Journal of Gerontology: MEDICAL SCIENCES 1992, Vol.47, No. 6, M183-MI88
Copyright 1992 by The Gerontulogical Society of America
Longitudinal Study of Psychotropic Drug Use by Elderly Nursing Home Residents Judith Garrard,1 Trudy Dunham,2 Lukas Makris,3 Susan Cooper,4 Leonard L. Heston,5 Edward R. Ratner,6 Daniel Zelterman,1 and Robert L. Kane1
In this longitudinal study of patterns of use of psychotropic drugs by a cohort of elderly nursing home residents (N = 5,752), drug use was examined upon admission, 3 months later, and at discharge/end of study. At each time point, 17% of the cohort used neuroleptics. Half of the subjects discontinued neuroleptics at each time point; however, a similar number were initiated on the drug. Benzodiazepines were used by 21%, 15%, and 15% at each of the three time points, respectively. Twice as many people were taken off benzodiazepines as initiated on them following admission. The 5% rate of antidepressant use was constant across the three time periods, although only half of those who took antidepressants upon admission were also taking them upon discharge/end of study. The amount of change due to discontinuation of these drugs and adjustment in dosage levels challenges the stereotype of the "neglected psychotropic drug user" in nursing homes.
their introduction in the 1950s, psychotropic drugs S INCE have played an important role in the treatment of psychosis, anxiety, depression, and other mental illnesses (1,2). Because the benefits of these drugs come with attendant risks, concerns about their use by nursing home residents have been raised by the public (3), the legislature (4), and scientific bodies (5). These concerns led to legislation (6) and recent federal regulation (7) of psychotropic drug use in Medicare and Medicaid certified nursing homes. Studies (8-15) of psychotropic drug use in long-term care facilities with sufficient numbers of residents or multiple sites are limited in number and, with two exceptions (16,17), have focused exclusively on residents already living in the nursing homes in which drug use was assessed crosssectionally rather than longitudinally. The two exceptions were a study of neuroleptic use by an admissions cohort followed up to 12-24 months (16) and another that included data on antidepressant use by an admissions cohort over a 12-month period (17). In general, there is a paucity of research about the initiation and discontinuation of psychotropic drugs, or modifications in drug dosages over time. This study used data from a 1976-84 evaluation of the nursing-home based geriatric nurse practitioner (GNP) model in which the outcome variable was quality of care, including psychotropic drug use (18). In this initial study, residents were followed in a retrospective record review over a 12- to 24-month period or until discharge, whichever occurred first. Results showed that employment of GNPs was cost neutral for the nursing homes, and their presence was associated with moderate improvement in quality of patient care (19). This and related research (20) provided a basis for federal legislation permitting fiscal reimbursement
of nurse practitioners in nursing home care. Using the same data base, we reported another study (16) in which we found that 50% of the admissions cohort who used neuroleptics lacked one or more of the diagnoses of mental illness as defined in the HCFA Interpretative Guidelines published in February, 1989(7). For this article, we use the same data base to describe previously unreported analyses of longitudinal use of neuroleptics, benzodiazepines, and antidepressants by nursing home elderly from their admission to the nursing home, 3 months later, and upon discharge or end of study. We also describe the statistical association between use of these three classes of psychotropic drugs and patient demographic characteristics, source of admission, source of payment, and restraint use. METHODS
The methods used in the parent study have been described in greater detail elsewhere (18) and can be summarized as follows: Subjects A convenience sample of 60 Medicare/Medicaid certified skilled nursing and intermediate care facilities were used: 30 (out of 102) facilities participating in a program to train GNPs were matched on the basis of bed size, type of ownership, affiliation, and rural/urban location to 30 homes without GNPs. None was a facility for the mentally ill or mentally retarded. The 60 nursing homes were located in eight western states of the United States (Arizona, California, Colorado, Idaho, Washington, Oregon, Montana, New Mexico). Compared to a national profile of skilled nursing M183
Downloaded from http://geronj.oxfordjournals.org/ at Cornell University Library on July 17, 2015
'School of Public Health, 2College of Education, 4College of Pharmacy, the University of Minnesota. 3 Merck Research Laboratories, West Point, Pennsylvania. 5 The Institute of Mental Illness Training and Research, University of Washington. 6 Abbott Northwestern Hospital, Minneapolis.
