Assessing Medication Exposures and Outcomes in the Frail Elderly: Assessing Research Challenges in Nursing Home Pharmacotherapy
Crystal, Stephen PhD*; Gaboda, Dorothy PhD*†; Lucas, Judith EdD*; Gerhard, Tobias PhD*‡; Chakravarty, Sujoy PhD*‡
From the *Center for Health Services Research on Pharmacotherapy, Chronic Disease Management and Outcomes, Institute for Health, Health Care Policy and Aging Research, Rutgers, the State University of New Jersey; †Center for State Health Policy; and ‡Ernest Mario School of Pharmacy, Rutgers University, New Brunswick, NJ.
Supported by the Agency for Healthcare Research and Quality through a cooperative agreement for the Center for Research and Education on Mental Health Therapeutics at Rutgers (HS016097), as part of AHRQ's Centers for Education and Research on Therapeutics (CERTs) program; with additional support from NIMH grants MH058984, MH60831, and MH076206; AHRQ grant HS011825; and The Retirement Research Foundation grant 2007-152.
Presented an earlier version at the AHRQ-sponsored symposium on Clinical and Comparative Effectiveness Research Methods II, June 1–2, 2006, Rockville, MD.
Reprints: Dorothy Gaboda, PhD, Center for Health Services Research on Pharmacotherapy, Chronic Disease, Management and Outcomes, Institute for Health, Health Care Policy and Aging Research, Rutgers University, 55 Commercial Avenue, New Brunswick, NJ 08901. E-mail: firstname.lastname@example.org.
Background: Large administrative datasets such as Medicare and Medicaid claims have much potential utility in clinical and comparative effectiveness (CE) studies. Among their advantages are the inclusion of clinically heterogeneous populations, without exclusions typical in clinical trials; the ability to study extremely large study populations with power to examine differential outcomes across individual drugs, treatment effect modification, and the risk of uncommon outcomes. However, claims data by themselves are subject to many limitations, notably, in their lack of information on such clinical characteristics as functional status, behaviors, and symptoms, which are important both as outcomes and as covariates.
Methods: We describe data from multiple sources including standardized, electronically recorded clinical and functional data from the Nursing Home (NH) Minimum Data Set; prescription drug data from Medicaid and Medicare claims; and facility data. We present the strengths and challenges of using merged data about the NH population to study prescription drug exposures and outcomes in the frail elderly, and suggest strategies to address methodological difficulties.
Results: Merged data from NH sources can support unique study designs in CE research and provide great power. However, given the differing longitudinal structure, timing of observations, and other complex features of the underlying data sources, such studies pose many challenging design and analytic issues.
Conclusions: Integrated data on the NH population have great potential for CE research among frail elderly persons, if methodological and measurement challenges can be adequately addressed.
Although people older than 65 years comprise only about 14% of the US population, they use more than one-third of all medications.1 Ninety percent of the elderly use 1 or more medications regularly, with >40% taking 5 or more different drugs concurrently.2 Within the older population, gaps in outcomes evidence are particularly severe for the elderly with multiple medical conditions and functional impairments, such as those residing in nursing homes (NHs). Even more than older people in general, they are typically excluded from randomized controlled trials (RCTs), the “gold standard” for efficacy and safety of medications, due to feasibility and ethical concerns. However, it may be hazardous to assume that safety and effectiveness findings from trials in other populations generalize to the frail elderly with multimorbidity.
