Full text of DOI:10.1093/geronb/55.5.S278

Journal of Gerontology: SOCIAL SCIENCES
Copyright 2000 by The Gerontological Society of America
2
000, Vol. 55B, No. 5, S278–S287
Nursing Home Staffing and
Its Relationship to Deficiencies
1
2
2
Charlene Harrington, David Zimmerman, Sarita L. Karon,
2
2
James Robinson, and Patricia Beutel
1Department of Social and Behavioral Sciences, University of California, San Francisco.
2Center for Health Systems Research and Analysis, University of Wisconsin, Madison.
Objectives. The authors examined the relationships between different types of nursing home staffing and nursing
home deficiencies to test the hypothesis that fewer staff hours would be associated with higher numbers of deficiencies.
Methods. Data were from the On-Line Survey, Certification, and Reporting System for all certified nursing homes in
the United States. Regression models examined total deficiencies, quality of care, quality of life, and other deficiencies.
Results. Fewer registered nurse hours and nursing assistant hours were associated with total deficiencies and quality
of care deficiencies, when other variables were controlled. Fewer nursing assistant staff and other care staff hours were
associated with quality of life deficiencies. Fewer administrative staff hours were associated with other deficiencies. Fa-
cilities that had more depressed and demented residents, that were smaller, and that were nonprofit or government-
owned had fewer deficiencies. Facilities with more residents with urinary incontinence and pressure sores and with
higher percentages of Medicaid residents had more deficiencies, when staffing and resident characteristics were con-
trolled.
Discussion. Facility characteristics and states were stronger predictors of deficiencies than were staffing hours and
resident characteristics. Because only a small portion of the total variance in deficiencies could be explained, much work
remains to explore factors that influence deficiencies.
SING registered nurses (RNs) and other nursing per-
sonnel as well as many other types of health profes-
proved resident functional outcomes. A recent study of
Minnesota nursing homes found that in the 1st year after a
patient’s admission to a nursing home, the licensed (but not
nonlicensed) nursing hours were significantly related to im-
proved functional ability, increased probability of discharge
home, and decreased probability of death (Bliesmer, Kane,
& Shannon, 1998). The role of professional nursing staff
disappeared, however, for chronic care residents.
Few studies have examined other types of care staff in
nursing homes such as physician and therapy services.
Some of these have found a positive relationship between
such services and outcomes (Chiodo, Gerety, Mulrow,
Rhodes, & Tuley, 1992; Karuza & Katz, 1994; Kochers-
berger, Hielema, & Westlund, 1994; Mulrow et al., 1994;
Przybylski et al., 1996).
In this study we had two aims. First, we described and
categorized the types of federal deficiencies issued to facili-
ties by state licensing and certification surveyors for differ-
ent types of care problems. Deficiencies are evaluations of
poor quality made by state surveyors under the federal nurs-
ing home certification regulations. Second, we used multi-
ple regression analyses to test the association of facility de-
ficiencies with facility staffing hours per resident day,
resident characteristics, facility characteristics, and states.
The general hypothesis we tested was that staff hours per
resident day in nursing homes would be negatively associ-
ated with higher numbers of deficiencies, when controlling
statistically for resident characteristics, facility characteris-
tics, and states.
U
sionals and care providers, nursing homes in the United
States provide care to about 1.6 million residents (Har-
rington, Carrillo, Thollaug, & Summers, 2000). Studies
have documented the importance of nursing in both the pro-
cess of care and the outcomes of nursing home care (Aaron-
son, Zinn, & Rosko, 1994; Cherry, 1991; Linn, Gurel, &
Linn, l977; Munroe, 1990; Nyman, 1988; Spector &
Takada, l991). Nyman (1988) found that nurse hours per pa-
tient day were positively related to facility quality measures.
Munroe (1990) found a positive significant relationship be-
tween the quality (measured by deficiencies) and higher ra-
tios of RN and licensed vocational or licensed practical
nurse (LVN/LPN) hours per resident day and concluded
that higher ratios of licensed nurses may be more important
than total nursing hours. Cherry (1991) found that increased
RN hours were positively associated with a composite of
good outcome measures (fewer decubitus ulcers, catheter-
ized residents, and urinary tract infections and less antibi-
otic use).
Spector and Takada (1991) reported that higher staff lev-
els and lower RN turnover rates were related to improve-
ments in resident functioning. Lower staffing was associ-
ated with high rates of urinary catheter use, low rates of skin
care, and low resident participation in activities. Cohen and
Spector (1996) found that higher ratios of RNs to residents,
adjusted for resident case mix, reduced the likelihood of
death and that higher ratios of LVN/LPNs significantly im-
S278

NURSING HOME STAFFING
S279
Deficiencies
lationships and hypotheses are presented as each indepen-
dent variable is discussed.
Deficiencies are issued to facilities by state surveyors as a
part of the federal survey process. The survey process is de-
signed to regulate quality of care, and the deficiencies repre-
sent surveyor evaluations about the type of quality problems
that exist. State survey agencies operate under contract with
the Health Care Financing Administration (HCFA) to moni-
tor nursing facilities on a periodic basis, about once a year
Staffing hours.—Federal standards have been estab-
lished for different types of staff in nursing facilities as a
means of ensuring adequate nursing home care. Nursing
home reform legislation in the Omnibus Reconciliation Act
of 1987 increased the minimum standards for nursing home
staffing. All nursing homes certified for Medicare and Med-
icaid residents must have an RN director of nursing, an RN
on duty for 8 hours a day 7 days a week, and a licensed
nurse (either an RN or LVN/LPN) on duty around the clock
(
between 9 and 15 months). These evaluations are the only
routinely collected and external source of information about
quality for all facilities. The deficiencies are entered in the
federal On-Line Survey, Certification, and Reporting (OS-
CAR) system by state staff after the surveys. If a facility
fails to meet a standard or regulation, a deficiency or cita-
tion is given to the facility (coded as 1 for a deficiency and 0
for no deficiency). OSCAR reports whether a deficiency is
issued for each survey item and the total number of defi-
ciencies issued to each facility.
(
1
Omnibus Budget Reconciliation Act of 1987 [OBRA],
987). In addition, the law requires sufficient nursing staff
to provide nursing and related services to residents to enable
residents to reach or maintain the highest practicable level
of physical, mental, and psychosocial well-being. Regula-
tions also require social activities; medically related social
services; dietary services; physician and emergency care
services; and pharmacy, dental, and rehabilitation services
including physical, speech, and occupational therapies (OBRA,
The OSCAR deficiency data are generally considered to
be accurate reflections of the actual deficiencies issued by
surveyors. The data also accurately reflect problems that ex-
ist in the facilities because deficiencies are subject to exten-
sive review by state officials and facilities and are subject to
appeal by providers through an administrative or judicial
process. Because providers do not want deficiencies on their
record, they challenge deficiencies they consider unwar-
ranted.