M184
GARRARD ET AL.
Demographic Variables Physicians' and nurses' notes, log or pass book, business office forms, and hospital discharge summaries were abstracted on the basis of a standardized record review form by nurse abstracters not affiliated with the nursing homes. For each subject, demographic variables included birthdate, source of admission (hospital, community, other nursing home), source of payment (presence or absence of Medicaid, which includes payments for nursing home care of indigent elderly), and documentation of restraint use. Gender identification was inadvertently omitted in the initial data collection; however, previous research has found no malefemale differences in use of psychotropics among nursing home elderly (11).
Drug Use Variable Based on medication records maintained by the nurse responsible for administering the drugs, use of up to 10 drugs per subject per time period was recorded in the form of trade name, dosage, and total number of doses per drug. Thus these data were based on actual use of the medications and not on prescription records or claims files. Data were collected for three two-week periods: upon admission, 3 months later, and the first 2 weeks of the month prior to discharge or end of study. The focus in the original study was on total drug dosage regardless of prescribing patterns, and no differentiation was made between use on a regular basis vs use as-needed (prn). We converted the most commonly used psychotropic agents to standard dose equivalents within each drug category (21). Neuroleptic agents (and dose equivalents) were: chlorpromazine (50 mg), haloperidol (1 mg), and thioridazine (50 mg). Benzodiazepines were: alprazolam (0.5 mg), clorazepate (7.5 mg), chlordiazepoxide (10 mg), diazepam (5 mg), flurazepam (15 mg), lorazepam (1 mg), oxazepam (10 mg), temazepam (15 mg), and triazolam (0.25 mg). Antidepressants consisted of amitriptyline (10 mg), doxepin (lOmg), and imipramine(lOmg). Dose equivalence was the basic datum. We did not distinguish among different agents within a class, e.g., between two neuroleptics such as thioridazine and chlorpromazine. Statistical Methods Results are described in the form of tree diagrams and univariate analyses in which the level of statistical significance was p < .05. For many variables it is impossible to infer cause and effect. For example, these data did not permit us to answer the question, does cessation of drug use lead to nursing home discharge or does impending discharge decrease the likelihood of drug use. RESULTS
General Characteristics Demographic characteristics. — This admissions cohort (N = 5,752) had a mean age of 83.3 years (SD = 7.8), 99% were Caucasian, and 28% were Medicaid recipients upon admission. Sixty percent entered the nursing home directly from a hospital, 21% from the community, and 19% transferred from another nursing home. By the end of the study, 19% had died, 15% were discharged alive to the community, 7% were discharged to a hospital with no further information, and 59% were still alive in the nursing home. Drug prevalence rates. — During the first 2 weeks after entrance into the nursing homes, 17% (n = 996) of the cohort used neuroleptics, 21% (n = 1,204) used benzodiazepines, and 5% (n = 291) used antidepressants. Approximately one-fourth of all psychotropic drug users took agents in two or more categories concurrently: 23% (/? = 231) of the neuroleptic users also used drugs in either or both of the other two categories, as did 21% (n = 249) of benzodiazepine users, and 29% (n = 206) of antidepressant users. Of special concern were 30 admissions who had
Downloaded from http://geronj.oxfordjournals.org/ at Cornell University Library on July 17, 2015
facilities participating in the Medicare/Medicaid programs in 1974 (8), the 60 nursing homes in this sample were larger in bed capacity (5% of our sample had < 50 beds compared to 19% nationally; 33% vs 41%, respectively, had 50-99 beds; and 62% vs 41%, respectively, had > 99 beds), underrepresented proprietary homes (53% in our sample vs 73% nationally), and overrepresented those with religious affiliations (22% in our sample vs 9% nationally). Differences between this sample and the national profile were statistically significant for bed size (x2 = 13.31, df = 2,p < .001) and type of ownership (x2 = 18.06, df = 3, p < .001). Details about staffing levels within the nursing homes, physician characteristics or their individual prescribing patterns were not collected in the initial study (18). An