Physician and Nurse Staffing in NursingHomes: The Role and Limitations of theOnline Survey Certification andReporting (OSCAR) System

Zhanlian Feng, PhD, Paul R. Katz, MD, Orna Intrator, PhD, Jurgis Karuza, PhD, and Vincent Mor, PhD

Objectives: To assess nursing home staffing data re-ported in the Online Survey Certification and Report-ing (OSCAR) system database for research and policy.

Design: Comparisons were made between OSCAR anda concurrent research survey of staffing data collectedfor the same facilities, using inter-rater agreementand correlation analyses.

Setting: Freestanding nursing homes from New YorkState (NYS) in 1997 (N � 327).

Measurements: Selected staffing variables were de-fined in comparable terms in both OSCAR and the NYSsurvey.

Results: The two data sources were in substantialagreement on the reported availability of a full-timephysician (other than medical director) and of a physi-cian assistant or nurse practitioner (Kappa �0.7), and

DOI: 10.1016/j.jamda.2004.12.008

ORIGINAL STUDIES

number of such staff (Spearman correlation �0.6). Thecorrelation was 0.8 for FTE registered nurses (RNs), 0.7for licensed practical nurses (LPNs), and 0.8 for certi-fied nurse aides (CNAs). In terms of average nursehours per patient day, separately for RNs, LPNs, CNAs,and all combined, the correlation was relatively weak(between 0.3 and 0.6). Overall staffing levels tendedto be lower in OSCAR than in NYS.

Conclusion: The OSCAR data are useful for exploringrelationships between staffing and various qualityof care outcomes, but may not be accurate enoughon a case-specific basis, or to determine policy re-garding minimal staffing levels using average nursehours per patient day measures. More systematicand timely efforts are needed to refine the OSCARcontent and survey methodology to document nurs-ing home staffing information. (J Am Med Dir Assoc2005; 6: 27–33)

they correlated well in the full-time equivalent (FTE) Keywords: OSCAR; nursing homes; staffing

Limited staffing in United States nursing homes, coupledwith increasing shortages of direct care workers,1,2 remains amajor challenge to the consistent delivery of high qualitycare. The need for accurate, facility-specific, staffing data iscritical if the necessary support is to be garnered and govern-

Center for Gerontology and Health Care Research, Brown University, Provi-dence, RI (Z.F., O.I., V.M.); University of Rochester School of Medicine andDentistry, Rochester, NY (P.R.K., J.K.); State University of New York BuffaloState College, Buffalo, NY (J.K.); Department of Community Health, BrownUniversity, Providence, RI (V.M.).

Research for this article was supported in part by NIA AG #11624 (Vincent Mor,PI) and HCFA Grant # 95-C090151/2-01 (Gerry Eggert, PI). An earlier draft ofthis article was presented at the 56th Annual Scientific Meeting of the Ger-ontological Society of America, November 21–25, 2003, San Diego, CA.

Address correspondence to Zhanlian Feng, PhD, Center for Gerontology andHealth Care Research, Brown University, Box G-ST, Providence, RI 02912.E-mail: Zhanlian_Feng@Brown.edu

Copyright ©2005 American Medical Directors Association

mental policy advanced to meet this challenge. Presently, theCenters for Medicare and Medicaid Services’ (CMS) OnlineSurvey Certification and Reporting (OSCAR) system is theonly uniform data source available to continually gauge thelevel of staffing in all certified nursing homes.3–5 To its credit,OSCAR data have been instrumental in informing long-termcare policy and quality assurance initiatives3,5–7 and in stim-ulating research regarding the structure, process, and qualityof care in nursing homes.8–13 The reliability and validity ofOSCAR staffing data have, however, been an ongoing con-cern.4,5,11,14–17 In OSCAR, each facility is required to reportthe number of staff hours, by staff category, worked in the 14days preceding the date of the annual survey. Despite suchstandardization, OSCAR data depend on self-report by facil-ity administrative staff and are subject to few confirmatoryaudits.