1
987; U.S. Department of Health and Human Services,
Health Care Financing Administration, 1995a, 1995b, 1995c).
Facilities have discretion in going beyond the minimal fed-
eral requirements in the numbers and types of staff they use
to provide care. Staffing types and hours vary considerably
across facilities (Harrington & Carrillo, 1999; Harrington,
Carrillo, Mullan, & Swan, 1998).
The federal regulations have 179 specific standards for
nursing facility care in 17 major categories. In this study,
we analyzed the total number of deficiencies for each facil-
ity. In addition, we grouped the deficiencies into three cate-
gories: (a) quality of care, (b) quality of life, and (c) other
deficiencies. Quality of care included 72 specific items in
the following federal survey categories: resident assess-
ment, quality of care, nursing services, dietary services,
physician services, rehabilitative services, dental services,
pharmacy services, and infection control. For example, pre-
vention of pressure sores, falls, and prevention of physical
decline are separate requirements under the quality of care
category. The quality of life category included 77 specific
items on resident’s rights; admission, transfer, and dis-
charge rights (including resident rights); resident behavior
and facility practices (includes resident rights); quality of
life; and physical environment (which could affect quality
of life). Other deficiencies included 30 specific items on ad-
ministration and medical records. Because surveyors have
some discretion in the specific individual deficiencies that
they apply to quality problems that they detect, grouping de-
ficiencies into a few broad categories has the advantage of
reducing some variability inherent in the survey process.
The three broad categories we used appeared to be concep-
tually separate and easier to analyze that the 17 categories
used by HCFA.
The major focus of this study was staffing hours per resi-
dent day in nursing facilities. The staffing was examined for
six separate categories. First, nursing staff was categorized
as all RN hours including nurse administrators, LVN/LPN,
and nursing assistant hours separately. Other care staff in-
cluded the professional staff and their assistants (e.g., physi-
cians, pharmacists, dietary staff, therapists, activities staff,
and social workers). All facility administrative staff (ex-
cluding nursing) was combined into a separate category.
Housekeeping and other staff formed a fourth category. Fi-
nally, all these groups of staff were combined for the total
staff in nursing facilities.
The general hypothesis was that staffing hours would be
inversely related to deficiencies, when resident characteris-
tics, facility characteristics, and states were controlled. Spe-
cifically, we hypothesized that fewer nursing staff hours per
resident day would be associated with higher numbers of to-
tal deficiencies and with quality of care and quality of life
deficiencies. We expected that fewer other care staff hours
(
e.g., therapists and activities staff) per resident day would
be related to more quality of care and quality of life defi-
ciencies. We expected that fewer administrative and house-
keeping staff hours per resident day would be related to
more deficiencies in the “other” deficiency category.
Resident characteristics.—Nursing homes vary in the
types of residents they serve, typically referred to as case
mix variation. A strong relationship among resident charac-
teristics, nurse staffing time requirements, and nursing costs
in nursing facilities has been shown in several studies (Ar-
ling, Nordquist, & Capitman, 1987; Fries & Cooney, 1985;
Independent Variables
We expected that four important types of factors would
be associated with deficiencies in nursing facilities: (a)
staffing hours; (b) resident characteristics; (c) facility char-
acteristics; and (d) state or regional factors. The specific re-

S280
HARRINGTON ET AL.
Fries et al., 1994; Schneider, Fries, Foley, Desmond, &
Gormley, l988). These studies have shown the importance
of activities of daily living (ADLs) and other resident char-
acteristics on facility resource requirements.
3. Certification status. There are four federal certification
categories of nursing facilities: skilled nursing facilities
(SNFs) for Medicare only (Category 4); nursing facilities
for Medicaid with a distinct part for SNF care (Category
3); nursing facilities dually certified for Medicaid and
Medicare (Category 2); and nursing facilities certified
for Medicaid only (Category 10). Medicare-only SNFs
(Category 4) have been shown to be related to high levels
of resident needs and staffing (Harrington et al., 1998;
Kanda & Mezey, 1991), because of residents’ eligibility
for Medicare skilled nursing reimbursements. Medicare
payment levels have traditionally been higher than those
in Medicaid facilities (Buchanan, Madel, & Persons,
1991). Because SNFs have higher payment rates than
other certification categories, they should be negatively
related to deficiencies, whereas Medicaid-only facilities,
which have the lowest reimbursement rates, may be pos-
itively associated with deficiencies. Controlling for staff-
ing hours and resident characteristics, we expected that
SNFs would have fewer deficiencies because of a num-
ber of factors not captured in the model, such as the ex-
pertise and experience of direct care staff, organizational
and administrative expertise, facility support services,
and other unmeasured factors.
Resident characteristics or case mix may increase the
number of deficiencies that a facility receives, when staff
hours and other factors are controlled. This may occur if
surveyors identify the resident characteristics or case mix
when they begin the survey and focus their survey atten-
tion on specific residents with problems. We expected that
resident characteristics such as functional impairment,
cognitive and emotional status, incontinence, and pressure
sores would be positively related to deficiencies, because
residents with these problems are probably at greater risk
of poor outcomes and generally require more effort and
expertise in caregiving. The hypothesis we tested was that
higher levels of case mix would be positively associated
with the number of deficiencies in total and with quality of
life and quality of care deficiencies in particular, when
staffing hours, facility characteristics, and states were con-
trolled. Because of the potential complex relationship be-
tween resident characteristics and deficiencies, we exam-
ined deficiencies with and without resident characteristics
in the models.
4
. Percentage of Medicaid residents. Many state Medicaid
programs have attempted to keep their level of Medi-
caid nursing home reimbursement rates low, and Medicaid
rates are generally lower than those of Medicare (Aaron-
son, Zinn, & Rosko, 1995; Buchanan et al., 1991; Cohen
Facility characteristics.—We examined the following
facility characteristics: size, whether or not a facility was
hospital based, certification status, percentage of Medicaid
residents, and type of ownership.
&
Dubay, 1990; Dor, 1989; Holahan & Cohen, l987;
1
. Size. The number of beds in a nursing home facility
size) is expected to be directly related to deficiencies.
Holahan, Rowland, Feder, & Heslam, 1993; Swan, Har-
rington, & Grant, 1993). Studies have shown that high
percentages of Medicaid residents result in lower staffing
levels and conversely that high percentages of Medicare
residents result in higher staffing levels (Fottler et al.,
1981; Harrington et al., 1998; Nyman, 1988; Zinn,
1993). Nyman (l988) found that facilities with greater
proportions of Medicaid residents appeared to have
lower quality of care. Facilities with greater proportions
of Medicaid residents may not have to compete on qual-
ity, and such facilities may have fewer resources because
Medicaid reimbursement rates are lower than those of
Medicare. Therefore, we hypothesized that higher per-
centages of Medicaid residents would be positively asso-
ciated with deficiencies even after controlling for staff-
ing hours, resident characteristics, and other factors.