Despite this gap in knowledge, the complex medical problems of NH residents lead them to use medication at especially high rates; for example, 1 study found that the average resident received 9 different medications per month.3 Because of the limitations of the available evidence base and the scarcity of outcome data in this population, the evidence base for safety and effectiveness for much pharmacotherapy among NH residents and comparably frail elders is weak and clinical decisions often rely on generalization from less-frail populations. The “generalizability gap” in the evidence base extends even to treatments that are targeted to the conditions most prevalent in the oldest old. For example, cholinesterase inhibitors and memantine are widely used among NH residents in the hopes of improving cognition and slowing progression of cognitive impairment among residents with dementia, but the preapproval clinical trials of these drugs for safety and effectiveness were mainly conducted in non-NH populations and, indeed, often used NH placement as the study end point. One review found that fewer than 10% of Alzheimer patients met typical eligibility criteria for RCTs of such medications.4 Thus, there is a pressing need for research investigating medication outcomes in the NH population. With more comorbidity and greater medication consumption, elderly patients have heightened central nervous system sensitivity and are at increased risk of exposure to potentially problematic medication, side effects, and drug interactions.5–7
In several cases, analyses of large claims databases have been used to examine outcomes and exposures for which RCTs are unavailable or inadequately powered. For example, analyses of administrative data from British Columbia and Pennsylvania have established that the mortality risk of first-generation APs among elderly with dementia is as great as for the second-generation drugs. This research contributed to the Food and Drug Administration 2008 extension of a black box warning to first generation drugs.8,9 However, the absence of key clinical information, such as measures of function, activities of daily living (ADL), cognitive status, and behavioral symptoms that may confound drug group comparisons, often imposes constraints on the ability for traditional claims-based studies to advance the evidence base. Observational designs that incorporate primary data collection can address these limitations but can be extremely costly and often lack the statistical power to compare alternative treatments (particularly individual medications) and explore treatment effect heterogeneity.
However, within the NH population, a standardized electronic health record, the Minimum Data Set (MDS), provides unique opportunities for comparative effectiveness (CE) research. The MDS includes measures of physical, psychological, and psychosocial functioning and active clinical diagnoses. When these data are merged with diagnosis, treatment, and outcome information from Medicare/Medicaid claims, they create great potential for examining patterns of medication use and select outcomes. However, these studies require an appreciation of the strengths and limitations of the underlying data sources and the complexities of merging data from multiple sources. In this article, we describe these datasets, present the strengths and challenges of using merged data about the NH population to study prescription drug exposures and outcomes in the frail elderly, and suggest strategies to address methodological difficulties.
Medicaid and Medicare claims data provide a detailed, longitudinal record of utilization, diagnoses, procedures, and prescriptions (Table 1). Most residents participate in 1 or both of these important programs.10 Medicare Part A/B claims reflect payments for physician, inpatient hospital, and other healthcare services used by almost all elderly NH residents (except for those who use capitated Medicare programs such as Medicare Advantage, which is an uncommon arrangement for NH residents), and Medicare functions as the first payer for residents eligible for both Medicare and Medicaid. Thus, these data are essential for providing complete diagnostic and utilization histories. Medicare Part A also pays for up to 100 days of postacute short-term NH rehabilitative and skilled nursing care.
Medicaid Analytic Extract data contain claims information standardized for all states. Calendar-year files on Medicaid eligibility, pharmacy claims, service utilization, and payment information contained in the Medicaid Analytic Extract data are extracted from the Medicaid Statistical Information System, the basic source of state-reported eligibility and claims data on the Medicaid population, their characteristics, utilization, and payments (Table 1). Through 2005, Medicaid was the primary payer for prescriptions for residents dually eligible for Medicare and Medicaid.
In 2006, Medicare beneficiaries began receiving prescription drug coverage (Part D) and, under the standard benefit for 2008, beneficiaries paid a deductible and 25% prescription drug costs up to the initial coverage limit of $2510; then, entering the coverage gap (“doughnut hole”), paid the entire cost of prescriptions up to the catastrophic threshold of $5,726. As of February 2009, 17.5 million Medicare beneficiaries were enrolled in stand-alone Part D plans; 6.3 million dually eligible for Medicaid and approximately 3.3 million receiving Low Income Subsidies received assistance during the coverage gap.11 Within NHs, 81% of residents were enrolled in Part D in 2006, whereas 16% had other drug coverage and 3% remained without drug coverage.12 For dual eligibles, comprising 66% of the NH population, the Medicare prescription coverage replaced their previous Medicaid drug coverage. These dual eligibles will not face a coverage gap entailing 100% cost sharing; however, they might be subject to restrictive utilization management practices such as prior authorization policies for the protected drug classes and lack of coverage for the unprotected drug classes, depending on the specific plan characteristics.
