Exposure to psychotropic medications prior to

0 downloads 0 Views 200KB Size Report
drug overdose during the study period, defined by the diagno- sis codes T360–T509 in the ICD-10. These diagnosis codes include all drug overdoses, not just ...
Psychopharmacology DOI 10.1007/s00213-015-3952-8

ORIGINAL INVESTIGATION

Exposure to psychotropic medications prior to overdose: a case-control study Yasuyuki Okumura 1 & Hisateru Tachimori 2 & Toshihiko Matsumoto 3 & Daisuke Nishi 2

Received: 4 February 2015 / Accepted: 29 April 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Rationale Little is known about psychotropic medication prescriptions prior to drug overdose. Objectives We aimed to examine the possible associations between the risk of overdose and exposures to various psychotropic medications. Methods We conducted a matched case-control study of 3 groups of patients aged 12–74 years, using a large-scale health insurance claims database in Japan (population 1.2 million). A total of 351 cases with drug poisoning were compared with two control groups without overdose: 1755 patients with any treatment (general controls) and those with depression (highrisk controls). Current, past, and nonusers were patients most recently exposed to psychotropic medications ≤90 days, 91– 180 days, and ≥180 days before the index date. Results Current sedative-hypnotic use was associated with an increased odds of overdose relative to general control nonusers (odds ratio [OR], 21.5; 95 % confidence interval

[CI], 9.7–47.8) and high-risk control nonusers (OR, 2.6; 95 % CI, 1.9–3.5). In the comparison of cases and high-risk controls, the ORs for overdose were higher among excessive dosage users than among usual dosage users (OR, 4.3; 95 % CI, 3.0–6.1), among barbiturate users than among benzodiazepine/Z-drug only users (OR, 4.5; 95 % CI, 2.3– 8.7), and among multiple provider episodes than among single provider episodes (OR, 4.4; 95 % CI, 1.7–11.0). Psychiatrists prescribed more than 77 % of potentially questionable prescriptions. Conclusions These results highlight the need for psychiatrists to monitor prescribed medications and balance the benefits and risks of pharmacological treatments. Keywords Drug poisoning . Acute poisoning . Drug dependence . Deliberate self-harm . Parasuicide . Suicide . Benzodiazepines . Psychotropic medications

Introduction Electronic supplementary material The online version of this article (doi:10.1007/s00213-015-3952-8) contains supplementary material, which is available to authorized users. * Yasuyuki Okumura [email protected] 1

Research Department, Institute for Health Economics and Policy, Association for Health Economics Research and Social Insurance and Welfare, Tokyo, Japan

2

Department of Mental Health Administration, National Institute of Mental Health, National Center of Neurology and Psychiatry, Tokyo, Japan

3

Department of Drug Dependence Research/Center for Suicide Prevention, National Institute of Mental Health, National Center of Neurology and Psychiatry, Tokyo, Japan

Drug overdose is a major public health concern worldwide. The annual incidence rate of drug poisoning in the USA has been estimated at 232 per 100,000 inhabitants (Xiang et al. 2012). Drug overdose is the one of the leading causes of admissions to emergency departments (Okumura et al. 2012b; Schwake et al. 2009), and more than 75 % of these cases involve intentional poisoning for suicidal purposes rather than accidental poisoning (Hirata et al. 1998; Hovda et al. 2008). Psychotropic medications such as benzodiazepines and antidepressants, which are approved for the treatment of unipolar depression, bipolar depression, schizophrenia, and anxiety disorders, are the most commonly ingested (Hirata et al. 1998; Xiang et al. 2012). Because of the differing availability of drugs, these patterns and causes of drug overdose vary by

Psychopharmacology

location and time period (Islambulchilar et al. 2009); therefore, it is important to maintain information that is up-to-date. Drug overdose prevention involves careful and appropriate prescriptions (Varnik et al. 2011). However, little is known about prescriptions prior to overdose. The only known epidemiological study, which involved a nested case-control study of 629 overdose cases in a cohort of psychiatric patients in Taiwan, found a 2.5-fold increased risk for overdose among users of benzodiazepines, highlighting the need for judicious evaluation of patients receiving benzodiazepines specifically (Shih et al. 2013). Given the limited information regarding other medications and prescription patterns, a more comprehensive picture of psychotropic prescriptions is needed to prevent drug overdoses. Therefore, we aimed to examine the possible associations between the risk of drug overdose and current and past exposure to various psychotropic medications using a large-scale claims database of employees and their family members. In addition, we aimed to explore the associations between the risk of drug overdose and specific prescription patterns of psychotropic medications such as excessive dosing and prescriptions by multiple providers.