(
Large facilities have been associated with higher staff-
ing, although findings have been mixed (Davis, 1993;
Nyman, 1988; Ullmann, 1987). Other studies have found
a negative relationship between size and staffing (Fottler,
Smith, & James, 1981). Controlling for both staffing
hours and resident characteristics, we expected that
larger sized nursing facilities would be related to defi-
ciencies because larger facilities may be more difficult to
manage and/or may result in a less positive environment
for residents.
. Hospital-based nursing facilities. Hospital-based nursing
facilities have traditionally had substantially higher nurs-
ing staff hours and other professional staff hours (thera-
pists, dietitians, pharmacists, and others) because hos-
pital-based residents may have higher acuity levels.
Moreover, hospital-based facilities generally may have
more Medicare residents and may receive higher reim-
bursement rates than Medicaid residents in nursing
homes. Because hospital-based facilities may have more
financial and staff resources, we expected that they
would have fewer deficiencies. Hospital-based facilities
may have fewer deficiencies because of unmeasured
variables, such as organizational philosophy, training
and skill of staff, experience in caring for complex resi-
dents, and/or surveyor expectations. We therefore hy-
pothesized that hospital-based nursing facilities would
have fewer deficiencies, controlling for staffing hours,
resident characteristics, and other factors.
2
5. Ownership. Nonprofit facilities have been reported to
have better nursing home outcomes in some studies
(Aaronson, et al., 1994; Davis, 1993; Spector, Selden, &
Cohen, 1998), so they should have fewer deficiencies.
We expected that nonprofit and government facilities
would have fewer deficiencies compared with for-profit
facilities, controlling for staffing hours, resident charac-
teristics, and other factors. We expected that unmeasured
variables for nonprofit facilities, such as organizational
philosophy, training and skill of staff, and turnover of
staff, may all result in fewer deficiencies.
State or regional factors.—Although the deficiency data
are generally considered accurate, there are some variations

NURSING HOME STAFFING
S281
in the surveyor procedures and practices for determining de-
ficiencies across the 50 states and the District of Columbia.
The U.S. General Accounting Office (1998a, 1998b) has
been critical of the state survey agencies in their poor identi-
fication of quality of care problems and their poor records in
enforcing the federal standards. The U.S. Department of
Health and Human Services, Health Care Financing Admin-
istration (1998a, 1998b) also identified problems with its
survey process and has been attempting to improve the pro-
cess. Thus, state and regional variables were included as
control variables in separate models. No hypotheses were
made about particular states.
skewed distribution, all facilities that reported staffing of
more than two standard deviations above the mean were re-
moved (159 facilities). Thus, the total number of facilities in
the study was 13,770. Excluded facilities were smaller,
more likely to be hospital based, more likely to be SNFs,
and had a lower proportion of Medicaid residents.
Each of the categories of staff included all the full-time,
part-time, and contract staff for that category. To compute
the staffing hours per resident day, we multiplied the total
staffing payroll full-time employees reported for a 2-week
period by 70 h and then divided by the total number of resi-
dents and by the 14 days in the reporting period (the proce-
dure used by HCFA). The staffing data were collected at the
time of each regular survey and were not audited by state
surveyors.
METHODS
Data Sources
The secondary data used in this study were drawn from the
federal OSCAR system for all certified nursing homes in the
United States. OSCAR data are collected in three separate
sets of files: (a) facility characteristics and staffing data, (b)
resident characteristics, and (c) survey deficiencies. The OSCAR
data are collected during the certification surveys conducted
every 9–15 months by state agencies under contract with
HCFA. Data in this article are from the surveys conducted
during the period July 1, 1995, through June 30, 1996. This
time period was selected because it was after HCFA imple-
mented procedural changes in the survey on July 1, 1995,
mandated by OBRA (1987) and because the data were the
most recent 12-month statistics available from HCFA at the
time of the study. Results from two sets of analyses for the first
Resident characteristics.—Several variables were se-
lected for resident characteristics from the OSCAR file. We
summed the scores for each of three ADLs (eating, toilet-
ing, and transferring) to construct an ADL index (1 ϭ low-
est need for assistance, 2 ϭ moderate need, 3 ϭ greatest
need). Thus, the summary ADL scale ranged from 3 to 9
with a score of 9 being the most dependent. Three mobility
variables were used for a mobility index: 1 ϭ independent
in ambulation, 2 ϭ in a chair all or most of the time, and 3 ϭ
bedfast all or most of the time, based on the number of resi-
dents at each level. The percentages of residents reported
with depression, dementia, behavioral symptoms, inconti-
nence, and pressure sores were calculated for the study.
6
and second 6 months of implementation did not differ sig-
Other independent variables.—For the size of a nursing
facility, the total beds were reported from OSCAR in a set
of four dichotomous variables: those with less than 60 beds,
with 60–119 beds, with 120–160 beds, and with more than
160 beds. A dichotomous variable indicated whether or not
the facility was hospital based. For certification status, nurs-
ing facilities for Medicaid with a distinct part for SNF care
(Category 3) and nursing facilities dually certified for Med-
icaid and Medicare (Category 2) were combined and used
as the comparison category. We used three dichotomous
variables to indicate each facility’s certification for SNFs,
facilities with a combination of Medicare and Medicaid res-
idents, and facilities with Medicaid-only residents. Non-
profit and government-owned facilities were combined and
indicated by a dichotomous variable and compared with
profit-making facilities. We included state dummy variables
in the model to take into account state variation in deficien-
cies. Census regions were indicated by a set of four dichoto-
mous variables (midwest, northeast, south, and west) and
were examined separately from states.
nificantly, so only the 12-month period is shown. The data for
the study were limited to the OSCAR data because these data
were available for all facilities in the United States and other
data sources were not available.
Data Cleaning
All federally certified nursing facilities for Medicare and
Medicaid in the United States were included in this study,
except for the trust territories and Puerto Rico. Intermediate
care facilities for the mentally retarded were excluded. We
eliminated duplicate provider records by matching facility
records by state, city, facility name, address, and telephone
number (372 facilities were eliminated). Another 1,496 fa-
cilities with survey dates prior to or after the study period
were removed. Facilities with incomplete information were
also removed (26 facilities). At that point, the resulting sam-
ple was 15,536 facilities. We corrected errors in reporting of
total beds and residents by setting the maximum number to
the number of certified skilled nursing beds in the facility
(for 2,315 facilities).