Because the initial assignment of dual eligibles to low cost benchmark plans was on a random basis, these differences in access generate variation in medication use that is exogenous to their health status, creating natural experiments that could be used for CE research. In addition, comparing health outcomes between residents in plans with and without the coverage gap, and other distinct plan characteristics, can inform the comparison of alternative coverage mechanisms. Detailed information on plan characteristics and beneficiary utilization enables careful analysis of the impact of commonly used management tools such as cost sharing, prior authorization, and step therapy on patient drug use and associated outcomes.
CLINICAL AND FACILITY DATA
The MDS is a national 350-item summary screening and assessment tool required of all Medicare and Medicaid-certified NHs providing data for all NH residents (Table 1). Extensive clinical and assessment data for individual residents, including physical and psychosocial functioning, active clinical diagnoses, treatments and mental health services, demographics, and payer source, are documented by nursing staff that have been trained using the standardized MDS administrative guidelines and submitted electronically.13,14 Previous studies found that interrater reliabilities of items and internal consistency of scales are generally good to excellent for the revised MDS (v 2.0).15 Good correspondence to research quality instruments has been found for cognition, ADL, and diagnoses, with reliability of .73 for the 62 items in the Disease Diagnosis section and .74 for the 8 items in the Medication Use section.16–18
All Centers for Medicare & Medicaid Services (CMS)-certified NHs are required to complete comprehensive MDS assessments on admission of each resident, annually, and when a resident shows “significant change in status.” A subset of the full MDS is conducted quarterly. The MDS contains select information on medication exposures, including the use of antipsychotics, antidepressants, antianxiety, and sedatives/hypnotics by class for the 7 days before assessment. Specific information about other classes, specific agents, dosage, or prescriber characteristics are not available from the MDS, but data can be supplemented with Medicaid and/or Medicare Part D prescriptions drug claims to provide more detailed information needed on exposure over time for CE and outcomes research. Using rigorous study designs that take into consideration the time-varying nature of antipsychotic use, confounding effects, and the hierarchical data structures inherent in NH research, studies have successfully examined excess risks of several outcomes (eg, hip fracture, venous thromboembolism, cerebrovascular events, ventricular arrhythmias, and diabetes onset) associated with antipsychotic agents and classes using these integrated datasets.19–21
A revised MDS 3.0, projected to be implemented October 2010, has been designed to improve the reliability, accuracy, and usefulness of the MDS.22 The MDS 3.0 data elements will continue to include select information by class on psychotropic medication exposures, such as antipsychotic use received any time during last 7 days or since admission/reentry if less than 7 days. Additional diagnostic codes such as the new code for psychotic disorders other than schizophrenia should increase specificity of recorded psychiatric conditions. In addition to existing resident clinical characteristics such as ADL score, MDS 3.0 will now include several new validated scales (Table 1).
Information about facility characteristics that can influence prescribing patterns is available from the Online Survey Certification and Reporting (OSCAR) system, a uniform computerized data system maintained by CMS.10 OSCAR contains facility and aggregated resident data from all Medicare/Medicaid certified NHs (Table 1). OSCAR data are reported by facilities before, and validated by state agencies as part of, the yearly recertification process. Data are entered into a uniform, computerized database and undergo extensive edit checks by CMS.23 Data on most facility characteristics are considered adequately reliable for research,24 and consequently OSCAR is widely used as the primary source of data on facility characteristics and deficiency citations.10 Facility characteristics such as acuity level and casemix adjust for the differences in clinical characteristics between NH residents, which can lead to selection effects. A variety of NH characteristics have been studied, including facility size, payer mix, staffing, and citations for deficiencies.25,26 Information on quality indicators such as deficiency citations, staffing levels, and physical restraints is extremely useful to test effects or control for variation in quality of care between facilities while assessing the impact of medications and other interventions on health outcomes.27,28 Previous studies have used combined MDS-OSCAR data to examine the impact of NH characteristics on resident outcomes.25,26,29–31
STRENGTHS AND COMPARATIVE ADVANTAGES
Analyses of claims data have several key strengths for studying prescription drug use, including very large numbers of covered lives with relatively comprehensive benefit information of the full continuum of care in most settings, strong representation of vulnerable populations, and diagnostic and treatment information from providers rather than consumers (Table 2). 32 The lack of clinical and contextual information in claims data can be partially ameliorated by supplementing these datasets with data from other routinely collected data sources such as the MDS and OSCAR. For example, MDS data have been linked to Medicare claims and other detailed medication records to study medication use and hospitalizations in NHs.33–36 Within the NH setting, medications are administered by staff, so that nonadherence is less an issue than with community-dwelling elderly. Combining prescription drug claims with MDS assessments substantially increases the potential to examine important details of medication use including dosage, duration, and use of specific drugs within medication classes, as in current work on guideline consistent use of antipsychotic medications in the NH.37
Because of the very large populations available to study, merged NH data offer unique opportunities to examine outcomes in particular subgroups of concern, to compare outcomes across individual drugs within and between classes, to assess risks of rare but serious adverse outcomes not detectable in typical clinical trials, and to examine outcomes over longer follow-up times. These data include large numbers of individuals in vulnerable subgroups such as minority, low-income, and/or individuals with disabilities, providing a vital resource for research on disparities. For example, preliminary analyses of 2005 MDS data by the authors indicate that among 690,000 residents aged 65 years and older with stays of at least 1 year, 12% were African American and 3% Hispanic, 61% had recorded diagnoses of dementia, and 50% were diagnosed with depression.