admission was defined as a person admitted to the nursing home for any reason during the study enrollment period of 12 months. This definition included individuals who had never lived in a nursing home and former nursing home residents discharged previously and readmitted during the study enrollment period. Only the individual's first admission during the study enrollment period was considered. No subject was used more than once, and no subject was replaced upon discharge. Eligibility criteria required that the individual be 65 years of age or older and have a length of stay in a study nursing home of 6 weeks or more. There were no restrictions by source of admission or source of payment. The initial study found no GNP/non GNP differences in psychotropic drug use by this admissions cohort (18,19). In order to generate a sufficient sample size within each nursing home, three random samples of admissions were drawn within each nursing home, one for each of three years. A nonproportionate, simple random sampling design was used, with the sampling frame consisting of all admissions over each 12-month period who met the subject criteria. The start date of the three-year study period varied between 1976-83 across these 60 facilities. Subjects were followed from admission to the nursing home until either discharge or end of study, whichever occurred first. Discharge was defined as death, discharge alive to community, transfer to another facility, or hospitalization with no return to nursing home. Each subject was thus followed for up to 12 months. The resulting random sample consisted of 5,752 subjects, approximately equally distributed among the 60 nursing homes.
M185
PSYCHOTROPIC DRUG USE
Dosage levels. — Mean dosage levels of neuroleptics were generally in a range considered moderate for elderly persons during each of the three assessment periods (22). Upon admission, 82% of neuroleptic users had an average daily dose equivalent of ^ 50 mg of chlorpromazine, and 93% used =s 100 mg. Of those on benzodiazepines, 74% used an average daily dose equivalent of ^ 5 mg of diazepam, and 92% used =s 10 mg. The level of antidepressant use was the equivalent of =s 50 mg of amitriptyline by 74% of users and ^ 100 mg by 96% on a daily basis.
and no difference was found. Of the 1,580 admissions on Medicaid, 18% used neuroleptics, 18% used benzodiazepines, and 5% used antidepressants. Of the 2,829 subjects who did not have Medicaid, the corresponding figures were 18%, 22%, and 5%. Patterns of Neuroleptic Use Over Time The rate of neuroleptic use, 17%, was the same at admission to the nursing home, three months later, and at discharge/end of study. As shown in Figure 1, however, these users were not the same individuals at each of the three time points because of discontinuation and initiation of these agents. Only 7% of the cohort used neuroleptics during all three assessment periods. Three months after admission, nearly one-third (30%) of the neuroleptic users at admission were no longer taking the drug, and approximately the same number of admissions who were not using these agents upon entrance to the nursing home had started them. Approximately 30% of those whose use was discontinued at 3 months had been reinitiated on these drugs by discharge/end of study. (It should be emphasized that 30% of neuroleptic users constituted 5.3% of the admissions cohort.) A total of 4,044 admissions had a length of stay of 3 months or more. Of these, 72% did not use neuroleptics at any of the three assessment periods. For those discharged less than 3 months after admission, 83% had no neuroleptic use at either entrance to or exit from the nursing home. The stereotype of the "neglected neuroleptic user" who is administered these drugs and continued on them at the same dosage level for the duration of his or her nursing home stay did notfitthe majority of neuroleptic users in this study. 3 Months after Admission
Admission to Nursing Home
Upon Discharge/End of Study
^
/
(N-509)
\
^ ^
(N-184) NE at admission (N-996)
,, NE re-fnrtlated
< ^ ^ \ ^ ~ ^ &-—-
NE discontinued at 3 months
Copyright 1992 by The Gerontulogical Society of America
Longitudinal Study of Psychotropic Drug Use by Elderly Nursing Home Residents Judith Garrard,1 Trudy Dunham,2 Lukas Makris,3 Susan Cooper,4 Leonard L. Heston,5 Edward R. Ratner,6 Daniel Zelterman,1 and Robert L. Kane1