Analytic Approach
Staffing data.—We cleaned staffing data using the total
We used bivariate analyses to examine relationships be-
tween key variables and deficiencies. Pearson product cor-
relations were conducted between the independent vari-
ables. Although some of the independent variables showed
relationships with each other, the correlations were modest,
suggesting that multicollinearity was not likely to be prob-
lematic. Tolerance tests in the regression analysis also did
not show multicollinearity to be a problem.
nursing data. Because facilities with 15 or fewer residents
had significantly higher staffing hours than did larger facili-
ties, to an extent judged to be unlikely to be accurate, 1,054
such facilities were eliminated. We then used means and
standard deviations to examine staffing for the four certifi-
cation categories. Because the staffing data showed a num-
ber of outliers that appeared to be erroneous and resulted in

S282
HARRINGTON ET AL.
There is no general agreement on the best way to measure
Table 1. Means and Standard Deviations for Dependent and
Independent Variables for 1995–96
and analyze deficiencies, so we tested several approaches.
We used ordinary least squares (OLS) regression modeling
with SPSS (1998) to examine factors associated with the
number of deficiencies issued to each facility to test the hy-
potheses for the study. We inspected the residual normality
of the total deficiencies distribution and found that there
was no strong violation of the normality assumptions for
OLS modeling. In this case, the distribution of residuals was
found to be more symmetric than the dependent variable.
The skewness in the dependent variable was due largely to
skewness in the distribution of the independent variables
and not to non-normality of the residuals.
We found the same distribution pattern when we used a
logistic regression to analyze whether any deficiency was
cited (deficiency ϭ 1, no deficiency ϭ 0). Then, using the
log of deficiencies, we computed an OLS regression on the
deficiencies excluding those facilities with zero deficien-
cies. The findings, which were similar, are reported in the
Results section.
Characteristic and Variable
M
SD
Deficiencies
Quality of care
Quality of life
Other
3.52
2.06
.34
3.75
2.51
.77
Total
5.92
6.17
Staff
Total staff (hours per resident day)
Nursing staff (hours per resident day)
Registered nurses
5.70
3.40
.59
2.93
1.77
.73
LVN/LPNs
.67
.62
Nursing assistants
2.14
1.26
.28
1.14
.99
Other care staff (hours per resident day)
Administrative staff (hours per resident day)
Housekeeping and other staff (hours per resident day)
Resident Characteristics
.57
.81
.76
ADL index (range 1–9, with 9 the most dependent)
Mobility index (range 1–3, with 3 the most dependent)
Depression (% of residents)
Dementia (% of residents)
Behavioral symptoms (% of residents)
Urinary incontinence (% of residents)
Pressure sores (% of residents)
Facility Characteristics
5.78
1.42
.72
.41
18.60
42.39
26.31
50.35
6.54
13.54
18.04
18.00
16.86
5.14
Finally, we also conducted alternative logistic regressions
including (a) whether the facility was in the quartile with
the greatest number of deficiencies (i.e., more than eight de-
ficiencies were cited) and (b) whether any deficiencies were
evaluated by surveyors as indicating substandard care. The
Fewer than 60 beds (yes ϭ 1)
19.37
43.01
21.69
15.94
8.70
77.89
3.76
18.35
9.68
6
1
0–119 beds (yes ϭ 1)
20–160 beds (yes ϭ 1)
2
pseudo R statistic for binary response variables is the pro-
portion of variation in the dependent variable “explained”
by the independent variables in the model. The logistic re-
gressions were significant and informative but not as effec-
tive and useful in targeting facilities with outlying deficien-
cies as the OLS regressions.
More than 160 beds (yes ϭ 1)
Hospital based (yes ϭ 1; n ϭ 1,198)
SNF/NF (Category 2, 3; yes ϭ 1; n ϭ 10,726)
SNF (Category 4; yes ϭ 1; n ϭ 517)
NF (Category 10; yes ϭ 1; n ϭ 2,527)
%
%
%
Medicare
Medicaid
private pay
16.83
23.74
19.87
46.65
65.57
24.75
68.01
RESULTS
The means and standard deviations of all dependent and
independent variables are shown in Table 1. The mean num-
ber of total deficiencies per facility was 5.92 during the pe-
riod, the median was 4.0, and the standard deviation was
% for-profit
Region
Midwest (yes ϭ 1)
Northeast (yes ϭ 1)
South (yes ϭ 1)
West (yes ϭ 1)
34.42
16.86
32.86
15.86
6
.17. A total of 81,534 deficiencies were cited for all facili-
ties during the 1-year study period. Of these, 48,431 defi-
ciencies were in the quality of care category, 28,368 were in
the quality of life category, and 4,735 were in the other defi-
ciency category (data not shown).
Notes: Data are from the On-Line Survey Certification and Reporting System
for all nursing homes for July 1, 1995–June 30, 1996, Health Care Financing
Administration. n ϭ 13,770 nursing facilities. ADL ϭ activities of daily living;
LVN/LPN ϭ licensed vocational nurse/licensed practical nurse; NF ϭ nursing
facility; SNF ϭ skilled nursing facility.
Descriptive Statistics
Table 1 shows that the total staff hours per resident day
was 5.70 for all nursing facilities. RNs were 0.59 hr per res-
ident day, LVN/LPNs were 0.67 hr per resident day, and
nursing assistants were 2.14 hr per resident day. Total nurs-
ing staff was 3.4 hr, or 60% of total staff hours. Other care
staff hours per resident day were 1.26 hr. These included di-
etary personnel, activity staff, physical therapists, social
workers, occupational therapists, and others (22% of total
hours; see Table 1). Housekeeping and other staff were 13%
of total hours per resident day (0.76 hr). Administrative staff
was 5% of the total, or 0.28 hr per resident day.
6.5% had pressure sores. Most nursing homes (62%) had
fewer than 119 beds. Only 8.7% of facilities were hospital
based, and 3.76% were SNFs. Eighteen percent were certi-
fied as nursing facilities with Medicaid-only residents. The
majority of residents were on Medicaid (65.6%), with 9.7%
on Medicare and the remainder private pay. Of the facilities,
68% were for-profit.
Regression Models for Deficiencies
Table 2 shows the results of OLS regression models of
deficiencies. As expected, lower levels of RN staff were
significantly associated with total care deficiencies and
quality of care but not with quality of life deficiencies and
other deficiencies. LVN/LPN hours were not related to defi-
ciencies. Lower levels of nursing assistant staff were associ-
Table 1 shows the distribution of resident characteristics.