Alternative sources of staffing data, such as cost reports and

payroll records, have been compared with OSCAR with the

Feng et al. 27

promise of greater reliability and consistency. Cost reports,however, are largely restricted to cost by staff category sum-marized over an entire fiscal year, although staffing levels arelikely to be variable over any period of time due to highturnover rates among nursing staff. Moreover, expendituredata may be a poor proxy for staffing given significant wageand benefit differentials across facilities and by region, rurallocation, and average staff tenure. In addition, cost reports donot contain nationally uniform staffing data because person-nel categories and definitions differ from state to state.4 Pay-roll data have important limitations as well. There is consid-erable variability in facility payroll records, and often it isdifficult to ensure that staffing data extracted from payrollrecords actually refer to hours worked rather than hours paidbut not worked (eg, sick leave, vacation).3 Furthermore, pay-roll records do not include information on the number ofpatients,4 and would either have to be supplemented by ormerged with other data.

In spite of persistent questions regarding OSCAR staffinginformation, consistent research findings on the relationshipbetween staffing and various outcomes have been docu-mented from studies using OSCAR, as compared to dataderived from smaller, research driven studies. For example,non-profit and hospital-based facilities have been observed tohave higher staffing levels, and nurse staff ratios have beenfound to be higher in facilities with more acute case mixprofiles.18–20 More importantly, lower staffing levels havebeen correlated with an increased frequency of deficienciesand other quality of care problems.3,11,12,21–28

Despite numerous complaints about the inaccuracy of OS-CAR staffing data, thus far there have been few studiesexamining the validity and reliability of these data. One studycomparing 1995 OSCAR data with 1995 and 1997 surveydata from the Ohio Department of Health found weak cor-relation between the two data sources on nurse staffing levelsand staffing hours per resident.17 Another study commis-sioned by the Health Care Financing Administration assessedthe quality of OSCAR staffing data as compared to payrollrecords (deemed as “gold standard”) and to Medicaid costreports from 78 facilities in Ohio during the 1995 to 1997period.5 This study reported relatively low correlation be-tween OSCAR and the payroll data on nurse staffing levels,and also found that the cost report data more strongly corre-lated with payroll information than with OSCAR data. Costreports, therefore, were deemed a more reliable and validsource of staffing information than OSCAR. This conclusionis likely premature, however, given the small sample size andselect geography and time of the sample upon which thecomparisons were made.

While there have been many studies addressing nurse staff-ing issues based on OSCAR data, there has been little, if any,analysis of OSCAR-derived physician staffing data. This mayrelate not only to concerns with data reliability but also anunder-appreciation of the important role of the physician innursing home care.29

The purpose of this article is to compare staffing data froma research survey of 327 nursing homes conducted in 1997 in

New York State (NYS) with OSCAR data obtained from the

28 Feng et al.

same facilities within 1 year of the research survey datacollection. The New York survey data collection, based onresponses from the medical director, director of nursing, andadministrator from each facility, is likely to be more reliablethan OSCAR, considering the study’s explicit design andquality control to maximize convergent validity. The resultsof these comparisons are presented as a rationale for CMS toinstitute a more systematic and timely approach to collectingfacility-level staffing information.

METHODS

Data Sources

The research survey used to compare with OSCAR wasconducted in NYS in 1997. Survey questionnaires were de-signed separately for the administrator, medical director, anddirector of nursing, recognizing the unique perspectives ofeach of these professions. The administrator survey focused onissues related to facility structure, financing, allied healthprofessionals staffing, and the range of treatment programsoffered in the facility. The medical director survey askedabout physician staff models, credentialing, call systems, andprocesses regarding clinical care. The director of nursing sur-vey focused on nurse staffing levels, care planning, and the useof care teams. The survey followed the principles of theDillman method30 to orchestrate a systematic data collectionprocess, with three waves of survey mailing implemented.Responses were obtained from 410 (63%) of the 646 nursinghomes operating at that time, according to the NYS Depart-ment of Health listings.

