David Zimmerman

Using indicators to structure quality improvement initiatives in long-term care.

Quality indicators and a quality-monitoring system developed for use in the regulatory survey process offer a quality improvement tool for nursing home staff. The systematic use of resident assessment data can aid in the identification of quality of care problems and the determination of the nature of those problems. This approach is particularly useful for continuous quality improvement efforts.

An Empirical Test of the Structure, Process, and Outcome Quality Paradigm Using Resident-Based, Nursing Facility Assessment Data

This study distinguishes between organizational char acteristics, regarded as exogenous structural indicators of quality, and those identified as endogenous indica tors of structural care (SC), and investigates the degree to which measures of SC vary by ownership mode (defined by four combinations of chain affiliation and profit status) for 142 certified and licensed nursing facilities (NFs) in a southern state. Structural care mea sures include: licensed and unlicensed staffing, licensed therapists, and case mix-adjusted direct care expendi tures. In addition, seven (four process and three out come) facility-level, risk-adjusted process, and outcome quality scales are developed from 39 resident-level qual ity indicators. A causal model of NF quality arranged according to the structure, process and outcome para digm is specified and estimated using path analysis. Or ganizational data derive from the 1991 Medicaid Cost Report; process and outcome quality measures were de veloped from the Minimum Data Set Plus Resident As sessment Instrument. Using the percentage of Medicaid and private pay residents as covariates, there was a significant overall multivariate effect due to ownership mode on the SC measures. Although there were several significant direct effects, the overall path model was disconfirmed. The multivariate results suggest that some organizational characteristics of structure quality may be more appropriately considered exogenous to causal quality models and therefore have indirect (ver sus direct) effects on process or outcome quality indica tors. The path analysis implies that the structure-pro cess-outcome paradigm may not accurately capture the way NF health care is delivered. Research which consid ers alternate NF quality paradigms needs to be done with samples that are more representative of national proportions of each ownership mode.

Longitudinal trends in antibiotic resistance in US nursing homes, 2000-2004.

We evaluated antibiotic resistance trends in US nursing homes using the Minimum Data Set. Significant increases in the number and proportion of infections caused by antibiotic-resistant bacteria were documented over the 5-year study. Further research on antibiotic resistance in nursing homes is urgently needed.

Residential care supply and cognitive and physical problem case mix in nursing homes.

Objectives: Arapid evolution has occurred in state policy and industry practices relative to assisted living and expanded use of residential care facilities for people with physical and cognitive frailty, yet relatively little is known about the interrelationship between this housing supply and nursing-home case mix. Methods:The association between residential care supply and the proportion of cognitively and physically impaired nursing facility residents was examined in more than 1,500 facilities in five states. Results:The proportion of nursing-home cases with only physical and cognitive impairment likely to be affected by emerging long-term care policy appears to be well under 10%. This effect is more persistent among admissions than continuing cases. Discussion:The findings raise caution about the optimistic assumptions of the interplay between residential care/assisted living policy and nursing-home use.

Antibiotic resistance in non-major metropolitan skilled nursing facilities: prevalence and interfacility variation.

