARP). Across the States: Profiles of Long-Term Care Systems. Washington, DC: Long-Term Care Policy Institute; 2000.
4. Bureau of Labor Statistics. Lost-Worktime Injuries and Illnesses: Characteristics and Resulting Time Away from Work, 1998. Available at: ftp://ftp.bls.gov/pub/news.release/History/osh2.04202000.news. Accessed October 10, 2000.
5. Bureau of Labor Statistics. Lost-Worktime Injuries and Illnesses: Characteristics and Resulting Time Away from Work, 1999. Available at: ftp://ftp.bls.gov/pub/news.release/history/osh2.03282001.news. Accessed May 17, 2001.
6. Collins JW, Owen BD. NIOSH research initiatives to prevent back injuries to nursing assistants, aides, and orderlies in nursing home. Am J Ind Med.1996;29:421–424.[CrossRef][ISI][Medline]
7. Personick ME. Nursing home aides experience increase in serious injuries. Mon Labor Rev.1990;113:30–37.[Medline]
8. Meyer J, Muntaner C. Injuries in home health care workers: An analysis of occupational morbidity from a state compensation database. Am J Ind Med.1999;35:295–301.[Medline]
9. SHARP: Safety and Health Assessment and Research for Prevention Program. Work-Related Disorders of the Back and Upper Extremity in Washington State, 1989–1996. Olympia, Wash: Washington State Department of Labor and Industries; 1998. Technical Report No.: 40-1-1997.
10. Allen A. On-the-job injury: a costly problem. J Post Anesth Nurs.1990;5:367–368.[Medline]
11. Brulin C, Gerdle B, Granlund B, Hoog J, Knutson A, Sundelin G. Physical and psychological work-related risk factors associated with musculoskeletal symptoms among home care personnel. Scand J Caring Sci.1998; 12:104–110.[CrossRef][ISI][Medline]
12. Marras WS, Davis KG, Kirking BC, Bertsche PK. A comprehensive analysis of low-back disorder risk and spinal loading during the transferring and repositioning of patients using different techniques. Ergonomics.1999;42:904–926.[Medline]
13. Sosnowitz B, Hriceniak J. Neonatal intensive care units can be hazardous to nurses’ health. J Perinatol.1998;8:253–257.
14. Trinkoff AM, Storr CL, Lipscomb JA. Physically demanding work and inadequate sleep, pain medication use, and absenteeism in registered nurses. J Occup Environ Med.2001;43:355–363.[ISI][Medline]
15. Bongers PM, De Winter CR, Kompier MAJ, Hildebrandt VH. Psychosocial factors at work and musculoskeletal disease. Scand J Work Environ Health.1993; 19:297–312.[ISI][Medline]
16. Shogren E, Calkins A. Findings of the Minnesota nurses Association Research Project on Occupational Injury Illness in Minnesota Between 1990–1994. St. Paul, Minn: Minnesota Nurses Association;1997.
17. Wunderlich GS, Sloan FA, Davis CK, eds. Nursing Staff in Hospitals and Nursing Homes. Washington, DC: National Academy Press; 1996.
18. Evidence-Based Practice Center (Oregon Health and Science University). The Effect of Healthcare Working Conditions on Patient Safety. Evidence Report/Technology Assessment No. 74. Rockville, Md: Agency for Health-care Research and Quality; 2003. AHRQ Publication No. 03-E024. Contract number 290-97-0018.
19. Muntaner C, Lynch J, Oates G. The social class determinants of income inequality and social cohesion. Int J Health Serv.1999;29:699–732.[CrossRef][ISI][Medline]
20. Myers D, Silverstein B, Nelson NA. Predictors of shoulder and back injuries in nursing home workers: a prospective study. Am J Ind Med.2002;41:466–476.[Medline]
21. Owen BD, Garg A. Patient handling tasks perceived to be most stressful by nursing assistants. In: Mital A, editor. Advances in Industrial Ergonomics and Safety 1. London, UK: Taylor & Francis; 1989:775–781.