On a 9-point scale, the mean ADL index was 5.78. The
mean resident mobility was 1.42 on a 3-point scale. Half of
the residents had urinary incontinence, 42% had dementia,
2
6% had behavioral symptoms, 18.6% had depression, and

NURSING HOME STAFFING
S283
Table 2. Ordinary Least Squares Regression Model of Staffing, Resident Characteristics, Facility Characteristics, and States as Predictors
of Nursing Home Deficiencies (SE in Parentheses)
Characteristic and Variable
Constant
Total Care Deficiencies
Quality of Care Deficiencies
Quality of Life Deficiencies
Other Deficiencies
4.412**
2.626**
(.667)
1.473**
(.444)
.106
(.147)
(1.081)
Staff
Registered nurses
Ϫ.184*
Ϫ.128**
(.049)
.087
Ϫ.047
(.032)
.042
Ϫ.008
(.011)
.002
(.079)
.131
LVN/LPNs
(
.084)
(.052)
Ϫ.072*
(.028)
Ϫ.063
(.036)
.045
(.054)
Ϫ.071
(.038)
(.034)
Ϫ.052**
(.019)
Ϫ.048*
(.024)
Ϫ.028
(.036)
Ϫ.024
(.025)
(.011)
Ϫ.004
(.006)
.007
(.008)
Ϫ.025*
(.012)
Ϫ.012
(.008)
Nursing assistants
Other care staff
Ϫ.127**
.046)
Ϫ.104
.058)
Ϫ.008
.087)
Ϫ.106
(
(
Administrative staff
Housekeeping and other staff
(
(.061)
Resident Characteristics
ADL index
.009
.091
(.052)
.012
(.078)
Ϫ.894**
(.239)
Ϫ.701**
(.185)
.007
(.186)
.417*
(.208)
8.444**
(.643)
Ϫ.072*
(.035)
.021
(.052)
Ϫ1.022**
(.159)
Ϫ.516**
(.123)
.403**
(.124)
.442**
(.138)
4.107**
(.428)
Ϫ.011
(.011)
Ϫ.035*
(.017)
Ϫ.161**
(.052)
Ϫ.126**
.041
Ϫ.008
(.041)
.084
(.046)
1.078**
.141
(.084)
Mobility index
Depression
Ϫ.002
.127)
Ϫ2.075**
.387)
Ϫ1.342**
(
(
Dementia
(
(
(
.299)
.402
.302)
.942**
.337)
Behavioral symptoms
Urinary incontinence
Pressure sores
13.636**
.042
1
Facility Characteristics
Fewer than 60 beds
Ϫ1.528**
.166)
Ϫ.880**
Ϫ1.011**
(.103)
Ϫ.506**
(.079)
.362**
(.097)
.081
(.135)
Ϫ.152
(.210)
Ϫ.110
(.091)
1.153**
(.176)
Ϫ.502
(.301)
Ϫ.316**
(.074)
51.39**
44.951**
.187
Ϫ.493**
(.068)
Ϫ.349**
(.052)
.318**
(.065)
.051
(.090)
.212
(.139)
Ϫ.020
(.061)
1.403**
(.117)
Ϫ.249
(.200)
Ϫ.267**
(.049)
48.35**
46.380**
.192
Ϫ.024
(.023)
Ϫ.026
(.017)
.038
(
60–119 beds
(
(
(
(
.127)
.718**
.158)
.135
.219)
.042
More than 160
(.021)
.003
Hospital based
(.030)
Ϫ.018
(.046)
Ϫ.019
(.020)
.192**
(.039)
Ϫ.115
(.066)
Ϫ.047**
(.016)
17.46**
16.682**
.076
SNF (Category 4)
Nursing Facility (Category 10)
.340)
Ϫ.148
.147)
2.747**
.286)
Ϫ.863
.488)
Ϫ.629**
.120)
(
%
%
Medicaid
Medicare
(
(
Nonprofit and government
(
State (df ϭ 50) F value
56.88**
52.062**
.211
F
Adjusted R2
Notes: Comparisons: Beds 120–159; nonhospital; Category 2, 3; private pay residents; for-profit; and Wyoming. n ϭ 13,770. ADL ϭ activity of daily living; LVN/
LPN ϭ licensed vocational nurse/licensed practical nurse; SNF ϭ skilled nursing facility.
*p Ͻ .05; **p Ͻ .01.
ated with more total care deficiencies, quality of care defi-
ciencies, and quality of life deficiencies but not with other
deficiencies. Lower levels of other care staff were associ-
ated with increased quality of life deficiencies but not with
total deficiencies or quality of care deficiencies. As ex-
pected, lower levels of administrative staff were associated
with higher deficiencies in the “other” category that in-
cluded administrative deficiencies. Contrary to the hypothe-
ses, lower levels of housekeeping and other staff had no re-
lationship to deficiencies.
Also contrary to our hypotheses, the average ADL depen-
dence index was negatively associated with quality of life

S284
HARRINGTON ET AL.
deficiencies, but there was no relationship with quality of
care deficiencies. Facilities with proportionately fewer resi-
dents with mobility limitations had higher rates of other de-
ficiencies. Facilities with higher percentages of residents
with depression and dementia had significantly fewer defi-
ciencies of all types. Facilities with a higher percentage of
residents with behavioral symptoms had more quality of life
deficiencies. Facilities with more incontinent residents had
more total deficiencies and more quality of care and quality
of life deficiencies. The percentage of residents with pres-
sure sores was positively associated with total care deficien-
cies and deficiencies in all deficiency categories. Because
some of the resident characteristics in the model could also
be considered outcomes or related to deficiencies, we ran
the regression models excluding the percentages of resi-
dents with depression, dementia, behavioral symptoms, in-
continence, and pressure sores. The regression models were
stable with and without these variables, although including
these characteristics added somewhat to the model.
Smaller facilities (16–60 beds and 60–119 beds) were
less likely to have quality of care and quality of life defi-
ciencies and had fewer total deficiencies than larger facili-
ties. Neither hospital affiliation nor certification status was
associated with deficiencies. Facilities with greater percent-
ages of Medicaid residents were more likely to have defi-
ciencies of all types. Nonprofit and government facilities
had fewer deficiencies of all types than for-profit facilities.
States were also significant predictors of deficiencies (see
Table 2 for summary F value). The regression model was
also tested with the four census regions as substitutes for the
state dummy variables (data not shown). The midwest,
south, and west each had higher numbers of all types of de-
ficiencies than did the northeast comparison group. The
highest coefficients were found in the western region.
Table 3 shows the alternative regression models for defi-
ciencies. A logistic regression shows the findings for facili-
ties that had any deficiency; these findings were similar to
those in Table 2 except that LVN/LPN hours were signifi-
cant and nurse assistant hours were not. When the regres-
sion for the log of deficiencies (excluding those facilities
with no deficiencies) shown in Table 3 was compared with
the OLS regression for the total deficiencies (Table 2), the
findings were similar. With respect to staffing levels, RN
and nurse assistant hours were negative predictors of defi-
ciencies. Overall, the alternative models yielded the same
general conclusions and showed the robustness of the OLS
model.
staff hours were associated with more quality of care defi-
ciencies. Fewer nursing assistant hours, as expected, had a
consistent, significant negative relationship with total, qual-
ity of care, and quality of life deficiencies. Fewer other care
staff hours were associated with more quality of life defi-
ciencies. On the other hand, staffing hours and resident
characteristics did not explain much of the variation in defi-
ciencies when case mix and facility characteristics and other
factors were controlled for in the model. Because we were
able to explain only a small portion of the variance, much
work remains to explore these important relationships with
deficiencies.