Skilled nursing facilities (SNFs) represent ideal environments for the emergence and spread of antibiotic resistance.1 Studies have found that residents in Veterans Administration (VA) SNFs2,3 and non-VA SNFs in major metropolitan areas4,5 are frequently colonized with antibiotic-resistant bacteria (ARB). The extent to which residents of non-urban SNFs are colonized with ARB remains poorly understood. Intrinsic differences in patient populations, referral patterns as well as other contextual factors may fuel very different patterns of antibiotic resistance in non-urban SNFs. Our group recently completed a longitudinal study to document patterns of antibiotic resistance in several SNFs located in non-urban counties of South Central Wisconsin. Herein, we present the colonization results of surveillance cultures performed at the inception of the study cohort in 2008–2009. The University of Wisconsin’s Institutional Review Board approved this study. A potential pool of 39 SNFs (size, ≥ 60 beds) located in 9 South Central Wisconsin counties was constituted from a directory of licensed facilities maintained by the State of Wisconsin. A randomly assigned number was used to determine the order in which facilities were approached by the research team. Six of the first 10 facilities approached agreed to participate. Variables describing characteristics of the facility and resident population were constructed from annual data collected during the state survey process as well as data collected from medical records of subjects at study entry. Residents of participating SNFs over the age of 18, including those with cognitive impairment, were eligible to participate. After obtaining written informed consent, multi-anatomical screening for colonization with methicillin-resistant Staphylococcus aureus (MRSA) and fluoroquinolone-resistant gram-negative bacteria (FQRGNB) was performed. Cultures of nares, skin of the axilla and groin, and perianal skin (or stool) were obtained from all subjects to detect MRSA colonization. Additional cultures of wounds, the insertion site of non-urinary indwelling medical devices, and urine collected from indwelling urinary device were obtained, when applicable. The same body sites, exclusive of nares and axilla/groin, were sampled to detect FQRGNB colonization. MRSA specimens were enriched in trypticase soy broth supplemented with 6.5% NaCl for 24 hours before plating onto selective media -- Mannitol Salt agar (Remel, Lenexa, KS) containing 4 µg/mL of cefoxitin. FQRGNB specimens were plated directly onto MacConkey agar (Remel, Lenexa, KS) containing 4 µg/mL of ciprofloxacin. All plates were incubated aerobically for 48 hours at 37°C and were identified to the species level using standard techniques. Cefoxitin and ciprofloxacin resistance was confirmed using the Kirby Bauer disk diffusion method. Point estimates and 95% confidence intervals of the proportion of residents colonized with MRSA and FQRGNB were calculated. Pearson chi-square tests were performed to identify if a significant difference in the proportion of subjects colonized with MRSA and FQRGNB across study locations was present. When applicable, visual inspection of confidence limits was performed to identify facility pairs accounting for those differences. The characteristics of the participating facilities, including characteristics of participating subjects in aggregate, are presented in Table 1. 449 of the 851 (53%) residents in the 6 participating SNFs were screened at baseline. An equal proportion of subjects were colonized with MRSA (22.3%; 95% CI 13.7 – 30.9%) and FQRGNB (21.3%; 95% CI 13.3 – 29.3%). Approximately 5% of participating subjects were co-colonized with MRSA and FQRGNB (95% CI 2.8 – 7.1%). Overall, 38.7% (95% CI 32.9 – 44.5%) of subjects screened were colonized with either MRSA and/or FQRGNB. Table 1 Facility Characteristics and Prevalence of Antibiotic-Resistant Bacteria for 6 Skilled Nursing Facilities in South Central Wisconsin. Significant variation in the proportion of subjects colonized with MRSA (Pearson chi-square = 14.6, P = 0.012) and FQRGNB (Pearson chi-square = 13.2, P = 0.022) was identified across the 6 facilities. Significant differences in the prevalence of MRSA were identified between Facility #3 (13.0%) and Facility #4 (33.7%). Significant differences in the prevalence of FQRGNB were identified between Facility #2 (29.1%) and Facility #6 (11.3%). The characteristics of facilities with the highest prevalence of MRSA or FQRGNB were not qualitatively different from facilities with a lower prevalence of MRSA or FQRGNB (Table 1). The generalizeability of our findings may be limited by the method in which study facilities were selected. Our study facilities, while representative of non-urban SNFs that cater to long-term stay residents requiring nursing services of low complexity, may not be representative of urban SNFs that provide a more complex level of nursing care.6 Nevertheless, the prevalence of MRSA in facilities in our study are not substantively different from those recently described for SNFs in a highly urbanized county in California.7 Comparable data on the prevalence of FQRGNB in other SNFs are not available. However, recently published studies describing sharp increases in the proportion of clinical isolates obtained from residents of Northeastern SNFs that were resistant to fluoroquinolone antibiotics8 as well as a high prevalence of FQRGNB colonization among SNF residents with an indwelling medical device9 support the generalizeability of our findings. In combination, these data suggest that a post-fluoroquinolone era has begun to emerge in U.S. SNFs. Few studies have attempted to measure the variation in antibiotic resistance across SNFs within the same geographic region.7,10 The twofold variation in FQRGNB prevalence and threefold variation in MRSA prevalence seen among SNFs in our study raise questions that require further study. Specifically, is variation being driven by differences in referral patterns, intra-facility antibiotic prescribing, intra-facility adherence to transmission-based precautions or some combination thereof? Pursuing the answers to these questions will be important for developing and implementing interventions to reduce the regional spread of antibiotic resistance. In summary, our study affirms the notion that residents of SNFs are commonly colonized with MRSA and FQRGNB, even in non-urban facilities that provide relatively low complexity of nursing care. Considerable variation in the prevalence of MRSA and FQRGNB in SNFs in the same geographic region exists. The explanations for this degree of inter-facility variation remain poorly understood and deserve further study.

Scope and severity index: a metric for quantifying nursing home survey deficiency number, scope, and severity adjusted for the state-related measurement bias.

OBJECTIVES To develop a metric (scope and severity index [SSI]) to measure nursing home deficiency number, scope, and severity adjusted for the state-related bias and to test its convergent and predictive validity. DESIGN We assigned scope and severity weights to each level of scope and severity (A-L). SSI was calculated as a sum of all weights per survey which was further corrected for the state-level bias by dividing by the state average number of health deficiencies and multiplying by the national average number of health deficiencies. Data source - National Online Survey, Certification, and Reporting system. SETTING All Medicare/Medicaid-certified skilled nursing facilities. MEASUREMENTS We correlated SSI with nursing home staffing levels (convergent validity) and denial of payment for new admissions (predictive validity). RESULTS The expert panel reached agreement on the scope and severity weights: Level A = 5, B = 10, C = 15, D = 20, E = 30, F = 40, G = 35, H = 50, I = 65, J = 55, K = 75, and L = 100 points. Scope and severity per deficiency was positively correlated with the number of deficiencies in that survey. SSI contained almost no state-related bias, but yet related state-level variability. It demonstrated strong face, convergent, and predictive validity. CONCLUSIONS SSI rendered a valuable metric to conduct quantitative analyses of nursing home deficiency number, scope, and severity across states. Future research should investigate the positive relationship between scope and severity of deficiencies and their number. Better understanding and correction of other factors introducing systematic bias to the survey results (e.g. regional impact) can further improve the accuracy of survey result evaluation.