22. Banaszak-Holl J, Hines MA. Factors associated with nursing staff turnover. Gerontologist.1996;36:512–517.[Abstract]
23. Remsburg RE, Armacost KA, Bennett RG. Improving nursing assistant turnover and stability rates in a long-term care facility. Geriatr Nurs.1999;20:203–208.[Medline]
24. Parsons SK, Simmons WP, Penn K, Furlough M. Determinants of satisfaction and turnover among nursing assistants. The results of a statewide survey. J Gerontol Nurs.2003;29:51–58.[Medline]
25. Cowles Research Group. Available at: http://www.longtermcareinfo.com/crg/. Accessed May 17, 2001.
26. Boden LI. Workers’ compensation in the United States: high costs, low benefits. Annu Rev Public Health.1995;16:189–218.[Medline]
27. Wickizer TM, Franklin G, Plaeger-Brockway R, Mootz RD. Improving the quality of workers’ compensation health care delivery: the Washington state occupational health services project. Milbank Q.2001;79:5–33.[Medline]
28. Bureau of Labor Statistics. Occupational Safety and Health Definitions. Available at: http://stats.bls.gov/iif/oshdef.htm. Accessed June 29, 2004.
29. Harrington C, Carrillo H, Thollaug S, Summers P, Wellin V. Nursing Facilities, Staffing, Residents, and Facility Deficiencies, 1993 Through 1999. Available at: http://www.cms.hhs.gov/medicaid/services/nursfac99.pdf. Accessed October 12, 2000.
30. Clarke SP, Sloane DM, Aiken LH. Effects of hospital staffing and organizational climate on needlestick injuries to nurses. Am J Public Health.2002;92:1115–1119.
31. Petrisek AC, Mor V. Hospice in nursing homes. Gerontologist.1999;39:279–290.[Abstract]
32. Robinson WS. Ecological correlations and the behavior of individuals. Am Sociol Rev.1950;15:351–357.[CrossRef]
33. Harrington C, Woolhandler S, Mullan J, Carrillo H, Himmelstein DU. Does investor-ownership of nursing homes compromise the quality of care. Int J Health Serv.2002;32:315–325.[Medline]
34. Straker J. Reliability of OSCAR Occupancy, Census and Staff Data: A Comparison with the Ohio Department of Health Annual Survey of Long-Term Care Facilities. Miami, Fla: Scripps Gerontology Center, Miami University; 1999. Technical Report Series 3-01.
35. Centers for Medicare and Medicaid Services. Report to Congress: Appropriateness of Minimum Nurse Staffing Ratios in Nursing Homes Phase II Final Report; 2001 Dec. Contract No.: 500-95-0062-T.O.3. Available at: http://www.cms.hhs.gov/medicaid/reports/rp1201home.asp. Accessed July 9, 2002.
36. Shulman B. The Betrayal of Work: How Low-Wage Jobs Fail 30 Million Americans. New York, NY: WW Norton & Co Inc; 2003.
37. Gass TE, Vladeck BC. Nobody’s Home: Candid Reflections of a Nursing Home Aide. New York, NY: Cornell University Press; 2004.
38. Johantgen M, Trinkoff A, Gray-Siracusa K, Muntaner C, Nielsen K. Using state administrative data to study nonfatal worker injuries: challenges and opportunities. J Safety Res.2004;35:309–315.[Medline]
39. National Quality Forum. National Voluntary Consensus Standards for Nursing-Sensitive Performance Measurement. Available at: http://www.qualityforum.org/nursing_sensitive_performance_measurement.html. Accessed February 4, 2004.
This article has been cited by other articles:
|
|
 |
|
  G. Sorensen, E. Barbeau, A. M. Stoddard, M. K. Hunt, K. Kaphingst, and L. Wallace Promoting Behavior Change Among Working-Class, Multiethnic Workers: Results of the Healthy Directions--Small Business Study Am J Public Health, August 1, 2005; 95(8): 1389 - 1395. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