The relationship between resident characteristics and de-
ficiencies was complex. Contrary to our hypotheses, lower
scores on ADL impairment were also related to more qual-
ity of life deficiencies, perhaps because surveyors consid-
ered that more independent residents need more activities
and stimulation than more impaired residents. Surprisingly,
depression and dementia were also associated with reduc-
tions in the number of deficiencies. This finding raises ques-
tions about whether quality problems associated with the
psychosocial characteristics are difficult for surveyors to
identify; or perhaps surveyors are less concerned about
quality of care and life for cognitively impaired residents,
and this is reflected in fewer deficiencies.
As expected, pressure sores were related to increased
numbers of deficiencies of all types because they are visible
and quantifiable measures of poor quality. As expected, uri-
nary incontinence was also associated with more deficien-
cies. Urinary incontinence is a treatable problem in nursing
homes and has also been the focus of the survey process,
which may explain the association with deficiencies. On the
other hand, surveyors may tend to cite facilities when they
observe problems that may be more of an indicator of case
mix than of quality problems. In other words, it may be dif-
ficult for surveyors to determine when poor outcomes are
related to the disease process or to poor and inadequate care.
As expected, higher percentages of behavioral problems,
urinary incontinence, and pressure sores were positively as-
sociated with quality of life deficiencies. These easily ob-
servable problems may have a negative impact on individ-
ual residents and could also have a negative impact on the
overall environment of the facility. Future studies should
examine these relationships in greater detail.
Consistent with our hypothesis, larger facilities were as-
sociated with more deficiencies (when staffing and resi-
dents were controlled), perhaps because larger facilities are
less able to monitor and ensure both quality of care and
quality of life requirements for residents. Alternatively, sur-
veyors’ judgments may not be very sensitive to facility size.
As expected, facilities with more Medicaid residents were
associated with more deficiencies, even when we controlled
for staffing and resident characteristics, suggesting poorer
care in such facilities. This finding is consistent with previ-
ous studies by Nyman (1988) that suggest poorer quality of
care occurs in facilities with higher proportions of Medicaid
residents, especially in areas where there is excess demand
for services. The findings may be related to the reduced
need to compete on the basis of quality or to lower Medi-
caid reimbursement rates compared with Medicare and pri-
Staffing hours alone predicted less than 1% of the total
variance in deficiencies (data not shown). Staffing and resi-
dent characteristics were significant predictors of total defi-
ciencies but only explained 3% of the variance when facility
characteristics were not included (data not shown). When
facility characteristics and region were added to the model
as shown in Table 2, the proportion of variance explained
increased to 21.1%.
DISCUSSION
Consistent with the hypotheses and previous studies (see
Institute of Medicine, Committee on the Adequacy of Nurse
Staffing in Hospitals and Nursing Homes, 1996), fewer RN

NURSING HOME STAFFING
S285
Table 3. Comparison of Alternative Regression Models Predicting Deficiencies
Log of Deficiencies Without
Any Substandard
Has Any Deficiencies
Zero Deficiencies
More Than Eight Deficiencies
Care Deficiency
Independent Variable
(Yes ϭ 1; n ϭ 13,770)
(n ϭ 11,708)
(Yes ϭ 1; n ϭ 13,770)
(Yes ϭ 1; n ϭ 13,770)
Staff
Nursing staff
Registered nurses
LVN/LPNs
.877**
1.156*
.953
0.954
1.026
.950
Ϫ.028*
.005
Ϫ.017*
Ϫ.018
Ϫ.004
Ϫ.019*
.983
1.011
.965
.974
.991
.686**
1.02
0.924
0.986
0.944
0.911
Nursing assistants
Other care staff
Administrative staff
Housekeeping and other staff
Resident Characteristics
ADL index
.916*
1.122*
1.117
.004
.003
.964
1.018
.864*
0.875
Mobility index
Depression
Dementia
Behavioral symptoms
Urinary incontinence
Pressure sores
.380**
.851
1.244
.703
12.439**
Ϫ.271**
Ϫ.187**
.028
.106*
2.091**
.375**
.593**
1.104
1.311
221.457**
0.654
0.914
1.631*
2.723**
191.586**
Facility Characteristics
Fewer than 60 beds
.775**
.842*
1.222*
.971
Ϫ.207**
Ϫ.128**
.070**
.031
.568**
.711**
1.239**
1.101
.592*
.780**
1.282*
0.906
60–119 beds
More than 160 beds
Hospital based
SNF (Category 4)
Nursing facility (Category 10)
.944
.981
3.250**
1.107
.787**
38.14**
Ϫ.048
Ϫ.030
.367**
Ϫ0.102
Ϫ.074**
45.59**
1.21
.622*
.903
2.778**
.607
.813**
35.75**
0.461
1.024
2.974**
2.852
0.932
%
%
Medicaid
Medicare
Nonprofit and government
State (df ϭ 50) F value
Constant
6.82**
Adjusted/pseudo R2
.102
.181
.152
0.035
Notes: Regression coefficients are shown for the ordinary least squares (OLS) model of total deficiencies without the zero deficiencies. All other numbers are odds
2
2
resulting from logistic regressions. The adjusted R is given for the OLS model. A pseudo R is shown for the logistic regression models. ADL ϭ activity of daily liv-
ing; LVN/LPN ϭ licensed vocational nurse/licensed practical nurse; SNF ϭ skilled nursing facility.
*p Ͻ .05; **p Ͻ .01.
vate pay rates. Confirming our hypothesis and the findings
from other studies, nonprofit and government facilities were
associated with fewer deficiencies (Aaronson et al., 1994;
Cohen & Spector, 1996). These facilities may be providing
higher quality of care than the for-profits because they may
reinvest their net revenues into the facilities. These facilities
may have more staff expertise and lower turnover rates and
other characteristics that were unobserved variables. On the
other hand, surveyors may be more sympathetic to nonprofit
and government nursing homes than the for-profit facilities.
State variations in deficiencies confirm that HCFA
should pay greater attention to achieving uniformity in the
state survey procedures. Lower levels of deficiencies in the
northeastern region compared with other regions may indi-
cate that facilities in the northeast have higher quality of
care or that the region has weaker survey procedures. Re-
gional differences could also be a proxy for other market
factors not included in the model. It would be troubling if
the differences were related to different survey procedures
and/or less enforcement, because the federal program
strives to ensure standard procedures and enforcement
across regions. The U.S. Department of Health and Human
Services, Health Care Financing Administration (1998a,
plementation of its survey process. HCFA should use find-
ings from this study to examine the differences in deficien-
cies across states and regions and to eliminate variation in
the survey and enforcement process.