The OSCAR is an administrative database maintainedby the CMS, which includes information on facility char-acteristics, patient conditions, and staffing levels routinelycollected from all Medicare/Medicaid certified nursinghomes as part of the licensure and certification process.CMS contracts with each state to conduct onsite inspec-tions, which occur on average about once a year but nolater than 15 months after the date of the previous standardsurvey. The information derived from OSCAR is self re-ported by each facility, reviewed by inspectors but notformally audited. In addition, this information changesfrequently as residents are discharged and admitted, orresidents’ conditions change.

The Matching Process

Each facility in the NYS survey was merged with an OS-CAR survey with the closest survey date, using the facility’sunique federal Medicare/Medicaid provider number. Facilitiesin NYS were surveyed in 1997, but information on the exactsurvey date was not available; instead, a uniform date, July 1,1997, was assigned to all facilities. All but 9 of the 410facilities in NYS were successfully matched with a closestOSCAR survey. Since freestanding and hospital-based facil-ities are different in their care orientation and staffing pat-terns, comparisons in this article were based on 327 freestand-

ing facilities among all the matched cases.

JAMDA – January/February 2005

Variables for Comparison

Staffing data in OSCAR are reported in the form of hoursworked in the 2-week period prior to the time of each stan-dard survey, separately for full-time, part-time, and contract,by staff category. A full-time employed staff is defined asworking 35 hours or more per week, and part-time if less than35 hours. The reported staff hours were converted to thenumber of full-time equivalents (FTEs) for each staff category,based on 35 hours per week over a 2-week period (ie, dividingthe total hours worked over a 2-week period by 70 hours).Total FTEs include both on-staff (full- or part-time) andcontract FTEs. The variables selected for comparisons pertainto physician staffing (full-time physicians, other than themedical director; physician extenders, including physicianassistants and nurse practitioners; and total hours the medicaldirector spends in the facility per week), nurse staffing (reg-istered nurses [RNs]; licensed practical nurses [LPNs]; certifiednurse aides [CNAs]; and average nurse hours per patient day,separately for RNs, LPNs, CNAs, and combined), and otherprofessional staffing (social workers; occupational therapists[OTs]; and physical therapists [PTs]). The corresponding mea-sures in NYS were defined in a comparable (though notalways identical) manner. The most obvious difference, forexample, lies in the fact that the facilities in NYS directlyreported the number of FTEs without giving informationabout actual hours worked. Definitions of each variable, alongwith differences in the wording of original survey questionsand instructions, are detailed in the appendix (Appendix 1).These nuances in measurement approaches are anticipated toaccount for part of the discrepancies in staffing levels asreported in the two data sets.

Analysis

Two statistical approaches were used. Since some staff, suchas full-time physicians (other than the medical director) andphysician extenders, are relatively rare in nursing homes, therelevant variables were dichotomized to indicate whether anysuch staff were available. Responses from NYS and OSCARto each variable were then cross-tabulated and agreementevaluated. For each cross-tabulation, the overall agreementrate is the total percent of cases where the two datasets agreewith each other (ie, all yes-yes and no-no pairs). The Kappacoefficient, a measure of inter-rater agreement, was reportedto indicate the strength of agreement: the closer the Kappa isto �1 (on a –1 to �1 scale), the stronger the agreement.31 For

Table 1. Availability of Selected Nursing Home Staff: Response Agre

Any full-time physician (other than medical director)Any physician extenderAny on-staff physician extender

NYS, New York State; OSCAR, Online Survey Certification and Rep* P � .01 for all Kappa coefficients.

staffing levels, correlation analysis was performed on the num-

ORIGINAL STUDIES

ber of FTEs or staff hours, conditional on their availability (ie,greater than zero) in both surveys. The Spearman rank-ordercorrelation was used, instead of Pearson, because the latter issensitive to outliers.

RESULTS

Table 1 presents the inter-survey agreement rates andKappa coefficients between the NYS survey and OSCARregarding the availability of selected physician staff, alongwith the percentage of facilities with each type of staff avail-able (as listed in the two columns under the headings “NYS”and “OSCAR”). In both surveys, about 8% of facilities re-ported having a full-time physician (other than the medicaldirector), and they concurred in the reporting of this infor-mation in 96% of cases. The NYS survey recorded morefacilities with physician extenders (28% with any and 17%with any on-staff) than did OSCAR (25% with any and 14%with any on-staff). The agreement rate was high on thereporting of physician extender availability (89% for any and93% for any on-staff). The Kappa coefficients in Table 1 allexceed 0.7, indicating strong inter-survey agreement in theavailability of these staff.

