REGULAR RESEARCH ARTICLES
Increased Risk Among Older Veterans of Prescribing Psychotropic Medication in the Absence of Psychiatric Diagnoses Ilse R. Wiechers, M.D., M.P.P., Paul D. Kirwin, M.D., Robert A. Rosenheck, M.D.
Objective: This study uses Veterans Health Administration (VHA) pharmacy and encounter claims to evaluate the use of psychotropic medications without a psychiatric diagnosis across age groups. Methods: National VHA administrative data for fiscal year 2010 (FY2010) were used to identify all veterans who filled a prescription for at least one psychotropic medication from VHA (N ¼ 1.85 million). Bivariate and multivariate analyses were used to compare the proportion of these veterans without any psychiatric diagnosis, across age groups, adjusting for possible medical indications. Analyses were repeated for six different classes of psychotropic medications and comparing mental health utilizers and nonemental health utilizers. Comparisons were made to prescribing of HIV and diabetes medications without an indicated diagnosis. Results: Of all VHA patients prescribed a psychotropic medication in FY2010, 30% had no psychiatric diagnosis, with highest proportions among veterans ages 65e85. This practice was most frequent among nonmental health utilizers and far more prevalent for psychotropic medications than for HIV or diabetes medications. Logistic regression analysis found that age greater than 65 was the strongest predictor of being prescribed a psychotropic without a psychiatric diagnosis. Adjustment for possible medical use of psychotropics and overall medical comorbidity did not substantially alter these trends. Conclusion: Older veterans, especially those not using specialty mental healthcare, are more likely to be prescribed psychotropic medications in the absence of a psychiatric diagnosis, perhaps representing unnecessary use, under-diagnosis of mental illness, or incomplete documentation. (Am J Geriatr Psychiatry 2014; 22:531e539) Key Words: Psychotropic medication, pharmacoepidemiology, veterans
INTRODUCTION Recent reports have generated concern by finding that certain classes of psychotropic medications are
frequently prescribed in the absence of a psychiatric diagnosis.1e7 Studies evaluating antidepressant prescribing have noted older patients are at increased risk for exposure to this practice.1e3 In a cohort of
Received July 10, 2013; revised October 4, 2013; accepted October 9, 2013. From the Robert Wood Johnson Foundation, Clinical Scholars Program (IRW) and Department of Psychiatry (PK, RR), Yale School of Medicine, New Haven, CT; and the VA Connecticut Healthcare System (IRW, PK, RR), West Haven, CT. Presented in part at the annual meeting of the American Association of Geriatric Psychiatry, Washington, DC, March 16e19, 2012. Send correspondence and reprint requests to Ilse R. Wiechers, M.D., M.P.P., Robert Wood Johnson Foundation Clinical Scholar, Yale School of Medicine & US Department of Veterans Affairs, P.O. Box 208088, 333 Cedar St., IE-61 SHM, New Haven, CT 06520-8088. e-mail: [email protected] Ó 2014 American Association for Geriatric Psychiatry http://dx.doi.org/10.1016/j.jagp.2013.10.007
Am J Geriatr Psychiatry 22:6, June 2014
531
Psychotropic Medication Use in Older Veterans noninstitutionalized older adults newly started on an antidepressant, anxiolytic, or antipsychotic, nearly half of the patients did not meet criteria for a mental health disorder.5 Other reports have found that multiple classes of psychotropics are commonly prescribed without evidence of a psychiatric diagnosis but have not fully explored the association of this practice with older age (greater than 65 years).7 In addition, the existing literature has not explored the risk to older patients of this phenomenon while considering the potential confounding effect of medical indications for psychotropic use (e.g., for pain, headache, seizures). Given older patients are more likely to have multiple medical comorbidities, accounting for this medical burden may help explain a portion of what has been reported previously in the literature as prescribing without a psychiatric diagnosis. These prescribing practices are a concern for several reasons. First, over the past two decades the prescribing of psychotropic medications has expanded substantially,8,9 with most prescribing coming from nonpsychiatric providers.10e12 Older patients are more likely to receive their mental health treatment in primary care settings,13 meaning a growing number of patients are potentially at risk for exposure to this practice. Second, elderly patients are at substantial risk of adverse events from psychotropic medications, particularly from anticholinergic effects (e.g., constipation, urinary retention, delirium), antihistaminic effects (e.g., sedation), and antiadrenergic effects (e.g., orthostatic hypotension).14e16 Psychotropic medications are also associated with increased morbidity and mortality, particularly related to falls in the elderly.17e20 Thus, unnecessary use of psychotropic medications in this population (i.e., use in the absence of a clear diagnostic indication) may pose a potential risk to health. This study uses national data from administrative records of outpatients treated by the Veterans Health Administration (VHA) to determine rates of psychotropic medication prescribing in the absence of psychiatric diagnosis across age groups. We aim to fill several gaps in the current literature: (1) to provide analyses with more granular older age groups and across six broad classes of psychotropics, (2) to include medical indications and severity of comorbid general medical illness in our analyses, and (3) to evaluate a large sample from a health system where
532
economic incentives and considerations for not giving a psychiatric diagnosis are not at issue. Additionally, we examined prescribing in the absence of a coded diagnostic indication for HIV and diabetes medications as an attempt to provide a comparison for poor documentation of diagnoses on encounter forms. We hypothesize that prescribing psychotropics in the absence of a psychiatric diagnosis increases with age, is attenuated by potential medical indications for use and medical comorbditiy, and is substantially more common than prescribing HIV or diabetes medications without an indicated diagnosis.