There are important issues to be addressed concerning the
quality of OSCAR data. The staffing data reported in OS-
CAR are collected for only a 2-week period at the time of
the annual survey and generally are not audited by survey-
ors. Some facilities may increase their staff during the pe-
riod around the survey. Thus, the reports may overstate the
actual staffing in facilities. Unfortunately, no other data
source on staffing across states is available. We have com-
pared the average staffing hours reported on OSCAR for all
facilities in California for the 1996–97 period with the an-
nual staffing data submitted to California on the facility cost
reports for the same period. This comparison did show that
the annual nursing staffing was on average only 0.1 h lower
than the OSCAR data, suggesting that the data are similar.
Another limitation is that the available staffing data do not
capture differences in education levels, capability, motiva-
tion, and experience of staff. It may be that these unmea-
sured factors—such as educational level, experience, num-
ber of nurses in advanced practice, turnover rates, and other
information—are more important than the hours and spe-
1
998b) has admitted it has problems with the effective im-

S286
HARRINGTON ET AL.
cific types of staff used. Staffing reports made on a quar-
terly basis would be more informative than those made for a
care may be occurring so that surveyors can examine these
residents during the survey process. Surveyors can also use
the QIs to identify specific facilities with higher or lower
proportions of residents with potential quality problems,
and this technology may improve the accuracy of identify-
ing problems and issuing deficiencies. Researchers should
continue to explore the relationships among staffing, resi-
dent characteristics, facility characteristics, and quality. As
databases are improved, perhaps the relationships may be-
come more clearly understood.
2
-week period. Thus, if staffing has a limited effect, it may
reflect some limitations in the data. Given the importance of
the issues surrounding staffing and its relationship to qual-
ity, HCFA should undertake a more detailed, rigorous study
of the appropriateness of current staffing data and the accu-
racy and representativeness of OSCAR staffing data.
Another issue is whether the deficiencies are an accurate
reflection of quality in the facilities. The identification of
quality problems in nursing homes is a difficult task for sur-
veyors. As discussed previously, there are good reasons to
believe that the deficiencies reported do identify problems
within the facilities. However, there may be additional
problems that are not reflected in any deficiencies. The U.S.
General Accounting Office (1998b) had independent sur-
veys conducted by a reputable research team. The research
team conducted surveys concurrently with California state
surveyors and found additional deficiencies that were not
identified by the California state survey agency, but they did
not identify false positive deficiencies. Thus, there may be false
negatives in surveyors’ identification of deficiencies (see also
U.S. General Accounting Office, 1999). There may, of course,
also be false positives with deficiencies that are undetected.
Deficiencies for quality violations are relatively infre-
quent, as we have shown. Moreover, the average number of
deficiencies per facility has been declining steadily from 8.8
deficiencies in l991 to 4.9 in l997 (Harrington & Carrillo,
Acknowl edgment s
This research was funded by the Health Care Financing Administration
and the Agency for Health Care Policy and Research. The article reflects
the opinions of the authors and not those of the funding agencies.
Address correspondence to Dr. Charlene Harrington, Professor, Department
of Social and Behavioral Sciences, University of California, 3333 California
Street, Suite 455, San Francisco, CA 94118. E-mail: chas@itsa.ucsf.edu
Refer ences
Aaronson, W. E., Zinn, J. S., & Rosko, M. D. (1994). Do for-profit and not-for-
profit nursing homes behave differently? The Gerontologist, 34, 775–786.
Aaronson, W. E., Zinn, J. S., & Rosko, M. D. (1995). Subacute care, Medi-
care benefits, and nursing home behavior. Medical Care Research and
Review, 52, 364–388.
Arling, G., Nordquist, R. H., & Capitman, J. A. (1987). Nursing home cost
and ownership type: Evidence of interaction effects. Health Services
Research, 22, 255–269.
Bliesmer, M. M., Kane, R. L., & Shannon, I. (1998). The relationships be-
tween nursing staffing levels and nursing home outcomes. Journal of
Aging and Health, 10, 351–371.
Buchanan, R. J., Madel, R. P., & Persons, D. (1991). Medicaid payment
policies for nursing home care: A national survey. Health Care Financ-
ing Review, 13, 55–72.
Cherry, R. L. (1991). Agents of nursing home quality of care: Ombudsmen
and staff ratios revisited. The Gerontologist, 31, 302–308.
Chiodo, L. K., Gerety, M. D., Mulrow, C. D., Rhodes, M. C., & Tuley,
M. R. (1992). The impact of physical therapy on nursing home patient
outcomes. Physical Therapy, 72, 168–173.
Cohen, J. W., & Dubay, L. C. (1990). The effects of Medicaid reimburse-
ment method and ownership on nursing home costs, case mix and
staffing. Inquiry, 27, 183–200.
Cohen, J. W., & Spector, W. D. (1996). The effect of Medicaid reimburse-
ment on quality of care in nursing homes. Journal of Health
Economics, 15, 23–48.
Davis, M. A. (1993). Nursing home ownership revisited: Market, cost, and
quality relationships. Medical Care, 31, 1062–1068.
Dor, A. (1989). The costs of Medicare patients in nursing homes in the
United States. Journal of Health Economics, 8, 253–270.
Fottler, M. D., Smith, H. L, & James, W. L. (1981). Profits and patient care
quality in nursing homes: Are they compatible? The Gerontologist, 21,
532–538.
Fries, B. E., & Cooney, L. (1985). Resources utilization groups: A patient
classification system for long-term care. Medical Care, 23, 110–122.
Fries, B. E., Schneider, D., Foley, W., Gavazzi, M., Burke, R., & Corne-
lius, E. (1994). Refining a case-mix measure for nursing homes:
Resource utilization groups (RUGS-III). Medical Care, 32, 668–685.
Harrington, C., & Carrillo, H. (1999). The regulation and enforcement of
federal nursing home standards. Medical Care Research and Review,
1
999; U.S. Office of the Inspector General, Department of
Health and Human Services, 1999). Thus, the accuracy and
measurement issues of deficiencies are even more impor-
tant. If HCFA is able to improve the reliability and consis-
tency of the survey process and the citing of deficiencies
within and across states, then the relationships between in-
dependent and dependent variables may be improved.
Finally, the technology of appropriately identifying qual-
ity problems in nursing facilities can be improved. It is
hoped that the use of new quality indicators (QIs) will im-
prove the identification of problems (Zimmerman, et al.,
1
995). These indicators, developed by Zimmerman and col-
leagues (1995), identify 30 resident problems with data
from the Minimum Data Set (MDS) assessment forms that
must be completed for each resident upon admission and
annually, with changes in conditions reported to HCFA on a
quarterly basis. Such MDS data on individual residents, al-
though it is also not regularly audited by state surveyors,
should be more accurate than the resident data reported on
the OSCAR report. Once the MDS data become available
for researchers, the data should be a valuable tool in study-
ing the relationships between staffing, resident characteris-
tics, and facility characteristics.