Table 2 reports the Spearman correlation coefficients onstaffing levels, along with variable means and standard devi-ations based on each survey. On average, the NYS surveyreported one FTE physician (other than the medical director)more than did OSCAR (4.5 and 3.4 FTEs, respectively). Thecorrelation for this variable was 0.66 (N � 20), indicatingreasonably high agreement. In conjunction with the results inTable 1, these findings suggest that the two data sources werein substantial agreement on the availability of full-time phy-sicians (other than the medical director), although they dif-fered substantially in the number of such physicians, if avail-able. Although the mean number of total physician extenders(N � 63) was higher in NYS (1.5 FTEs) than in OSCAR (1.1FTEs), the mean number of on-staff physician extenders (N �39) was about the same (1.5 FTEs); for both, the correlationwas quite reasonable (0.65 and 0.68, respectively). The twodatasets were moderately correlated on total medical directorhours per week (0.69), but the NYS survey reported morehours, on average, than did OSCAR (15 and 11 hours,respectively).

There was fairly strong correlation on the FTE numbers ofRNs, LPNs, and CNAs (0.78, 0.71, and 0.80, respectively). Interms of average nurse hours per patient day, however, the

Between the NYS Survey and OSCAR (N � 327)

S)

OSCAR(%)

Agreement(%)

Kappa*

.0 8.3 96.0 0.73

.1 24.8 88.7 0.71

.5 14.4 93.0 0.73

.

ement

NY(%

82816

orting

correlation was relatively weak (0.57 for RNs, 0.45 for LPNs,

Feng et al. 29

in bo

0.29 for CNAs, and 0.27 for all combined). There was mod-erate to strong correlation for social workers, OTs, and PTs(0.62, 0.63 and 0.68, respectively).

DISCUSSION

In light of different measurement approaches, different datacollection times and turnover rates among the administratorsin New York State, surprisingly, the staffing data from theNYS survey and from OSCAR appeared to be comparable.The correlation between the two data sources for most staffingvariables was fairly strong.

The assumption being made is that the NYS survey gener-ated more reliable and valid data than OSCAR because, as aresearch survey, the method and protocol were more precise.All in all, rather than defining one dataset or the other as the“gold standard,” the present study should be viewed from theclassic perspective of construct validity32 where each measure-ment approach is understood to generate its own unique“method variance” that distorts the validity and accuracy ofthe data. Quantifying and understanding this method vari-ance is essential to establishing the utility, biases, and limita-tions of any nursing home dataset, including the OSCAR.This is especially relevant to OSCAR, given its preeminencein long-term care policy, quality assurance, and research.

While the overall limitations of OSCAR staffing data arewell known, a number of specific measurement and designissues need to be addressed. For example, physician assistantsand nurse practitioners are lumped together in OSCAR, de-spite the fact that they are trained differently, perform differ-ent functions, and more importantly, they are licensed andsupervised differently.33 Therefore, data on these two types of

Table 2. Correlation Between the NYS Survey and OSCAR on Selec

NYS M(SD)

Physician staffing# Full-time physician (other than

medical director) FTEs 4.5 (3# Total physician extender FTEs 1.5 (1# On-staff physician extender FTEs 1.5 (1# Medical director hours per week 14.6 (1

Nurse staffing# RN FTEs 20.6 (2# LPN FTEs 23.4 (1# CNA FTEs 75.4 (5Average RN hours per patient day 0.5 (0Average LPN hours per patient day 0.7 (0Average CNA hours per patient day 2.0 (0Average total nurse hours per patient

day3.2 (1

Other professional staffing# Social worker FTEs 2.6 (2# Occupational therapist FTEs 1.1 (1# Physical therapist FTEs 1.3 (1

NYS, New York State; FTEs, full time equivalents; OSCAR, Onlinlicensed practical nurses; CNA, certified nurse aides.