Very large numbers also increase the ability to examine treatment effect heterogeneity, construct narrowly tailored new-treatment cohorts for pharmacoepidemiologic studies, and facilitate advanced statistical methods such as high dimensionality propensity scoring and instrumental variables. These designs are remarkably demanding of power and may require multiyear national data to generate statistical variation, which facilitates identification and provide sufficient numbers for controlled study of rare conditions or comparisons between specific medications. These data allow extended longitudinal follow-up utilization, diagnoses, procedures, and prescriptions for clinically heterogeneous populations, including those with multiple morbidities. Longitudinal modeling allows within-person analysis of medication utilization and outcomes over time, which can shed light on processes and potential modifiable factors leading to adverse health outcomes.
LIMITATIONS, CHALLENGES, AND STRATEGIES
The Medicare, Medicaid, and MDS data mentioned earlier contain individual beneficiary and provider identifiers, which allow the data to be linked and are subject to the Privacy Act and other Federal government rules and regulations. CMS uses strict security measures to safeguard individual privacy and requires review of data requests by a CMS Privacy Board. All requests for identifiable data must be developed and reviewed with the assistance of the Research Data Assistance Center, and include a Data Use Agreement, detailing specific procedures to assure protection of the data (Table 1). Obtaining CMS data, particularly for multiple years, represents a substantial financial investment and may take several months. Use of these identifiable data likely will also require approval by the Institutional Review Board of the researcher's organization.
In addition to privacy and financial concerns, the datasets require a substantial investment of time in becoming familiar with the associated documentation. A researcher must become familiar with comprehensive data dictionaries detailing diagnosis codes, procedure codes, and National Drug Codes. In planning studies, researchers need to become familiar with published treatment recommendations38,39 and/or team with clinical partners familiar with treatment practices in NH. In addition, projects using these complex data require analysts with advanced programming skills to process the data accurately and efficiently.
Using administrative data requires understanding and accounting for certain well-known limitations of the underlying data sources.29 In the NH context, it is particularly important to use information from multiple sources to account for transitions in settings and payer sources (Table 3). For example, use of medications cannot be observed in claims data for periods of inpatient hospitalization or short-term rehabilitative NH stays funded by Medicare Part A. Such stays often follow a hospitalization, creating a time period during which observation of medication use should be treated as censored. Careful attention is also needed to state variations in policies, particularly before 2006. For example, before Medicare Part D, the New York Medicaid program bundled reimbursement for some drugs (maintenance medications) into Medicaid per diem rates for NH care, and certain states did not report days supplied on Medicaid prescription drug claims in all years.