The reporting of QIs will allow surveyors and researchers
to identify the condition of residents upon admission and
then over time. Future studies should examine changes in
resident conditions over time, controlling for the initial con-
dition of residents. This may improve the accuracy of de-
tecting true quality problems that are the result of poor care,
taking into account the resident’s initial condition and/or
resident co-morbidities. Surveyors can use the QIs to target
individual residents in facilities where potential quality of
5
6, 471–494.
Harrington, C., Carrillo, H., Mullan, J., & Swan, J. H. (1998). Nursing fa-
cility staffing in the states: The 1991–95 period. Medical Care
Research and Review, 55, 334–363.
Harrington, C., Carrillo, H., Thollaug, S., & Summers, P. (2000). Nursing
facilities, staffing, residents, and facility deficiencies, 1992–98. Report
Prepared for the Health Care Financing Administration. San Fran-
cisco, CA: University of California. Available Worldwide Web:
www.hcfa.gov/medicaid/ltchomep/htm.

NURSING HOME STAFFING
S287
Holahan, J., & Cohen, J. (1987). Nursing home reimbursement: Implica-
bursement for nursing homes, 1978–88. Health Care Financing
Review, 14, 111–131.
tions for cost containment, access, and quality. The Milbank Quarterly,
6
5(1), 112–147.
Ullmann, S. G. (1987). Ownership, regulation, quality assessment, and per-
formance in the long-term health care industry. The Gerontologist, 27,
233–239.
U.S. Department of Health and Human Services. (1998a, July 21). HHS
fact sheet: Assuring the quality of nursing home care. Washington, DC:
Author.
U.S. Department of Health and Human Services (1998b, July). Report to
Congress: Study of private accreditation. deeming of nursing homes,
regulatory incentives and non-regulatory initiatives, and effectivess of
the survey and certification system (Vol. 1–4). Washington, DC: Author.
U.S. Department of Health and Human Services, Health Care Financing
Administration. (1995a, April). State operations manual (Provider Cer-
tification. Transmittal No. 272) Washington, DC: Author.
U.S. Department of Health and Human Services, Health Care Financing
Administration. (1995b, June). State operations manual (Provider Cer-
tification. Transmittal No. 273) Washington, DC: Author.
U.S. Department of Health and Human Services, Health Care Financing
Administration. (1995c, June). State operations manual (Provider Cer-
tification. Transmittal No. 274) Washington, DC: Author.
U.S. General Accounting Office. (1998a, July). California nursing homes:
Care problems persist despite federal and state oversight. report to the
special committee on aging, U.S. Senate (GAO/HEHS-98-202). Wash-
ington, DC: U.S. Government Printing Office.
Holahan, J., Rowland, D., Feder, J., & Heslam, D. (1993, Fall). Explaining
the recent growth in Medicaid spending. Health Affairs, 12(3), 177–193.
Institute of Medicine, Committee on the Adequacy of Nurse Staffing in Hos-
pitals and Nursing Homes (1996). Nursing staff in hospitals and nursing
homes: Is it adequate? Washington, DC: National Academy Press.
Kanda, K., & Mezey, M. (1991). Registered nurse staffing in Pennsylvania
nursing homes: Comparison before and after implementation of Medi-
care’s prospective payment system. The Gerontologist, 31, 318–324.
Karuza, J., & Katz, P. R. (1994). Physician staffing patterns correlates of
nursing home care: An initial inquiry and consideration of policy impli-
cations. Journal of the American Geriatrics Society, 42, 787–793.
Kochersberger, G., Hielema, F., & Westlund, R. (1994). Rehabilitation in
the nursing home: How much, why, and with what results. Public
Health Reports, 109, 372–376.
Linn, M., Gurel, L., & Linn, B. A. (1977). Patient outcome as a measure of
quality of nursing home care. American Journal of Public Health, 67,
3
37–344.
Mulrow, C. D., Gerety, M. B., Kanten, D., Cornell, J. E., DeNino, L. A.,
Chiodo, L., Aguilar, C., O’Neil, M. B., Rosenberg, J., & Solis, R. M.
(
1994). A randomized trial of physical rehabilitation for very frail nurs-
ing home residents. Journal of the American Medical Association, 271,
19–524.
5
Munroe, D. J. (1990). The influence of registered nursing staffing on the
U.S. General Accounting Office. (1998b). California nursing homes: Fed-
eral and state oversight inadequate to protect residents in homes with
serious violations. W. J. Scanlon, Testimony to the Special Committee
on Aging, U.S. Senate (GAO/HEHS-98-219). Washington, DC: Author.
U.S. General Accounting Office. (1999). Nursing homes: Additional steps
needed to strengthen enforcement of federal quality standards. Report
to the Special Committee on Aging, U.S. Senate (GAO/HEHS-99-46).
Washington, DC: Author.
quality of nursing home care. Research in Nursing and Health, 13,
2
63–270.
Nyman, J. A. (1988). Improving the quality of nursing home outcomes:
Are adequacy- or incentive-oriented policies more effective? Medical
Care, 26, 1158–1171.
Omnibus Budget Reconciliation Act of 1987, Pub. L. No. 100-203, Subtitle
C: Nursing Home Reform (1987).
Przybylski, B. R., Dumont, E. D., Watkins, M. E., Warren, S. A., Beaulne,
A. P., & Lier, D. A. (1996). Outcomes of enhanced physical and occu-
pational therapy service in a nursing home setting. Archives of Physical
Medicine and Rehabilitation, 77, 554–561.
Schneider, D. P., Fries, B. D., Foley, W. J, Desmond, M., & Gormley, W.
J. (1988). Case mix for nursing home payment: Resource utilization
groups, version II. Health Care Financing Review (Suppl.), 39–52.
Spector, W. D., Selden, T. M., & Cohen, J. W. (1998). The impact of own-
ership type on nursing home outcomes. Health Economics, 7, 639–653.
Spector, W. D. & Takada, H. A. (1991). Characteristics of nursing homes
that affect resident outcomes. Journal of Aging and Health, 3, 427–454.
SPSS. (1998). SPSS Base 8.0. Chicago, IL: SPSS, Inc.
U.S. Office of the Inspector General, Department of Health and Human
Services. (1999). Nursing home survey and certification: Deficiency
trends (OEI-02-98-00331). Washington, DC: Author.
Zimmerman, D. R., Karon, S. L., Arling, G., Clark, B. R., Collins, T.,
Ross, R., & Sainfort, F. (l995). Development and testing of nursing
home quality indicators. Health Care Financing Review, 16, 107–127.
Zinn, J. S. (1993). The influence of nurse wage differentials on nursing home
staffing and resident care decisions. The Gerontologist, 33, 721–729.
Received April 9, 1999
Accepted April 12, 2000
Swan, J. H., Harrington, C., & Grant, L. A. (1993). State Medicaid reim-