* P � .01 for all Spearman correlation coefficients.† Number of facilities with # FTEs or staff hours greater than zero

staff should be collected separately. Besides, the licensure

30 Feng et al.

varies from state to state, making it very difficult to comparephysician extenders across the states. It would make bettersense to define the categories for physician extenders based onlevel of supervision (whether a physician must be directlysupervising) and how they function in addition to the type ofphysician extender. Furthermore, a clinical nurse specialistcategory should be added to the OSCAR database becausethey may be hired by the facility as a consultant.

Potential ambiguities in staff definitions need to be scruti-nized and corrected. For example, in OSCAR “other physi-cian” is defined as “a salaried physician, other than the med-ical director, who supervises the care of residents when theattending physician is unavailable, and/or a physician(s)available to provide emergency services 24 hours a day.” Froma clinical perspective, this definition may be confusing andperhaps irrelevant—taken on face value, who really caresabout physician backup for the medical director (ie, coveringfor emergencies)? It may be more important to know thenumber of physicians actually caring for patients. To be fair,the differences in physician FTEs may not be real but speak tothe inadequacy of OSCAR definitions.

A number of recommendations are proposed here to refinethe collection of staffing data through OSCAR. First, it isrecommended that staffing data be collected quarterly. Thismay overlap most facilities’ current cycle of quarterly staffingexpenditure reports within their own fiscal year. For consum-ers examining the CMS web site (Nursing Home Compare),quarterly data would be a better reflection of the consistencyof staffing levels in a given home. For the inspectors, thisinformation would be in a form more readily subject to auditand not subject to the “staffing up” phenomenon in anticipa-

rsing Home Staffing Variables

OSCAR Mean(SD)

SpearmanCorrelation*

N†

3.4 (2.9) 0.66 201.1 (1.6) 0.65 631.5 (1.9) 0.68 39

11.0 (11.3) 0.69 277

18.9 (18.3) 0.78 31222.4 (18.0) 0.71 31071.2 (52.4) 0.80 3120.5 (0.2) 0.57 3100.6 (0.3) 0.45 3092.0 (0.5) 0.29 3113.1 (0.8) 0.27 312

2.1 (2.5) 0.62 3050.9 (1.4) 0.63 2751.3 (1.6) 0.68 310

ey Certification and Reporting System; RN, registered nurse; LPN,

th NYS and OSCAR.

ted Nu

ean

.7)

.4)

.4)3.4)

4.1)9.1)6.7).4).3).6).1)

.2)

.4)

.6)

e Surv

tion of annual inspections, as many have speculated.34 For

JAMDA – January/February 2005

researchers and for continuous quality improvement, theavailability of quarterly staffing data would allow for greaterprecision in addressing the relationship between staffing andquality. Second, the collection of staff turnover data should beconsidered in OSCAR, given high turnover rates and theirimpact on resident care.35 Finally, staffing data should beaudited by CMS regional offices to ensure accuracy.

In the nursing home setting, medical directors are respon-sible for implementing resident care policies and coordinatingmedical care in the facility. They can also help ensure thatregulatory requirements are consistent with evidence-basedclinical information and therefore should play a more activerole in the survey process.29 Moreover, important questionsregarding physician practice in nursing homes, such as therelationship between staffing and clinically relevant outcomesremain unanswered. While previous work has demonstrated acorrelation between medical staff organization (ie, closed vs.open staff) and intensity of care,36 the relationship to qualityhas yet to be addressed. The putative relationship amongphysician practice, quality care, and nurse efficiency/turnover,as described by some observers, highlights the importance ofthe physician in nursing home care as well as the need formore research in this arena.

This study also examined the widely used measures ofaverage nurse hours per patient day, separately by nurse cat-egory and combined. Although these measures were close atthe means between the two data sources, their correlation wasrelatively weak, suggesting that these measures can not alwaysbe substantiated using alternative data sources. Given theenormous policy implications of these measures, it is recom-mend that CMS conduct a more comprehensive and rigorousreview of the accuracy of OSCAR staffing data as well as theappropriateness of using such data for public reporting andstandard setting purposes.