State variations also exist in MDS data. Most notably, state requirements for quarterly assessments vary, so that less information may be available on quarterly assessments in states not participating in the Prospective Payment System, limiting longitudinal tracking of some measures (eg, comorbidities and psychotic symptoms) to comprehensive assessments only. An important issue that needs further exploration is the extent of facility or state variation in the recording of MDS information, because some states currently base reimbursement for Medicaid and Medicare patients on functional status of residents. Such use of MDS clinical assessment information for payment and for quality measurement has the potential to influence recording.15 In addition, NHs and states have provided varying resources and training for completing the MDS, which may affect reporting of information.40 Differences have been found between facilities in interrater reliability levels of MDS assessment items, but it is unclear which facility characteristics are related to these differences.17
As discussed earlier, many studies have supported the reliability and validity of many of the MDS 2.0 data elements and summary scales. However, other studies have questioned the validity of some elements such as pain frequency and intensity, measures of mood, visual acuity, and incontinence.41–45
Longitudinal tracking of clinical information about NH residents can be complex. Although assessments are completed at approximately 90-day intervals, assessments are also completed when a resident experiences a “significant change” in function, which may change the timing of subsequent assessments. Hospitalizations of residents also interrupt the timing and detail of assessments. Although MDS assessments occur at roughly 90-day intervals, they must be matched to daily drug claims to construct meaningful medication histories. Meaningful assembly of covariates over time is challenging, as health status changes between assessment points, potentially resulting in misclassification of MDS-based measures. To address this issue, we have, for example, limited analyses to MDS assessments within 2 weeks before claims-based exposure/outcome assessment to avoid misclassification of clinical measures from the MDS; however, this may diminish study samples.
In addition, combining MDS data with claims requires careful specification of measures of outcomes, clinical covariates, and treatments. MDS assessments generally reflect status within the 7 days before the assessment. However, some sections specify other time frames. For example, continence items refer to status in the past 14 days, and some items in the mood and behavior patterns section refer to the past 30 days. Therefore, specifying models requires careful attention to temporal relationships and clinical understanding of relevant conditions.
For CE research, the addition of clinical information from MDS assessments performs 2 primary functions: it provides (1) clinical detail that improves ability to control for confounding between comparison groups and (2) outcome measures not available in claims data. MDS-based clinical variables can be integrated as covariates into CE analyses in the same way as variables based on diagnostic codes or drug utilization from claims. Assuming a new-user design, both types of variables would be assessed for a defined preinitiation period (commonly the same period that established new exposure treatment episodes) and used for baseline adjustments between the comparison groups, either using traditional multivariate methods or summary variable approaches such as propensity scores. When particular concern exists about misclassification of MDS-based variables from more distant assessments (eg, when clinical practice suggests that treatment initiation is commonly preceded by sudden changes in important patient characteristics observed in the MDS), analyses can be limited to only those subjects with MDS assessments within a shorter time period (eg, 7 or 14 days) before initiation. Of course, addition of MDS to claims data will only improve confounder adjustment in areas where MDS collects data, ie, largely functional and behavioral measures and does not, for example, provide laboratory or imaging results. Thus, the greatest benefit of merged NH data lies in analyses where behavioral and functional measures constitute important confounders, such as mental health interventions; however, they may also be useful in other clinical areas as more general proxies for frailty. Similarly, MDS-based variables may be used as outcome measures for either binary outcomes (eg, falls, infections, dysphagia, or feeding tube use), or—with adjustment for their baseline values—for continuous variables such as body mass index, cognition, social engagement, physical function, and ADL performance.
In conclusion, RCT data are often inapplicable to populations of frail elderly with complex comorbidities, such as the typical resident population in NHs. A better understanding of prescription drug use and outcomes in the NH population is not only important in its own right but can also serve as an important test bed for outcomes in similarly frail but less well-observed elderly in other settings. The combination of administrative data available for NH residents presents unique opportunities to combine clinical and functional assessments with diagnosis and medication exposure information. The high level of detail on clinical characteristics, functioning, and interventions has the potential for overcoming some of the traditional limitations of administrative data by providing richer clinical information and can support a variety of statistical methods to reduce confounding. New-user designs, advanced statistical methods, such as use of instrumental variables, examination of effect modifiers in heterogeneous populations, and comparisons across individual drugs are all extremely demanding of power. Cell sizes dwindle rapidly, and national multiyear data may be necessary to address rare outcomes.