In this longitudinal study of patterns of use of psychotropic drugs by a cohort of elderly nursing home residents (N = 5,752), drug use was examined upon admission, 3 months later, and at discharge/end of study. At each time point, 17% of the cohort used neuroleptics. Half of the subjects discontinued neuroleptics at each time point; however, a similar number were initiated on the drug. Benzodiazepines were used by 21%, 15%, and 15% at each of the three time points, respectively. Twice as many people were taken off benzodiazepines as initiated on them following admission. The 5% rate of antidepressant use was constant across the three time periods, although only half of those who took antidepressants upon admission were also taking them upon discharge/end of study. The amount of change due to discontinuation of these drugs and adjustment in dosage levels challenges the stereotype of the "neglected psychotropic drug user" in nursing homes.
their introduction in the 1950s, psychotropic drugs S INCE have played an important role in the treatment of psychosis, anxiety, depression, and other mental illnesses (1,2). Because the benefits of these drugs come with attendant risks, concerns about their use by nursing home residents have been raised by the public (3), the legislature (4), and scientific bodies (5). These concerns led to legislation (6) and recent federal regulation (7) of psychotropic drug use in Medicare and Medicaid certified nursing homes. Studies (8-15) of psychotropic drug use in long-term care facilities with sufficient numbers of residents or multiple sites are limited in number and, with two exceptions (16,17), have focused exclusively on residents already living in the nursing homes in which drug use was assessed crosssectionally rather than longitudinally. The two exceptions were a study of neuroleptic use by an admissions cohort followed up to 12-24 months (16) and another that included data on antidepressant use by an admissions cohort over a 12-month period (17). In general, there is a paucity of research about the initiation and discontinuation of psychotropic drugs, or modifications in drug dosages over time. This study used data from a 1976-84 evaluation of the nursing-home based geriatric nurse practitioner (GNP) model in which the outcome variable was quality of care, including psychotropic drug use (18). In this initial study, residents were followed in a retrospective record review over a 12- to 24-month period or until discharge, whichever occurred first. Results showed that employment of GNPs was cost neutral for the nursing homes, and their presence was associated with moderate improvement in quality of patient care (19). This and related research (20) provided a basis for federal legislation permitting fiscal reimbursement
of nurse practitioners in nursing home care. Using the same data base, we reported another study (16) in which we found that 50% of the admissions cohort who used neuroleptics lacked one or more of the diagnoses of mental illness as defined in the HCFA Interpretative Guidelines published in February, 1989(7). For this article, we use the same data base to describe previously unreported analyses of longitudinal use of neuroleptics, benzodiazepines, and antidepressants by nursing home elderly from their admission to the nursing home, 3 months later, and upon discharge or end of study. We also describe the statistical association between use of these three classes of psychotropic drugs and patient demographic characteristics, source of admission, source of payment, and restraint use. METHODS
The methods used in the parent study have been described in greater detail elsewhere (18) and can be summarized as follows: Subjects A convenience sample of 60 Medicare/Medicaid certified skilled nursing and intermediate care facilities were used: 30 (out of 102) facilities participating in a program to train GNPs were matched on the basis of bed size, type of ownership, affiliation, and rural/urban location to 30 homes without GNPs. None was a facility for the mentally ill or mentally retarded. The 60 nursing homes were located in eight western states of the United States (Arizona, California, Colorado, Idaho, Washington, Oregon, Montana, New Mexico). Compared to a national profile of skilled nursing M183
Downloaded from http://geronj.oxfordjournals.org/ at Cornell University Library on July 17, 2015
'School of Public Health, 2College of Education, 4College of Pharmacy, the University of Minnesota. 3 Merck Research Laboratories, West Point, Pennsylvania. 5 The Institute of Mental Illness Training and Research, University of Washington. 6 Abbott Northwestern Hospital, Minneapolis.