METHODS Sample and Data Sources The sample includes all outpatients who received at least one prescription for a psychotropic medication (defined below) in VHA nationally during fiscal year (FY) 2010 (October 1, 2009 to September 30, 2010). Sociodemographic and diagnostic data were obtained from the outpatient encounter files and data on filled prescriptions from the Decision Support System pharmacy file. Measures Psychotropic medication prescriptions were classified into six groups as follows and measures were constructed to represent any use of each class and the number of prescriptions from within each class: 1. Antidepressants included amitriptyline, amoxapine, clomipramine, desipramine, doxepin, imipramine, nortriptyline, protriptyline, trimipramine, isocarboxazid, phenelzine, selegeline, tranylcypromine, bupropion, citalopram, desvenlafaxine, duloxetine, escitalopram, fluoxetine, fluvoxamine, maprotiline, mirtazapine, nefazodone, paroxetine, sertraline, trazodone, and venlafaxine. 2. Antipsychotics included chlorpromazine, fluphenazine, perphenazine, thioridazine, thiothixene, trifluoperazine, aripiprazole, clozapine, haloperidol, loxapine, molindone, olanzapine, paliperidone, quetiapine, risperidone, and ziprasidone. 3. Benzodiazepine/sedatives/hypnotics included alprazolam, chlordiazepoxide, chlorazepate, clonazepam, diazepam, estazolam, flurazepam, lorazepam, oxazepam, temazepam, triazolam,
Am J Geriatr Psychiatry 22:6, June 2014
Wiechers et al. buspirone, chloral hydrate, eszopiclone, meprobamate, zaleplon, and zolpidem. 4. Stimulants included amphetamine, dextroamphetamine, lisdexamphetamine, methamphetamine, dexmethylphenidate, and methylphenidate. 5. Anticonvulsant mood stabilizers included carbamazepine, gabapentin, lamotrigine, oxcarbazepine, topirimate, valproate sodium, valproic acid, and divalproex sodium. 6. Lithium. HIV medications included lamivudine/zidovudine, emtricitabine/tenofovir disoproxil, tenofovir disoproxil, abacavir/lamivudine, efavirenz, efavirenz/ emtricitabine/tenofovir disoproxil, lopinavir/ritonavir, ritonavir, darunavir, atazanavir, and raltegravir. Diabetes medications included metformin, glyburide, glipizide, and all varieties of insulin. Dichotomous measures were constructed to represent any use of any of these medication classes, similar to the psychotropic class measures above. These medications were chosen in consultation with colleagues in internal medicine, because they represent classes of medications for which there is little off-label use and for which it was expected there would be high rates of same-year diagnoses. Thus, these classes of medications served as a proxy for the base rate of potential incomplete documentation in the system. Available sociodemographic data included age, gender, income, VA serviceeconnected disability and disability pension status, and rural/urban designation based on Rural-Urban Commuting Area codes (depts.washington.edu/uwruca). Clinical data included diagnoses based on International Classification of Diseases, Ninth Revision (ICD-9) codes. A dichotomous variable identified veterans who used any mental health specialty care during the year, defined by use of at least one outpatient encounter with a mental health specialty clinic or an inpatient hospitalization in a psychiatric bed section. Data on age were categorized as follows: less than age 40, ages 40e49, ages 50e64, ages 65e74, ages 75e85, and greater than age 85. Data on psychiatric diagnoses included all ICD-9 codes 290 through 319 as well as 331.00 (Alzheimer dementia; other dementia codes captured in the psychiatric ICD-9 codes above). Categories of medical indications for use of various psychotropic medications were created based on U.S. Food and Drug Administrationeapproved indications for use of the
Am J Geriatr Psychiatry 22:6, June 2014
above listed medications and included: seizures (345.xx), insomnia (780.51, 780.52, 327.0x), migraine and headache (346.x, 784.0), pain and neuropathy (250.6, 357.2, 337.1, 338.x, 719.4, 780.96, 729.1, 728.85, 781.0, 053.12, 729.2, 352.1, 350.1), narcolepsy (347), nausea and vomiting (787.0x), and pruritus (698.x). HIV/AIDS ICD-9 codes included 042 through 044, and diabetes mellitus codes included 2500 through 2509. The Charlson Comorbidity Index, a weighted index determined by combining scores for components based on outpatient diagnoses and patient age, was used to assess the overall severity of medical comorbidity.21 Because age was central to our analyses, we only used the component of the index based on outpatient diagnoses.