CMS, pushed by advocates, has already begun to publishnursing home staffing data from OSCAR on its “NursingHome Compare” Web site (http://www.medicare.gov/NHCompare). Results from this analysis are consistent withothers in demonstrating that these data are not really accurateon a case specific basis.3,5 Yet the staffing information pro-vided by OSCAR is critically important, given the observedassociations between lower staffing levels and poor quality ofcare outcomes.3,11,12,21–28 Reliable and valid nursing homestaffing data are also necessary to understand the impact offederal policies and to properly calibrate the level of staffingneeded to achieve minimal levels of resident outcome quality.Unfortunately, at this time neither CMS nor the states at-tempt to verify the accuracy of the staffing data on a regularbasis,4 on the grounds that it is unrealistic to incorporateverification of staffing levels into the survey process based onthe current state of facility records.3 In order to make theOSCAR data more valuable for public reporting efforts, it isincumbent upon the government to design a more compre-hensive and systematically structured mechanism for docu-menting staffing information. To this end, the recommenda-tions proposed above should be considered as the first

necessary step.

ORIGINAL STUDIES

CONCLUSION

The OSCAR data appear to be useful for exploring rela-tionships in the nursing home setting, although not necessar-ily representing “truth” in public reports that characterize theperformance of individual facilities. Results from this analysisand from previous studies should not be interpreted as anindictment of OSCAR, but they do point to the need tocontinue examining the construct validity of OSCAR and itslimits, and to fine-tune its content and methodology. As thetheory and empirical data in long-term care become morerefined and advanced, it is essential that OSCAR, as a datasystem, keeps pace, if it is to continue to inform rationaldecision-making at the consumer and facility level and effec-tive policy formation at the state and national level.

REFERENCES1. General Accounting Office (GAO). Nursing Workforce: Emerging

Nurse Shortages Due to Multiple Factors. GAO-01-944. Washington,DC: GAO, 2001. Available at: http://www.gao.gov/new.items/d01944.pdf. Accessed December 23, 2004.

2. Department of Health and Human Services’ Office of the AssistantSecretary for Planning and Evaluation (ASPE), the Centers for Medicareand Medicaid Services (CMS), and Health Resource and Services Ad-ministration (HRSA) and the Department of Labor’s Office of theAssistant Secretary for Policy, Bureau of Labor Statistics (BLS) andEmployment and Training Administration (ETA). The Future Supply ofLong-term Care Workers in Relation to the Aging Baby Boom Gener-ation: Report to Congress. 2003. Available at: http://www.aspe.hhs.gov/daltcp/reports/Itcwork.pdf. Accessed December 23, 2004.

3. Centers for Medicare and Medicaid Services (CMS). Report to Congress:Appropriateness of Minimum Nurse Staffing Ratios in Nursing Homes,Phase II. Baltimore, MD: CMS, 2001. Available at: http://www.cms.hhs-.gov/medicaid/reports/rp1201home.asp. Accessed December 23, 2004.

4. General Accounting Office (GAO). Skilled Nursing Facilities: AvailableData Show Average Nursing Staff Time Changed Little After MedicarePayment Increase. GAO-03-176. 2002. Available at: http://www-.gao.gov/new.items/d03176.pdf. Accessed December 23, 2004.

5. Health Care Financing Administration (HCFA). Report to Congress:Appropriateness of Minimum Nurse Staffing Ratios in Nursing Homes,Phase I. 2000. Available at: http://www.cms.hhs.gov/medicaid/reports/rp700hmp.asap. Accessed December 23, 2004.

6. Harrington C, Kovner C, Mezey M, et al. Experts recommend minimumnurse staffing standards for nursing facilities in the United States. Ger-ontologist 2000;40:5–16.

7. Office of Inspector General (OIG). Quality Assurance Committees inNursing Homes. OEI-01-01-00090. 2003. Available at: http://www.oig.hhs.gov/oei/reports/oei-01-01-00090.pdf. Accessed December 23,2004.