Methods Data source We used data from the health insurance claims database developed by the Japan Medical Data Center (JMDC) Co. Ltd., Tokyo, Japan (Kimura et al. 2010). All Japanese citizens are required to be enrolled in one of three types of health insurances: occupation-based, municipality-based, or a separate system for persons aged 75 years. All insured medical and pharmaceutical institutions issue medical and pharmacy claims for each patient every month to ensure reimbursement for the costs of health-related service. In medical claims, each medical institution records clinical information such as sex, age, date of treatment, and diagnosis codes based on the International Statistical Classification of Diseases, 10th Revision (ICD-10). In pharmacy claims, each pharmaceutical institution records procedural information such as drug name, dispensing date, and dosage. Almost all medical and pharmaceutical institutions are obligated to use an online claims system. In Japan, almost all patients who use psychotropic medications are reimbursed by public insurance because the majority of psychotropic medications are prescription drugs rather than over-the-counter drugs. In January 2005, the JMDC started to collect anonymous medical and pharmacy claims from over 50 occupation-based health insurance societies for corporate employees and their family members. The database included 1,720,344 recipients aged 0–74 years as of October 2012, representing

approximately 1.5 % of the total population in Japan (Statistics Bureau 2013). The institutional review board of the JMDC approved the analysis in the present study. Design and setting We conducted a case-control study. The study cohort included patients aged 12–74 years during the 13-month study period (October 1, 2012 to November 30, 2013). Because we planned to retrieve information on prescription patterns within the 6 months before the index date, we further required that the patients had been enrolled in their health insurance at least 6 months before the index date. We compared cases with two control groups. Cases were defined as patients who had an initial definite diagnosis of drug overdose during the study period, defined by the diagnosis codes T360–T509 in the ICD-10. These diagnosis codes include all drug overdoses, not just overdoses from psychotropic medications. We included any type of overdose (i.e., intentional, accidental, and undetermined intent overdose), as in previous studies (Heyerdahl et al. 2009; Okumura et al. 2012a; Reith et al. 2003) because data on external causes (ICD-10 codes: V01-Y98) are not recorded in the database. For cases, the index date was set as the date of admission for overdose. The first control group, or the general controls, consisted of patients who were being treated for any condition and had no history of overdose during the study period. General controls were selected because both cases and general controls were representative of the study cohort. The second control group, or the high-risk controls, consisted of patients who had a definite diagnosis of unipolar depression (ICD-10 codes: F32– F33 and F341) with no history of overdose during the study period. High-risk controls were selected because they were likely to be more similar to cases with respect to potential cofounders than general controls. Unipolar depression was selected as the diagnosis for high-risk controls because it is the most prevalent mental and behavioral disorder diagnosis in patients who overdose (Okumura et al. 2012a). For both control groups, we randomly selected five control participants for each case from the eligible source population matched by sex and age within 5 years. For controls, the index dates were set as the initial date of outpatient or inpatient visits during the study period for any treatment and depression, respectively. Exposure definition For cases and controls, we identified all psychotropic medications prescribed within 6 months prior to the index date, including 101 psychotropic medications classified as sedativeshypnotics, antidepressants, antipsychotics, and mood stabilizers according to the one of the most popular prescription

Psychopharmacology

handbooks in Japan (Urabe et al. 2012). A fixed-combination drug of chlorpromazine-promethazine-phenobarbital (Vegetamin®) was counted as both a sedative-hypnotic and an antipsychotic. Because sulpiride has been approved for the treatment of depression at low doses and schizophrenia at high doses, sulpiride 15 mg/day, imipramine equivalents >300 mg/day, and chlorpromazine equivalents >450 mg/day, respectively, based on the maximum recommended daily doses in the package inserts (Alfresa Pharma 2013; Shionogi 2013; Takeda Pharmaceutical Company 2012). Then, we categorized current users of psychotropic medications into two groups: usual (reference category) and excessive dose.