Analyses First, we present descriptive data on the entire sample including sociodemographic characteristics and diagnostic and psychotropic medication frequencies. Subsequent analyses were completed separately for all veterans, for those using mental health specialty care at least once during the fiscal year (mental health utilizers), and for those not using any mental health specialty care (nonemental health utilizers). The latter group included all nonemental health clinics and thus includes primary care and medical and surgical specialty clinics. Analyses were based on two assumptions intended to avoid overestimation of the use of psychotropics without a diagnosis: (1) that a diagnosis of any mental illness may justify the use of any psychotropic medication and (2) that any medical illness that is an indication for use of a particular psychotropic medication may justify use of any psychotropic medication. Among veterans receiving any psychotropic medication, we completed cross-tabulations of age by presence of any psychiatric diagnosis for the entire sample and separately for mental health utilizers and nonemental health utilizers. Because of to the large sample size, virtually all c2 analyses were highly statistically significant (p
Increased Risk Among Older Veterans of Prescribing Psychotropic Medication in the Absence of Psychiatric Diagnoses Ilse R. Wiechers, M.D., M.P.P., Paul D. Kirwin, M.D., Robert A. Rosenheck, M.D.
Objective: This study uses Veterans Health Administration (VHA) pharmacy and encounter claims to evaluate the use of psychotropic medications without a psychiatric diagnosis across age groups. Methods: National VHA administrative data for fiscal year 2010 (FY2010) were used to identify all veterans who filled a prescription for at least one psychotropic medication from VHA (N ¼ 1.85 million). Bivariate and multivariate analyses were used to compare the proportion of these veterans without any psychiatric diagnosis, across age groups, adjusting for possible medical indications. Analyses were repeated for six different classes of psychotropic medications and comparing mental health utilizers and nonemental health utilizers. Comparisons were made to prescribing of HIV and diabetes medications without an indicated diagnosis. Results: Of all VHA patients prescribed a psychotropic medication in FY2010, 30% had no psychiatric diagnosis, with highest proportions among veterans ages 65e85. This practice was most frequent among nonmental health utilizers and far more prevalent for psychotropic medications than for HIV or diabetes medications. Logistic regression analysis found that age greater than 65 was the strongest predictor of being prescribed a psychotropic without a psychiatric diagnosis. Adjustment for possible medical use of psychotropics and overall medical comorbidity did not substantially alter these trends. Conclusion: Older veterans, especially those not using specialty mental healthcare, are more likely to be prescribed psychotropic medications in the absence of a psychiatric diagnosis, perhaps representing unnecessary use, under-diagnosis of mental illness, or incomplete documentation. (Am J Geriatr Psychiatry 2014; 22:531e539) Key Words: Psychotropic medication, pharmacoepidemiology, veterans
INTRODUCTION Recent reports have generated concern by finding that certain classes of psychotropic medications are
frequently prescribed in the absence of a psychiatric diagnosis.1e7 Studies evaluating antidepressant prescribing have noted older patients are at increased risk for exposure to this practice.1e3 In a cohort of
