BRIEF REPORT Accidental Poisoning Mortality Among Patients in the Department of Veterans Affairs Health System Amy S.B. Bohnert, PhD,*w Mark A. Ilgen, PhD,*w Sandro Galea, MD, PhD,z John F. McCarthy, PhD,*w and Frederic C. Blow, PhD*w (Med Care 2011;49: 393-396) Background: Accidental poisoning mortality is an increasingly important concern, particularly for health systems, which distribute potentially poisoning medications and treat substance use disorders. Objectives: To describe the rate of accidental poisoning mortality in the Veterans Health Administration (VHA) during fiscal year 2005, assess differences with rates observed in the general US population, and describe the frequency with which specific drugs and medications were mentioned on the death records of accidental poisoning decedents. Research Design: Cohort study. Subjects: All 5,567,621 individuals aged 18+ who received VHA inpatient or outpatient services in fiscal year 2004 (October 1, 2003 to September 30, 2004) or fiscal year 2005 and were alive at the start of fiscal year 2005. Measures: The National Death Index indicated vital status and cause of death, the National Patient Care Database indicated who used VHA services and consequently was in the study cohort, and the Web-based Injury Statistics Query and Reporting System indicated poisoning mortality rates in the general US population. Results: The crude rate of accidental poisoning mortality in the VHA for fiscal year 2005 was 19.85 deaths per 100,000 personyears. After accounting for gender and age distribution, VHA patients had nearly twice the rate of fatal accidental poisoning compared with adults in the general US population (standardized mortality ratio = 1.96; 95% confidence interval: 1.83, 2.08). Opioid medications and cocaine were frequently mentioned as the agents causing poisoning on death records. Conclusions: The present work indicates that a substantial need exists for interventions to reduce the risk of accidental poisoning among VHA patients. Key Words: accidental poisoning, mortality, veterans From the *Department of Veterans Affairs, HSR&D COE and SMITREC; wDepartment of Psychiatry, University of Michigan Medical School, Ann Arbor, MI; and zDepartment of Epidemiology, Columbia University, New York, NY. Supported by funding from VHA's Office of Mental Health Services; VA Health Services Research and Development (HSR&D; grant number CDA09-204); and the VHA's HSR&D and Office of Academic Affiliations. Reprints: Amy S.B. Bohnert, PhD, University of Michigan, 4250 Plymouth Road, Box 5765, Ann Arbor, MI 48104. E-mail: amybohne@med.umich.edu. Copyright r 2011 by Lippincott Williams & Wilkins ISSN: 0025-7079/11/4904-0393 Medical Care  Volume 49, Number 4, April 2011 U nintentional poisoning is the second most common cause of accidental death among adults in the United States (US).1 Unintentional poisoning is an increasingly important public health concern; between 1990 and 2004, the age-adjusted rate of accidental poisoning mortality in the United States increased 62.5%.1 This increase appears to be largely accounted for by sharp increases in the rate of prescription medication poisonings.2,3 Monitoring accidental poisoning mortality is highly relevant to health care systems. Over 95% of poisoning deaths are due to medications, alcohol, and illegal drugs.4 Health care systems both distribute medications that can lead to poisoning (eg, opioids), and treat individuals with drug and alcohol use problems. However, there are limited data available on the surveillance of accidental poisoning within health systems. Examining accidental poisoning mortality is particularly important for the Veterans Health Administration (VHA). Research from the Vietnam Experience Study5 and the Agent Orange Registry6 suggests that veterans may be at greater risk of fatal poisoning than nonveterans. Additionally, the VHA has recently implemented a national suicide prevention program,7 which involves increased screening, referral, and treatment for suicide risk. Because some suicides by poisoning are misclassified as accidental poisonings,8 examining rates of fatal poisoning is relevant to VHA suicide prevention efforts. To date, however, no studies have examined system-wide rates of accidental poisoning deaths among VHA health system users. This study describes the rate of accidental poisoning mortality in the VHA patient population during fiscal year 2005, assesses differences with rates observed in the general US population, and describes the drugs and medications responsible for accidental poisoning deaths in the patient population. MATERIALS AND METHODS Sample and Data Sources The study cohort included all individuals who received VHA inpatient or outpatient services in fiscal year 2004 (October 1, 2003 to September 30, 2004) or fiscal year 2005, were alive and age 18+ at the start of fiscal year 2005, and who resided in the United States (4749 individuals were excluded because they resided in Manila). All study protocols were approved by the local