M184
GARRARD ET AL.
Demographic Variables Physicians' and nurses' notes, log or pass book, business office forms, and hospital discharge summaries were abstracted on the basis of a standardized record review form by nurse abstracters not affiliated with the nursing homes. For each subject, demographic variables included birthdate, source of admission (hospital, community, other nursing home), source of payment (presence or absence of Medicaid, which includes payments for nursing home care of indigent elderly), and documentation of restraint use. Gender identification was inadvertently omitted in the initial data collection; however, previous research has found no malefemale differences in use of psychotropics among nursing home elderly (11).
Drug Use Variable Based on medication records maintained by the nurse responsible for administering the drugs, use of up to 10 drugs per subject per time period was recorded in the form of trade name, dosage, and total number of doses per drug. Thus these data were based on actual use of the medications and not on prescription records or claims files. Data were collected for three two-week periods: upon admission, 3 months later, and the first 2 weeks of the month prior to discharge or end of study. The focus in the original study was on total drug dosage regardless of prescribing patterns, and no differentiation was made between use on a regular basis vs use as-needed (prn). We converted the most commonly used psychotropic agents to standard dose equivalents within each drug category (21). Neuroleptic agents (and dose equivalents) were: chlorpromazine (50 mg), haloperidol (1 mg), and thioridazine (50 mg). Benzodiazepines were: alprazolam (0.5 mg), clorazepate (7.5 mg), chlordiazepoxide (10 mg), diazepam (5 mg), flurazepam (15 mg), lorazepam (1 mg), oxazepam (10 mg), temazepam (15 mg), and triazolam (0.25 mg). Antidepressants consisted of amitriptyline (10 mg), doxepin (lOmg), and imipramine(lOmg). Dose equivalence was the basic datum. We did not distinguish among different agents within a class, e.g., between two neuroleptics such as thioridazine and chlorpromazine. Statistical Methods Results are described in the form of tree diagrams and univariate analyses in which the level of statistical significance was p < .05. For many variables it is impossible to infer cause and effect. For example, these data did not permit us to answer the question, does cessation of drug use lead to nursing home discharge or does impending discharge decrease the likelihood of drug use. RESULTS
General Characteristics Demographic characteristics. — This admissions cohort (N = 5,752) had a mean age of 83.3 years (SD = 7.8), 99% were Caucasian, and 28% were Medicaid recipients upon admission. Sixty percent entered the nursing home directly from a hospital, 21% from the community, and 19% transferred from another nursing home. By the end of the study, 19% had died, 15% were discharged alive to the community, 7% were discharged to a hospital with no further information, and 59% were still alive in the nursing home. Drug prevalence rates. — During the first 2 weeks after entrance into the nursing homes, 17% (n = 996) of the cohort used neuroleptics, 21% (n = 1,204) used benzodiazepines, and 5% (n = 291) used antidepressants. Approximately one-fourth of all psychotropic drug users took agents in two or more categories concurrently: 23% (/? = 231) of the neuroleptic users also used drugs in either or both of the other two categories, as did 21% (n = 249) of benzodiazepine users, and 29% (n = 206) of antidepressant users. Of special concern were 30 admissions who had
Downloaded from http://geronj.oxfordjournals.org/ at Cornell University Library on July 17, 2015