Received July 10, 2013; revised October 4, 2013; accepted October 9, 2013. From the Robert Wood Johnson Foundation, Clinical Scholars Program (IRW) and Department of Psychiatry (PK, RR), Yale School of Medicine, New Haven, CT; and the VA Connecticut Healthcare System (IRW, PK, RR), West Haven, CT. Presented in part at the annual meeting of the American Association of Geriatric Psychiatry, Washington, DC, March 16e19, 2012. Send correspondence and reprint requests to Ilse R. Wiechers, M.D., M.P.P., Robert Wood Johnson Foundation Clinical Scholar, Yale School of Medicine & US Department of Veterans Affairs, P.O. Box 208088, 333 Cedar St., IE-61 SHM, New Haven, CT 06520-8088. e-mail: [email protected] Ó 2014 American Association for Geriatric Psychiatry http://dx.doi.org/10.1016/j.jagp.2013.10.007
Am J Geriatr Psychiatry 22:6, June 2014
531
Psychotropic Medication Use in Older Veterans noninstitutionalized older adults newly started on an antidepressant, anxiolytic, or antipsychotic, nearly half of the patients did not meet criteria for a mental health disorder.5 Other reports have found that multiple classes of psychotropics are commonly prescribed without evidence of a psychiatric diagnosis but have not fully explored the association of this practice with older age (greater than 65 years).7 In addition, the existing literature has not explored the risk to older patients of this phenomenon while considering the potential confounding effect of medical indications for psychotropic use (e.g., for pain, headache, seizures). Given older patients are more likely to have multiple medical comorbidities, accounting for this medical burden may help explain a portion of what has been reported previously in the literature as prescribing without a psychiatric diagnosis. These prescribing practices are a concern for several reasons. First, over the past two decades the prescribing of psychotropic medications has expanded substantially,8,9 with most prescribing coming from nonpsychiatric providers.10e12 Older patients are more likely to receive their mental health treatment in primary care settings,13 meaning a growing number of patients are potentially at risk for exposure to this practice. Second, elderly patients are at substantial risk of adverse events from psychotropic medications, particularly from anticholinergic effects (e.g., constipation, urinary retention, delirium), antihistaminic effects (e.g., sedation), and antiadrenergic effects (e.g., orthostatic hypotension).14e16 Psychotropic medications are also associated with increased morbidity and mortality, particularly related to falls in the elderly.17e20 Thus, unnecessary use of psychotropic medications in this population (i.e., use in the absence of a clear diagnostic indication) may pose a potential risk to health. This study uses national data from administrative records of outpatients treated by the Veterans Health Administration (VHA) to determine rates of psychotropic medication prescribing in the absence of psychiatric diagnosis across age groups. We aim to fill several gaps in the current literature: (1) to provide analyses with more granular older age groups and across six broad classes of psychotropics, (2) to include medical indications and severity of comorbid general medical illness in our analyses, and (3) to evaluate a large sample from a health system where
532
economic incentives and considerations for not giving a psychiatric diagnosis are not at issue. Additionally, we examined prescribing in the absence of a coded diagnostic indication for HIV and diabetes medications as an attempt to provide a comparison for poor documentation of diagnoses on encounter forms. We hypothesize that prescribing psychotropics in the absence of a psychiatric diagnosis increases with age, is attenuated by potential medical indications for use and medical comorbditiy, and is substantially more common than prescribing HIV or diabetes medications without an indicated diagnosis.