8. Angelelli J, Mor V, Intrator O, et al. Oversight of nursing homes:Pruning the tree or just spotting bad apples? Gerontologist 2003;43:67–75.

9. Castle NG. Providing outcomes information to nursing homes: Can itimprove quality of care? Gerontologist 2003;43:483–492.

10. Harrington C, Carrillo H. The regulation and enforcement of federalnursing home standards, 1991–1997. Med Care Res Rev 1999;56:471–494.

11. Harrington C, Zimmerman D, Karon SL, et al. Nursing home staffing andits relationship to deficiencies. J Gerontol B Psychol Sci Soc Sci 2000;55:S278–287.

12. Zhang X, Grabowski DC. Nursing home staffing and quality under thenursing home reform act. Gerontologist 2004;44:13–23.

13. Zinn JS, Mor V, Intrator O, et al. The impact of the prospective paymentsystem for skilled nursing facilities on therapy service provision: A

transaction cost approach. Health Serv Res 2003;38:1467–1485.

Feng et al. 31

14. Harrington C, Carrillo H, Mullan J, Swan JH. Nursing facility staffing inthe states: The 1991 to 1995 period. Med Care Res Rev 1998;55:334–363.

15. Harrington C, O’Meara J, Collier E, Schnelle JF. Nursing indicators ofquality in nursing homes. A Web-based approach. J Gerontol Nurs2003;29:5–11.

16. Schnelle JF. Determining the relationship between staffing and quality.Gerontologist 2004;44:10–12.

17. Straker JK. Reliability of Oscar Occupancy, Census and Staff Data: AComparison with the Ohio Department of Health Annual Survey ofLong-Term Care Facilities. Oxford, OH: Scripps Gerontology Center,Miami University, 1999, pp. 3–01.

18. Aaronson WE, Zinn JS, Rosko MD. Do for-profit and not-for-profitnursing homes behave differently? Gerontologist 1994;34:775–786.

19. Harrington C, Swan JH. Nursing home staffing, turnover, and case mix.Med Care Res Rev 2003;60:366–392, discussion 393–369.

20. Zinn JS. Market competition and the quality of nursing home care.J Health Polit Policy Law 1994;19:555–582.

21. Agency for Healthcare Research and Quality (AHRQ). The Effect ofHealth Care Working Conditions on Patient Safety. Washington, DC:US Department of Health and Human Services, 2003, pp. 74.

22. General Accounting Office (GAO). Nursing Homes: Quality of CareMore Related to Staffing than Spending. GAO-02-431R. Washington,DC: GAO, 2002. Available at: http://www.gao.gov/new.items/d02431r.pdf. Accessed December 23, 2004.

23. Institute of Medicine (IOM). Nursing Staff in Hospitals and NursingHomes: Is it adequate? Washington, DC: National Academy Press, 1996.

24. Kayser-Jones J, Schell E, Lyons W, et al. Factors that influence end-of-lifecare in nursing homes: The physical environment, inadequate staffing,

and lack of supervision. Gerontologist 2003;43:76–84.

32 Feng et al.

25. Munroe DJ. The influence of registered nurse staffing on the quality ofnursing home care. Res Nurs Health 1990;13:263–270.

26. Office of Inspector General (OIG). Quality of Care in Nursing Homes:An Overview. Washington, DC: US Department of Health and HumanServices, 1999.

27. Schnelle JF, Simmons SF, Harrington C, et al. Relationship of nursinghome staffing to quality of care. Health Serv Res 2004;39:225–250.

28. Spector WD, Takada HA. Characteristics of nursing homes that affectresident outcomes. J Aging Health 1991;3:427–454.

29. American Medical Directors Association (AMDA). White paper on thesurvey process. 2002. M02. Available at: http://www.amda.com/library/whitepapers/survey.htm. Accessed December 23, 2004.

30. Dillman D. Mail and Telephone Surveys: The Total Design Method.New York: John Wiley and Sons, 1978.

31. Guggenmoos-Holzmann I. The meaning of kappa: Probabilistic conceptsof reliability and validity revisited. J Clin Epidemiol 1996;49:775–782.