We identified current users with potentially high-toxicity prescriptions, which included barbiturates, tricyclic antidepressants (TCAs), and lithium because overdose with these agents could lead to severe toxicity and be fatal (Henry et al. 1995; Sarchiapone et al. 2011; Urban et al. 2013). First, we categorized current users of sedatives-hypnotics into three groups: benzodiazepines (BZDs)/Z-drugs only (reference category), barbiturates, and others. Second, we categorized current users of antidepressants into three groups: selective serotonin reuptake inhibitors (SSRIs)/serotonin-noradrenaline reuptake inhibitors (SNRIs) only (reference category), TCAs, and others. Third, we categorized current users of mood stabilizers into two groups: others (reference category) and lithium. We also defined multiple provider episodes (i.e., doctorshopping) as receiving overlapping prescriptions from two or more providers lasting more than 30 days before the index date. We categorized current users of psychotropic medications into two groups: single and multiple providers. Potential confounders We classified a history of mental and behavioral disorders within 180 days before the index date using the criteria developed by the Global Burden of Disease study (Harvard Initiative for Global Health et al. 2008) with some modifications: unipolar depression, bipolar depression, schizophrenia, alcohol use disorders, drug use disorders, anxiety disorders, somatoform disorders, other neurotic disorders, eating disorders, personality disorders, autism spectrum disorders, intellectual disability, and others (Table 1). We also classified a history of comorbid conditions within 180 days before the index date using the Sundararajan version of the Charlson Comorbidity Index (Sundararajan et al. 2007). Statistical analyses We used conditional logistic regression analysis to estimate the odds ratios (ORs) and their 95 % confidence intervals (CIs) for overdose among current and past users of psychotropic medications compared with nonusers. We estimated the ORs in separate comparisons of cases with each of the control groups. First, we estimated the ORs without adjustment for covariates. Then, we calculated the adjusted ORs, controlling for sex, age groups, history of mental and behavioral disorders, and history of comorbid conditions. We repeated all analyses for sedatives-hypnotics, antidepressants, antipsychotics, and mood stabilizers. We further conducted stratified analyses by primary provider, prescription duration, excessive dosing, high-toxicity prescription, and overlapping prescription. All data were analyzed using R version 3.0.3 with the R package survival.

Psychopharmacology

Results Study participants We identified 1,197,466 patients who were 12–74 years old, had been enrolled in their health insurance at least 6 months before the study period, and received any

Table 1

treatment during the study period (Fig. 1). Of the cohort, we identified 351 drug overdose cases (62 % women) during the study period (Table 1), with a median age of 37 years (interquartile range, 25–46 years). Unipolar depression was the most prevalent history of mental and behavioral disorders (45 %), followed by schizophrenia (28 %), anxiety disorders (22 %), and

Demographic characteristics

Characteristics

Provider typea Clinic Hospital Admission status Outpatient Admission to non-ICU Admission to ICU Sex Male Female Age group, years 12–19 20–29 30–39 40–49 ≥50 History of mental and behavioral disorders (ICD-10 code) Any (F04–F99) Unipolar depression (F32–F33, F34.1) Bipolar depression (F30–F31) Schizophrenia (F20–F29) Alcohol use disorders (F10) Drug use disorders (F11–F16, F18–F19) Anxiety disorders (F40–F44) Somatoform disorders (F45) Other neurotic disorders (F48) Eating disorders (F50) Personality disorders (F60–F69) Intellectual disability (F70–F79) Autism spectrum disorders (F84.0–F84.1, F84.5) Others Charlson Comorbidity Index 0 ≥1

Cases

General controls

High-risk controls

(N=351)

(N=1755)

(N=1755)

Number

Percent

Number

Percent

Number

Percent

59 292

16.8 83.2

1430 325

81.5 18.5

1311 444

74.7 25.3

193 101 57

55.0 28.8 16.2

1745 10 0

99.4 0.6 0.0

1734 18 3

98.8 1.0 0.2

132 219

37.6 62.4

660 1095

37.6 62.4

660 1095

37.6 62.4

29 85 83

8.3 24.2 23.6

145 425 415

8.3 24.2 23.6

145 425 415

8.3 24.2 23.6

85 69

24.2 19.7

425 345

24.2 19.7

425 345

24.2 19.7

232 157 59 98 13 8 78 32 42 9 17 2 3 29

66.1 44.7 16.8 27.9 3.7 2.3 22.2 9.1 12.0 2.6 4.8 0.6 0.9 8.3

56 25 3 11 0 0 15 7 13 1 0 0 1 4

3.2 1.4 0.2 0.6 0.0 0.0 0.9 0.4 0.7 0.1 0.0 0.0 0.1 0.2

1430 1305 128 205 13 1 431 166 232 25 14 7 15 88

81.5 74.4 7.3 11.7 0.7 0.1 24.6 9.5 13.2 1.4 0.8 0.4 0.9 5.0

219 132

62.4 37.6

1644 111

93.7 6.3

1245 510

70.9 29.1

ICU intensive care unit a

A clinic is a medical institution having 19 or fewer inpatient beds, where as a hospital is defined as a medical institution with 20 or more inpatient beds