facilities participating in the Medicare/Medicaid programs in 1974 (8), the 60 nursing homes in this sample were larger in bed capacity (5% of our sample had < 50 beds compared to 19% nationally; 33% vs 41%, respectively, had 50-99 beds; and 62% vs 41%, respectively, had > 99 beds), underrepresented proprietary homes (53% in our sample vs 73% nationally), and overrepresented those with religious affiliations (22% in our sample vs 9% nationally). Differences between this sample and the national profile were statistically significant for bed size (x2 = 13.31, df = 2,p < .001) and type of ownership (x2 = 18.06, df = 3, p < .001). Details about staffing levels within the nursing homes, physician characteristics or their individual prescribing patterns were not collected in the initial study (18). An admission was defined as a person admitted to the nursing home for any reason during the study enrollment period of 12 months. This definition included individuals who had never lived in a nursing home and former nursing home residents discharged previously and readmitted during the study enrollment period. Only the individual's first admission during the study enrollment period was considered. No subject was used more than once, and no subject was replaced upon discharge. Eligibility criteria required that the individual be 65 years of age or older and have a length of stay in a study nursing home of 6 weeks or more. There were no restrictions by source of admission or source of payment. The initial study found no GNP/non GNP differences in psychotropic drug use by this admissions cohort (18,19). In order to generate a sufficient sample size within each nursing home, three random samples of admissions were drawn within each nursing home, one for each of three years. A nonproportionate, simple random sampling design was used, with the sampling frame consisting of all admissions over each 12-month period who met the subject criteria. The start date of the three-year study period varied between 1976-83 across these 60 facilities. Subjects were followed from admission to the nursing home until either discharge or end of study, whichever occurred first. Discharge was defined as death, discharge alive to community, transfer to another facility, or hospitalization with no return to nursing home. Each subject was thus followed for up to 12 months. The resulting random sample consisted of 5,752 subjects, approximately equally distributed among the 60 nursing homes.
M185
PSYCHOTROPIC DRUG USE
Dosage levels. — Mean dosage levels of neuroleptics were generally in a range considered moderate for elderly persons during each of the three assessment periods (22). Upon admission, 82% of neuroleptic users had an average daily dose equivalent of ^ 50 mg of chlorpromazine, and 93% used =s 100 mg. Of those on benzodiazepines, 74% used an average daily dose equivalent of ^ 5 mg of diazepam, and 92% used =s 10 mg. The level of antidepressant use was the equivalent of =s 50 mg of amitriptyline by 74% of users and ^ 100 mg by 96% on a daily basis.
and no difference was found. Of the 1,580 admissions on Medicaid, 18% used neuroleptics, 18% used benzodiazepines, and 5% used antidepressants. Of the 2,829 subjects who did not have Medicaid, the corresponding figures were 18%, 22%, and 5%. Patterns of Neuroleptic Use Over Time The rate of neuroleptic use, 17%, was the same at admission to the nursing home, three months later, and at discharge/end of study. As shown in Figure 1, however, these users were not the same individuals at each of the three time points because of discontinuation and initiation of these agents. Only 7% of the cohort used neuroleptics during all three assessment periods. Three months after admission, nearly one-third (30%) of the neuroleptic users at admission were no longer taking the drug, and approximately the same number of admissions who were not using these agents upon entrance to the nursing home had started them. Approximately 30% of those whose use was discontinued at 3 months had been reinitiated on these drugs by discharge/end of study. (It should be emphasized that 30% of neuroleptic users constituted 5.3% of the admissions cohort.) A total of 4,044 admissions had a length of stay of 3 months or more. Of these, 72% did not use neuroleptics at any of the three assessment periods. For those discharged less than 3 months after admission, 83% had no neuroleptic use at either entrance to or exit from the nursing home. The stereotype of the "neglected neuroleptic user" who is administered these drugs and continued on them at the same dosage level for the duration of his or her nursing home stay did notfitthe majority of neuroleptic users in this study. 3 Months after Admission
Admission to Nursing Home
Upon Discharge/End of Study
^
/
(N-509)
\
^ ^
(N-184) NE at admission (N-996)
,, NE re-fnrtlated
< ^ ^ \ ^ ~ ^ &-—-
NE discontinued at 3 months