32. Campbell DT, Fiske DW. Convergent and discriminant validation bythe multitrait-multimethod matrix. Psychol Bull 1959;56:81–105.

33. Hooker RS, Berlin LE. Trends in the supply of physician assistants andnurse practitioners in the United States. Health Aff (Millwood) 2002;21:174–181.

34. Office of Inspector General (OIG). Nursing Home Survey and Certifi-cation: Overall Capacity. 1999. Available at: http://www.oig.hhs.gov/oei/reports/oei-02-98-00330.pdf. Accessed December 23, 2004.

35. Zimmerman S, Gruber-Baldini AL, Hebel JR, et al. Nursing home facilityrisk factors for infection and hospitalization: Importance of registerednurse turnover, administration, and social factors. J Am Geriatr Soc2002;50:1987–1995.

36. Karuza J, Katz PR. Physician staffing patterns correlates of nursing homecare: An initial inquiry and consideration of policy implications. J Am

Geriatr Soc 1994;42:787–793.

JAMDA – January/February 2005

APPENDIX 1

Definition of Variables According to the NYS Survey and OSCAR

Variable NYS OSCAR

Full-time physicians, otherthan the medical director

“How many physicians working full-time at thefacility, other than yourself (medicaldirector), currently receive a salary from thenursing home?” Measured innumbers/counts.

Full-time other physicians, where“other physician” refers to “asalaried physician, other thanthe medical director, whosupervises the care of residentswhen the attending physician isunavailable, and/or aphysician(s) available to provideemergency services 24 hours aday.” Measured in FTEs.

Total physician extenders Sum of the following, in FTEs:a. “How many full time equivalent (FTE) nurse

practitioners primarily involved in clinicalactivities are on your staff?”

b. “How many full time equivalent (FTE) nursepractitioners are on contract to provideclinical services?”

c. “How many full time equivalent (FTE)physician assistants are on your staff?”

d. “How many full time equivalent (FTE)physician assistants are on contract toprovide service?”

Physician extender refers to “anurse practitioner, clinical nursespecialist, or physician assistantwho performs physiciandelegated services.” Note thatthese staff are combined in asingle category and notseparately reported. Measuredin total FTEs (staff or contract).

On-staff physician extenders Sum of (a) and (c) above. On-staff FTEs, following abovedefinition.

Medical director hours perweek

“How many total hours in a typical week doyou (medical director) spend in this facility?”

Total medical director FTEsmultiplying 35 hours.

Registered nurses “What is the actual number of FTE registerednurses currently in your facility (including thedirector of nursing, “agency”, and per diemnurses)?”

Sum of the following, in FTEs (on-staff or contract):

a. Registered nursesb. Director of nursingc. Nurses with administrative

dutiesLicensed practical nurses “What is the actual number of FTE LPNs

currently in your facility (include “agency”and per diem nurses)?”

Number of LPN FTEs (staff orcontract).

Nurse aides “What is the actual number of FTE nursingaides currently in your facility (include“agency” and per diem nursing aides)?”

Number of certified nurse aideFTEs (staff or contract).

Average RN hours perpatient day

Total RN FTEs multiplying 5 hours divided bytotal number of residents.

Same as NYS.

Average LPN hours perpatient day

Total LPN FTEs multiplying 5 hours divided bytotal number of residents.

Same as NYS.

Average CNA hours perpatient day

Total CNA FTEs multiplying 5 hours divided bytotal number of residents.

Same as NYS.

Average total nurse hoursper patient day

Combined total RN, LPN, and CNA hoursmultiplying 5 hours divided by total numberof residents

Same as NYS.

Social workers Number of FTE social workers (staff orcontract)

Same as NYS.

Occupational therapists Number of FTE occupational therapists (staff orcontract)

Same as NYS.

Physical therapists Number of FTE physical therapists (staff orcontract)

Same as NYS.

NYS, New York State; OSCAR, Online Survey Certification and Reporting System; FTEs, full time equivalents; RN, registered nurse; LPN,licensed practical nurses; CNA, certified nurse aides.

ORIGINAL STUDIES Feng et al. 33