Merging multiple data sources has great potential but can be challenging to implement for an individual study given the complexity and labor intensiveness of this work. Because highly detailed information about individuals is necessary, data are potentially identifiable, requiring rigorous procedures to assure protection of confidentiality. More effective and systematic linkage of these rich data resources has considerable potential to advance the state of the art in CE, medication safety, and other health services research on the frail elderly.
1.Avorn J. Powerful Medicines: The Benefits, Risks, and Costs of Prescription Drugs. New York: Vintage Books; 2005.
2.Kaufman DW, Kelly JP, Rosenberg L, et al. Recent patterns of medication use in the ambulatory adult population of the US: The Slone Survey. JAMA. 2002;287:337–344.
3.Simoni-Wastila L, Stuart BC, Shaffer T. Over-the-counter drug use by Medicare beneficiaries in nursing homes: implications for practice and policy. J Am Geriatr Soc. 2006;54:1543–1549.
4.Schneider LS, Olin JT, Lyness SA, et al. Eligibility of Alzheimer's disease clinic patients for clinical trials. J Am Geriatr Soc. 1997;45:923–928.
5.Gurwitz JH, Field TS, Avorn J, et al. Incidence and preventability of adverse drug events in nursing homes. Am J Med. 2000;109:87–94.
6.Lau D, Kasper J, Potter DEB, et al. HSR: Health Serv Res. 2004;39:1257–1276.
7.Jeste DV, Blazer D, Casey D, et al. ACNP white paper: Update on use of antipsychotic drugs in elderly persons with dementia. Neuropsychopharmacology. 2008;33:957–970.
8.Schneeweiss S, Setoguchi S, Brookhart A, et al. Risk of death associated with the use of conventional versus atypical antipsychotic drugs among elderly patients. CMAJ. 2007;176:627–632.
9.Wang PS, Schneeweiss S, Avorn J, et al. Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N Engl J Med. 2005;353:2335–2341.
10.Harrington C, Carrillo H, LaCava C. Nursing Facilities, Staffing, Residents, and Facility Deficiencies, 1999 Through 2005. San Francisco, CA: Department of Social and Behavioral Sciences, University of California; 2006.
11.The Henry J. Kaiser Family Foundation. The Medicare Prescription Drug Benefit. Fact sheet #7044-10. 2009. Available at: http://www.kff.org
. Accessed April 6, 2010.
12.Briesacher BA, Soumerai SB, Field TS, et al. Nursing home residents and enrollment in Medicare Part D. J Am Geriatr Soc. 2009;57:1902–1907.
15.Mor V. A comprehensive clinical assessment tool to inform policy and practice: Applications of the Minimum Data Set. Med Care. 2004;42:50–59.
16.Gruber-Baldini AL, Zimmerman SI, Mortimore E, et al. The validity of the Minimum Data Set in measuring the cognitive impairment of persons admitted to nursing homes. J Am Med Dir Assoc. 2000;48:1601–1606.
17.Mor V, Angelelli J. Jones R, et al. Inter-rater reliability of nursing home quality indicators in the US. BMC Health Serv Res. 2003;3:20.
18.Hawes C, Morris JN, Phillips CD, et al. Reliability estimates for the Minimum Data Set for nursing home resident assessment and care screening (MDS). Gerontologist. 1995;35:172–178.
19.Liperoti R, Pedrone C, Lapane K, et al. Venous thromboembolism among elderly patients treated with atypical and conventional antipsychotic agents. Arch Intern Med. 2005;165:2677–2682.
20.Liperoti R, Gambassi G, Lapane KL, et al. Conventional and atypical antipsychotics and the risk of hospitalization for ventricular arrhythmias or cardiac arrest. Arch Intern Med. 2005;165:696–701.
21.Jalbert J, Eaton C, Miller S, Lapane K. Antipsychotic use and the risk of hip fracture among older adults afflicted with dementia. J Am Med Dir Assoc. 2010;11:120–127.
23.Harrington C, Zimmerman D, Karon SL, et al. Nursing home staffing and its relationship to deficiencies. J Gerontol Soc Sci. 2000;55B:S278–S287.
24.Feng Z, Katz PR, Intrator O, et al. Physician and nurse staffing in nursing homes: The role and limitations of the Online Survey Certification and Reporting (OSCAR) system. J Am Med Dir Assoc. 2005;6:27–33.