Psychopharmacology Fig. 1 Flow diagram of included participants

1,720,344 recipients aged 0-74 years as of October 2012

1,538,423 eligible recipients who were 12-74 years old and had been enrolled in their health insurance before March 31, 2012

1,197,466 patients with any treatment during the study period (October 1, 2012 to November 30, 2013)

351 overdose cases

1,755 general controls (without overdose and with any treatment)

1,755 high-risk controls (without overdose and with depression)

bipolar depression (17 %). From the cohort, we matched 1755 controls for each of the general and high-risk groups. Cases and both control groups were well matched for sex and age groups (Table 1).

controls, current use of antipsychotics (OR, 5.0; 95 % CI, 3.6–6.8) and mood stabilizers (OR, 3.4; 95 % CI, 2.4–4.8) were both associated with increased odds for overdose, and these associations were attenuated among past users (Table 2).

Psychotropic medications among overdose cases

Stratified analyses of current users

Among the 351 overdose cases, the most frequently used psychotropic medication was sedatives-hypnotics (62 %), followed by antidepressants (44 %), antipsychotics (31 %), and mood stabilizers (20 %) (Table 2). Of psychotropic medication users, 93–96 % were current users. Of 210 cases currently using sedatives-hypnotics, 171 (81 %) received a prescription from psychiatrists, 169 (80 %) were long-term users, 18 (9 %) used barbiturates, 82 (39 %) received excessive doses, and 9 (4 %) had concurrent prescriptions from multiple providers (Table 3).

Table 3 shows the stratified analyses of current use of psychotropic medications. In the comparison of cases and general controls, prescriptions by psychiatrists did not result in significant ORs for overdose. Long-term users of sedatives-hypnotics (OR, 15.1; 95 % CI, 2.9– 77.7) and antidepressants (OR, 9.2; 95 % CI, 1.4–59.2) had greater odds for overdose when compared with new users. No significant associations were observed between cases and general controls with respect to excessive dosing, high-toxicity prescription, and overlapping prescription. In the comparison of cases and high-risk controls (Table 3), the ORs for overdose were higher among any users prescribed by psychiatrists than among those prescribed by non-psychiatrists, among excessive sedative-hypnotic users than among usual dosage users (OR, 4.3; 95 % CI, 3.0–6.1), among barbiturate users than among BZD/Z-drug only users (OR, 4.5; 95 % CI, 2.3–8.7), among TCA users than among SSRI/SNRI only users (OR, 1.8; 95 % CI, 1.2–2.8), and among sedativehypnotic users with multiple providers than among those with a single provider (OR, 4.4; 95 % CI, 1.7–11.0). Of cases with a potentially questionable prescription, 89 % of excessive sedative-hypnotic and 72 % of barbiturate prescriptions were from psychiatrists (Table 4). Of nine doctor-shopping users, 67 % received a prescription both from psychiatrists and nonpsychiatrists.

Current and past exposures to psychotropic medications In the comparison of cases and general controls, current (OR, 21.5; 95 % CI, 9.7–47.8) and past (OR, 46.5; 95 % CI, 3.8–568.6) use of sedatives-hypnotics were both associated with increased odds for overdose, although the confidence interval for past use was wide (Table 2). When cases and high-risk controls were compared in a similar manner, only current use of sedatives-hypnotics (OR, 2.6; 95 % CI, 1.9–3.5) was associated with an increased odds for overdose. Compared with general controls, current use of antidepressants (OR, 11.1; 95 % CI, 4.6–26.9) and antipsychotics (OR, 12.6; 95 % CI, 3.5–45.8) were both associated with increased odds for overdose (Table 2). Compared with high-risk

41.0 2.8 56.1 29.1 2.3 68.7 18.5 1.1 80.3

102 8 241

65 4 282

59.8 2.6 37.6

144 10 197

210 9 132

0 0 1755

3 0 1752

12 0 1743

23 1 1731

a

0.0 0.0 100.0

0.2 0.0 99.8

0.7 0.0 99.3

1.3 0.1 98.6

12.6 (3.5–45.8)* – ref – – ref

– – ref

11.1 (4.6–26.9)* – ref

21.5 (9.7–47.8)* 46.5 (3.8–568.6)* ref

AOR (95 % CI)a

183.3 (58.2–577.2)* – ref

94.8 (46.6–193.0)* – ref

114.3 (58.6–222.9)* 351.0 (32.2–3821.6)* ref

OR (95 % CI)

Adjusted for sex, age groups, history of mental and behavioral disorders, and history of comorbid conditions

*p