Institutional Review Board. www.lww-medicalcare.com | 393 Bohnert et al Data were obtained from the following 3 sources: (a) the National Death Index (NDI), (b) the VHA's National Patient Care Database, and (c) the Centers for Disease Control and Prevention's (CDC) Web-based Injury Statistics Query and Reporting System (WISQARS). Data sources and analyses generally follow those described in more detail by McCarthy et al9 The National Patient Care Database was used to identify gender and age; patient race/ethnicity was not included due to incomplete data. NDI and WISQARS data are obtained from the same data collection process and are both maintained by the CDC; the NDI contains detailed information from individual death records and is provided to researchers, while WISQARS is on online reporting system of basic mortality statistics. For both, the CDC compiles data from state viral statistics offices. The NDI was used to determine vital status and cause of death for the VHA study sample, and has the greatest sensitivity in determining vital status among population-level data sources.10 Past studies indicated that the sensitivity of the NDI data for vital status has ranged between 87% and 98%.10 NDI searches were conducted for all those who received VHA services in fiscal year 2004 and/or fiscal year 2005 (October 1, 2003 to September 30, 2005) but did not have subsequent VHA utilization through June 2006; individuals with VHA use after the end of fiscal year 2005 were assumed to have survived the study period and they were not included in NDI searches. Established procedures identified "true" matches when the search resulted in multiple potential matches.11 For the present study, we used WISQARS data4 to ascertain the rate of fatal accidental poisoning in the general US population in the year 2005 by age-sex group. We did so by creating a report of the number of deaths and crude rate of accidental poisoning deaths using the report generator available for fatal injuries on the WISQARS website. We repeated this process for subgroups defined by age and sex. Fatal accidental poisoning was defined using the International Classification of Diseases-10 codes X40 to X49.12 Deaths that result in these codes are those that are the Medical Care  Volume 49, Number 4, April 2011 result of poisoning by legal and illegal drugs (X40 to X44), alcohol (X45), organic solvents (X46), gases (X47), pesticides (X48), and other chemicals (X49) and that are ruled accidental in intent. We further examined T-codes when available from the death records of individuals who died of accidental poisoning in the VHA study sample. T-codes are additional indicators used by medical examiners to describe drugs and medications that contributed to poisoning death. The specific T-codes used in this study are described in Table 1. Analytic Strategy To calculate rates of accidental poisoning mortality, we used the approach described by McCarthy et al9 to calculate the person-years contributed by each individual in the VHA patient population for fiscal year 2005. For individuals with VHA use in fiscal year 2004, risk time began on the first day of FY05 (October 1, 2004). For those without VHA use in fiscal year 2004 but with use in fiscal year 2005, risk time began on the day of their first use of VHA services in fiscal year 2005. Risk time ended on the day of death (from any cause) or the end of fiscal year 2005, whichever came first.9 We calculated standardized mortality ratios (SMRs) and 95% confidence intervals (CIs) using the Indirect Standardization Method.13 This method involves calculating the expected number of deaths for the sample by summing the number of deaths that would be expected in each age-sex stratum based on the number of person-years in the sample and the rate in the reference population for that stratum. The SMR is calculated as a ratio of the number of observed deaths to the expected number of deaths, and the 95% CI is the SMR ? 1.96 ? [SMR/O[observed number of deaths)]. RESULTS The study cohort included 5,567,621 VHA patients, among whom 1013 died of accidental poisoning in fiscal year 2005 (16 individuals could not be included because they were missing data on sex). The crude accidental poisoning TABLE 1. Drugs and Medication Causing Poisoning Mentioned via T-Codes on the Death Records of VHA Patients Who Died by Accidental Poisoning in Fiscal Year 2005* Drug/Medication Typew Nonopioid analgesics, antipyretics, and antirheumatics Heroin Other opioids Methadone Other synthetic narcotics Cocaine Other and unspecified narcotics Barbiturates Benzodiazepines Antiparkinsonism drugs Antidepressants Antipsychotics and neuroleptics Psychostimulants with abuse potential T-Code(s) n (%) T39.0-T39.9 T40.1 T40.2 T40.3 T40.4 T40.5 T40.6 T42.3 T42.4 T42.8 T43.0-T43.2 T43.3-T43.5 T43.6 29 68 201 150 47 253 92 4 82 5 88 21 48 (2.7) (6.3) (18.5) (13.8) (4.3) (23.3) (8.5) (0.4) (7.5) (0.5) (8.1) (1.9) (4.4) *A total of 732 cases had a total of 1088 T-code mentions. wNonmutually exclusive categories. VHA indicates veterans health administration. 