Nursing home case mix in Wisconsin. Findings and policy implications.

Along with many other states, Wisconsin is considering a case mix approach to Medicaid nursing home reimbursement. To support this effort, a nursing home case mix model was developed from a representative sample of 410 Medicaid nursing home residents from 56 facilities in Wisconsin. The model classified residents into mutually exclusive groups that were homogeneous in their use of direct care resources, i.e., minutes of direct care time (weighted for nurse skill level) over a 7-day period. Groups were defined initially by intense, Special, or Routine nursing requirements. Within these nursing requirement categories, subgroups were formed by the presence/absence of behavioral problems and dependency in activities of daily living (ADL). Wisconsins current Skilled/Intermediate Care (SNF/ICF) classification system was analyzed in light of the case mix model and found to be less effective in distinguishing residents by resource use. The case mix model accounted for 48% of the variance in resource use, whereas the SNF/ICF classification system explained 22%. Comparisons were drawn with nursing home case mix models in New York State (RUG-II) and Minnesota. Despite progress in the study of nursing home case mix and its application to reimbursement reform, methodologic and policy issues remain. These include the differing operational definitions for nursing requirements and ADL dependency, the inconsistency in findings concerning psychobehavioral problems, and the problem of promoting positive health and functional outcomes based on models that may be insensitive to change in resident conditions over time.

Applying Performance Measures to Long Term Care

BACKGROUND Nine nursing facilities in Mississippi participated in the American Health Care Associations Quality Indicator Index and Education (QUIIX-Ed) project to apply quantitative performance measurements to continuous quality improvement in long-term care. Begun in May 1994, with the first collection of performance measurement data in March 1995, the project used 15 measures of clinical care based on standardized assessments of resident conditions. METHODOLOGY QUIIX-Ed assessed the impact of quality indicator information in nursing facilities based on Minimum Data Set resident assessments. Comparison reports allow each facility to profile its performance against that of other providers in the project. The project does not standardize the quality improvement process: the focus is on the incorporation of information into existing nursing care and quality improvement practices. OPERATIONAL ISSUES Each month the facilities updated their quality indicator scores and exported those scores to a central database that aggregated scores to generate benchmarking statistics for each facility, including project medians, percentiles, and facility performance rankings. EXAMPLES Two case studies on bedfast residents and restraint use illustrate the process of examining care issues identified by quality indicators, developing appropriate interventions, and evaluating the efficacy of the intervention. DISCUSSION The limited intervention of introducing quantitative care measurements had a significant impact on resident outcomes and the nursing process. These measurements are readily incorporated into existing nursing practices and facility improvement efforts, regardless of the extent to which the quality improvement process has been developed.

Mental Illness and Psychotropic Medication Use in the Nursing Home

The authors examined mental illness and psychotropic medications use among nursing home residents. Data were drawn from the Texas Long-Term Care Reimbursement Project, a 1986 study of nearly 2,000 residents in 49 nursing homes. The study measured the use of antipsychotics and other psychotropic medications, physical health conditions, mental illness diagnoses, behavior, and nursing and other direct-care time for sampled residents. The findings indicated that 45% of the sample was receiving an antipsychotic or other psychotropic medication. Although psychotropics were prescribed more extensively for those with a psychiatric diagnosis, nearly one half of persons without a psychiatric diagnosis were receiving psychotropic medications at the time of the survey. Moreover, psychotropics were quite prevalent among those with unstable medical conditions and/or severe activities of daily living impairment. Neither a mental illness diagnosis, evidence of a behavioral problem, nor use of psychotropics was significantly correlated with the amount of nursing or other direct-care time received by residents. The findings raise concerns about the wide-spread prescribing of these medications, especially among residents who have no supporting psychiatric diagnosis and/or who have physical health conditions making them vulnerable to adverse drug effects.

A nurse-staffing taxonomy for decision making in long-term care nursing facilities.

A nurse-staffing taxonomy is proposed to facilitate informed staffing decisions in long-term care nursing facilities and to set forth construct components for empirically related research. Recommendations from an expert workgroup were synthesized with current staffing research to define a staffing taxonomy. Refinements were made, incorporating on-site nursing home quality assessments and concepts founded on psychometric theory and Donabedians model. A quality monitoring protocol, based on the staffing taxonomy, was used to assess quality improvement systems. Results from 48 US nursing facilities indicate that most long-term care facilities struggle with staffing allocation and the integration of staffing into a quality monitoring process.