25.Lapane KL, Hughes CM. Which organizational characteristics are associated with increased management of depression using antidepressants in US nursing homes? Med Care. 2004;42:992–1000.
26.Levin CA, Wei W, Akincigil A, et al. Prevalence and treatment of diagnosed depression among elderly nursing home residents in Ohio. J Am Med Dir Assoc. 2007;8:585–594.
27.Mor V, Berg K, Angelelli J, et al. The quality of quality measurement in US nursing homes. Gerontologist. 2003;43:37–46.
28.O'Neill C, Harrington C, Kitchener M, et al. Quality of care in nursing homes: An analysis of relationships among profit, quality, and ownership. Medical Care. 2003;41:1318–1330.
29.Hughes CM, Lapane KL, Mor V. Influence of facility characteristics on use of antipsychotic medications in nursing homes. Med Care. 2000;38:1164–1173.
30.Gerhard T, Lucas JA, Olfson M, et al. Effect of regulatory warnings on antipsychotic prescribing for nursing home residents with dementia. PDS Pharmacoepidemiol Drug Saf. 2008;17:S120–S121.
31.Intrator O, Castle NG, Mor V. Facility characteristics associated with hospitalization of nursing home residents: Results of a national study. Med Care. 1999;37:228–237.
32.Crystal S, Akincigil A, Bilder S, et al. Studying prescription drug use and outcomes with Medicaid claims data: strengths, limitations, and strategies. Med Care. 2007;45:S58–S65.
33.Gruneir A, Miller SC, Feng Z, et al. Relationship between state Medicaid policies, nursing home racial composition, and the risk of hospitalization for black and white residents. Health Serv Res. 2008;43:869–881.
34.Intrator O, Zinn J, Mor V. Nursing home characteristics and potentially preventable hospitalizations of long-stay residents. J Am Geriatr Soc. 2004;52:1730–1736.
35.Lapane KL, Spooner JJ, Mucha L, et al. Effect of nonsteroidal anti-inflammatory drug use on the rate of gastrointestinal hospitalizations among people living in long-term care. J Am Geriatr Soc. 2001;49:577–584.
36.Quilliam BJ, Lapane KL. Clinical correlates and drug treatment of residents with stroke in long-term care. Stroke. 2001;32:1385–1393.
37.Lucas JA, Bowblis J, Gerhard T, Chakravarty S, Crystal S. Antipsychotic medication (APM) use among nursing home residents: Appropriate or Inappropriate? PDS: Pharmacoepidemiology Drug Saf. 2009;18:S143.
38.American Medical Directors Association. Clinical Practice Guidelines in the Long Term Care Setting. AMDA; 2003.
39.American Geriatrics Society (AGS) and American Association for Geriatric Psychiatry (AAGP). Consensus statement on improving the quality of mental health care in U.S. nursing homes: Management of depression and behavioral symptoms associated with dementia. J Am Geriatr Soc. 2003;51:1287–1298.
40.Shin JH, Scherer Y. Advantages and disadvantages of using MDS data in nursing research. J Gerontol Nurs. 2009;35:7–17.
41.Fisher SE, Burgio LD, Thorn BE, et al. Pain assessment and management in cognitively impaired nursing home residents: Association of certified nursing assistance pain report, Minimum Data Set pain report, and analgesic medication use. J Am Geriatr Soc. 2002;50:152–156.
42.McCurran C. Assessment for depression among nursing home elders: Evaluation of the MDS mood assessment. Geriatr Nurs. 2002;23:103–108.
43.Engle VF, Graney MJ, Chan A. Accuracy and bias of licensed practical nurse and nursing assistant ratings of nursing home residents’ pain. J Gerontol A Bio Sci Med Sci. 2001;56:M405–M411.
44.Bharucha AJ, Vasilescu M, Dew MA, et al. Prevalence of behavioral symptoms: comparison of the Minimum Data Set assessment with research instruments. J Am Med Dir Assoc. 2008;9:244–250.
45.Swanson MW. Care plan assessment for visual status and evaluated vision among nursing home residents. Optom Vis Sci. 1995;72:151–154.
This article has been cited 1 time(s).
© 2010 Lippincott Williams & Wilkins, Inc.