METHODS Sample and Data Sources The sample includes all outpatients who received at least one prescription for a psychotropic medication (defined below) in VHA nationally during fiscal year (FY) 2010 (October 1, 2009 to September 30, 2010). Sociodemographic and diagnostic data were obtained from the outpatient encounter files and data on filled prescriptions from the Decision Support System pharmacy file. Measures Psychotropic medication prescriptions were classified into six groups as follows and measures were constructed to represent any use of each class and the number of prescriptions from within each class: 1. Antidepressants included amitriptyline, amoxapine, clomipramine, desipramine, doxepin, imipramine, nortriptyline, protriptyline, trimipramine, isocarboxazid, phenelzine, selegeline, tranylcypromine, bupropion, citalopram, desvenlafaxine, duloxetine, escitalopram, fluoxetine, fluvoxamine, maprotiline, mirtazapine, nefazodone, paroxetine, sertraline, trazodone, and venlafaxine. 2. Antipsychotics included chlorpromazine, fluphenazine, perphenazine, thioridazine, thiothixene, trifluoperazine, aripiprazole, clozapine, haloperidol, loxapine, molindone, olanzapine, paliperidone, quetiapine, risperidone, and ziprasidone. 3. Benzodiazepine/sedatives/hypnotics included alprazolam, chlordiazepoxide, chlorazepate, clonazepam, diazepam, estazolam, flurazepam, lorazepam, oxazepam, temazepam, triazolam,
Am J Geriatr Psychiatry 22:6, June 2014
Wiechers et al. buspirone, chloral hydrate, eszopiclone, meprobamate, zaleplon, and zolpidem. 4. Stimulants included amphetamine, dextroamphetamine, lisdexamphetamine, methamphetamine, dexmethylphenidate, and methylphenidate. 5. Anticonvulsant mood stabilizers included carbamazepine, gabapentin, lamotrigine, oxcarbazepine, topirimate, valproate sodium, valproic acid, and divalproex sodium. 6. Lithium. HIV medications included lamivudine/zidovudine, emtricitabine/tenofovir disoproxil, tenofovir disoproxil, abacavir/lamivudine, efavirenz, efavirenz/ emtricitabine/tenofovir disoproxil, lopinavir/ritonavir, ritonavir, darunavir, atazanavir, and raltegravir. Diabetes medications included metformin, glyburide, glipizide, and all varieties of insulin. Dichotomous measures were constructed to represent any use of any of these medication classes, similar to the psychotropic class measures above. These medications were chosen in consultation with colleagues in internal medicine, because they represent classes of medications for which there is little off-label use and for which it was expected there would be high rates of same-year diagnoses. Thus, these classes of medications served as a proxy for the base rate of potential incomplete documentation in the system. Available sociodemographic data included age, gender, income, VA serviceeconnected disability and disability pension status, and rural/urban designation based on Rural-Urban Commuting Area codes (depts.washington.edu/uwruca). Clinical data included diagnoses based on International Classification of Diseases, Ninth Revision (ICD-9) codes. A dichotomous variable identified veterans who used any mental health specialty care during the year, defined by use of at least one outpatient encounter with a mental health specialty clinic or an inpatient hospitalization in a psychiatric bed section. Data on age were categorized as follows: less than age 40, ages 40e49, ages 50e64, ages 65e74, ages 75e85, and greater than age 85. Data on psychiatric diagnoses included all ICD-9 codes 290 through 319 as well as 331.00 (Alzheimer dementia; other dementia codes captured in the psychiatric ICD-9 codes above). Categories of medical indications for use of various psychotropic medications were created based on U.S. Food and Drug Administrationeapproved indications for use of the
Am J Geriatr Psychiatry 22:6, June 2014
above listed medications and included: seizures (345.xx), insomnia (780.51, 780.52, 327.0x), migraine and headache (346.x, 784.0), pain and neuropathy (250.6, 357.2, 337.1, 338.x, 719.4, 780.96, 729.1, 728.85, 781.0, 053.12, 729.2, 352.1, 350.1), narcolepsy (347), nausea and vomiting (787.0x), and pruritus (698.x). HIV/AIDS ICD-9 codes included 042 through 044, and diabetes mellitus codes included 2500 through 2509. The Charlson Comorbidity Index, a weighted index determined by combining scores for components based on outpatient diagnoses and patient age, was used to assess the overall severity of medical comorbidity.21 Because age was central to our analyses, we only used the component of the index based on outpatient diagnoses.
Analyses First, we present descriptive data on the entire sample including sociodemographic characteristics and diagnostic and psychotropic medication frequencies. Subsequent analyses were completed separately for all veterans, for those using mental health specialty care at least once during the fiscal year (mental health utilizers), and for those not using any mental health specialty care (nonemental health utilizers). The latter group included all nonemental health clinics and thus includes primary care and medical and surgical specialty clinics. Analyses were based on two assumptions intended to avoid overestimation of the use of psychotropics without a diagnosis: (1) that a diagnosis of any mental illness may justify the use of any psychotropic medication and (2) that any medical illness that is an indication for use of a particular psychotropic medication may justify use of any psychotropic medication. Among veterans receiving any psychotropic medication, we completed cross-tabulations of age by presence of any psychiatric diagnosis for the entire sample and separately for mental health utilizers and nonemental health utilizers. Because of to the large sample size, virtually all c2 analyses were highly statistically significant (p