394 | www.lww-medicalcare.com r 2011 Lippincott Williams & Wilkins Medical Care  Volume 49, Number 4, April 2011 Accidental Poisoning Deaths Among Veterans TABLE 2. Rate and Standardized Mortality Ratio of Accidental Poisoning Mortality Among VHA Patients in Fiscal Year 2005 Compared to the General US Population in 2005, Adults Aged 18 and Older* US VHA Accidental Poisonings Males 18-29 y 30-64 y 65+ y Females 18-29 y 30-64 y 65+ y Total Crude Rate per 100,000 Person-Years Accidental Poisonings Crude Rate per 100,000 Person-Years Standardized Mortality Ratiow 95% Confidence Interval 15,679 3390 11,826 463 7647 1024 6155 468 23,326 14.51 13.45 17.53 3.01 6.68 4.28 8.91 2.19 10.49 960 28 841 91 53 1 50 2 1013 20.62 21.86 36.81 4.06 11.82 1.70 15.34 3.14 19.85 1.98 1.62 2.10 1.35 1.61 N/A 1.72 N/A 1.96 1.85, 2.10 1.02, 2.23 1.96, 2.24 1.07, 1.63 1.18, 2.04 N/A 1.24, 2.20 N/A 1.83, 2.08 *VHA rates were calculated for fiscal year 2005, while US rates were for calendar year 2005. wSMR and 95% confidence interval calculated using indirect standardization. N/A indicates nonapplicable; VHA, veterans health administration; US, United States. mortality rate in fiscal year 2005 was 19.85 per 100,000 person-years for VHA users (Table 2), which was significantly higher than the crude rate of 10.49 for the US population as a whole (P<0.0001 in a 2-sided exact significance test). Similar relationships of demographic characteristics with accidental poisoning mortality rates were observed in the VHA patient population and the US population. The rate of accidental poisoning mortality was higher for men than women, higher for individuals aged between 30 and 64 years as compared with individuals aged 18 to 29 or 65+, and higher for individuals 18 to 29 than individuals aged 65+ in both the VHA population and the US population (exact 2sided P<0.0001 for all sex- and age-group comparisons within each population). Compared with adults in the general US population, VHA patients had nearly twice the risk of fatal accidental poisoning after accounting for gender and age distribution (SMR = 1.96; 95% CI: 1.83, 2.08; Table 2). SMRs could not be calculated for women aged between 18 and 29 and 65+ due to low numbers in these groups. For all other strata, the SMR and 95% CI indicated a significantly higher risk for VHA users than the same age-sex group in the US population (Table 2). Among the 1013 VHA patients whose underlying cause of death was accidental poisoning, 732 had a total of 1088 Tcode mentions available in their death records (Table 1). Opioid medications (including methadone) made up 32.3% of all T-codes reported. Cocaine was also common, representing 23.3% of all mentions. Other types of medications and heroin each represented less than 10% of all mentions in this sample. Additionally, 12 individuals died of poisoning due to alcohol (underlying cause of death code X45). DISCUSSION The present study is the first to our knowledge to examine the rate of accidental poisoning mortality in a large health system. We found a greater rate of accidental poisoning deaths among veterans who use VHA services compared with persons in the general US population. In r 2011 Lippincott Williams & Wilkins comparison to the rate of suicide among VHA patients, which was 39.80 per 100,000 person-years,9 the rate of accidental poisoning observed in the present study (19.83 per 100,000 person-years) was lower. However, after accounting for age and gender, the SMR was slightly higher than that for suicide (1.95 compared with 1.66 for suicide9). These findings indicate that, although patients treated at the VHA have a greater risk of suicide than accidental poisoning death, their risk of accidental poisoning death relative to the general population is larger than that of suicide. Despite the concern by Congress and within the VHA regarding risk for suicide among VHA patients (described in9), the present results suggest that there is actually greater excess compared with the general population in accidental poisoning mortality than in suicide mortality among VHA patients. We also found that opioid medications were frequently the substance responsible for accidental poisoning death among VHA patients. There are numerous possible causes for the observed elevated rate of accidental poisoning mortality among VHA patients, including psychosocial and demographic differences between VHA patients and the general population, greater access to medications through health system use, increased prevalence of psychiatric and substance use disorders (which increases risk of drug and medication misuse), and the effects of combat exposure. Determining the specific underlying causes of the elevated rate among VHA patients is beyond the scope of the present study. Nonetheless, the present work highlights the need for interventions to reduce the risk of accidental poisoning among VHA patients. The present study has several limitations. First, classification of cause of death was based on medical examiners' records. This source of data may contain some bias. In particular, differentiating between suicidal and accidental poisonings is difficult.8 There is likely variation in amount of information available at the time of death across cases and well as variation among medical examiners in method to rule overdose intent. Furthermore, interviews with nonfatal poisoning victims indicate that intention does www.lww-medicalcare.com | 395 Bohnert et al not always clearly fit into categories of suicidal and accidental.14 Veteran status is not available through NDI or WISQARS data, which would greatly aid research on differences in mortality risk for veterans and between veterans who do and do not use VHA services. Because veterans gain eligibility to use VHA services by financial need or having a physical or mental condition related to military service, VHA users are likely to be at a higher risk of death from drug- and alcohol-related conditions than veterans who do not use VHA services. Also, these data only cover deaths that occurred within the United States and do not reflect deaths that occurred overseas, and only represent 1 year and rates in sex-age groups that are smaller may fluctuate due to random variation over time. Despite these limitations, the present study highlights the importance of studying poisoning among veterans. Preventing accidental poisoning is important to the mission of health care organizations such as the VHA that prescribe medications and treat potentially high-risk individuals. In recent years, the VHA has greatly increased mental health-, substance use-, and suicide-related services,7 and increased monitoring of accidental poisoning risk through these services could potentially impact the rate of poisoning mortality in the future. Future work should examine the relationship between prescribing patterns of specific medications within the VHA and overdoses due to those medications among patients. The elaboration of International Classification of Diseases-10-CM codes will likely improve future surveillance of nonfatal poisoning events due to medications and drugs as well. Such work may suggest the necessity of interventions and policies such as increased patient education, limiting the maximum size fill of particular medications, switching to medication packaging that discourages taking more pills than intended, or creating automatic alerts within the electronic medical record system when particular combinations of medications are prescribed. The present work will serve as a baseline for understanding 396 | www.lww-medicalcare.com Medical Care  Volume 49, Number 4, April 2011 how future VHA-wide interventions may impact the accidental poisoning death. REFERENCES 1. Centers for Disease Control and Prevention. Unintentional poisoning deaths-United States, 1999-2004. Morb Mortal Wkly Rep. 2007;56:93-96. 2. Paulozzi LJ, Budnitz DS, Xi Y. Increasing deaths from opioid analgesics in the United States. Pharmacoepidemiol Drug Saf. 2006;15: 618-627. 3. Caravati EM, Grey T, Nangle B, et al. Increase in poisoning deaths caused by non-illicit drugs- Utah 1991-2003. Morb Mortal Wkly Rep. 2005;54:33-36. 4. Centers for Disease Control and Prevention. Web-based Injury Statistics Query and Reporting System (WISQARS). Available at: http:// www.cdc.gov/ncipc/wisqars. (Accessed June 29, 2010). 5. Centers for Disease Control and Prevention. Post-service mortality among Vietnam veterans. The Centers for Disease Control Vietnam Experience Study. JAMA. 1987;257:790-795. 6. Bullman TA, Kang HK. Posttraumatic stress disorder and the risk of traumatic deaths among Vietnam veterans. J Nerv Ment Dis. 1994;182: 604-610. 7. Department of Veterans Affairs, Office of Inspector General. Healthcare Inspection: Implementing VHA's Mental Health Strategic Plan Initiatives for Suicide Prevention. Washington, DC: VA Office of Inspector General; 2007. Report 06-03706-126. 8. Phillips DP, Ruth TE. Adequacy of official suicide statistics for scientific research and public policy. Suicide Life Threat Behav. 1993; 23:307-319. 9. McCarthy JF, Valenstein M, Kim HM, et al. Suicide mortality among patients receiving care in the veterans health administration health system. Am J Epidemiol. 2009;169:1033-1038. 10. Cowper DC, Kubal JD, Maynard C, et al. A primer and comparative review of major US mortality databases. Ann Epidemiol. 2002;12: 462-468. 11. Sohn MW, Arnold N, Maynard C, et al. Accuracy and completeness of mortality data in the Department of Veterans Affairs. Popul Health Metr. 2006;4:2. 12. World Health Organization. International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10). 2nd ed. Geneva, Switzerland: World Health Organization; 2004. 13. Osborn CE. Essentials of Statistics in Health Information Technology. Boston, MA: Jones and Bartlett Publishers; 2007. 14. Bohnert AS, Roeder K, Ilgen MA. Unintentional overdose and suicide among substance users: a review of overlap and risk factors. Drug Alcohol Depend. 2010;110:183-192. r 2011 Lippincott Williams & Wilkins