International Journal of Drug Policy - 2014

Volume 25 Issue 3 May 2014

Issue link:

Contents of this Issue


Page 69 of 153

570 A. Siegler et al. / International Journal of Drug Policy 25 (2014) 569–574 strategies such as naloxone distribution to laypersons and first responders (Centers for Disease Control and Prevention, 2012). In addition, Good Samaritan laws aim to increase calls for emergency response to drug overdose by protecting witnesses from arrest for drug possession when calling emergency services. The identification of opioid overdose settings can aid in the strategic enhancement of opioid overdose prevention efforts. Previ- ous studies have explored drug overdose settings (Bernstein et al., 2007; Bohnert, Tracy & Galea, 2009; Davidson et al., 2003; Cerdá et al., 2013), and have reported varying findings, ranging from 28% (Davidson et al., 2003) to 83% overdosing in private homes (Cerdá et al., 2013). However, these studies did not evaluate if demo- graphics or drug use characteristics were associated with overdose settings. Drug use is a socially and culturally bound phenomenon, making it plausible that an individual's demographic characteris- tics and the drugs used may in fact affect the setting and risk level of overdosing. The discovery of common opioid overdose settings, as well as any differences by opioid type and demographic charac- teristics, can help inform the development and implementation of targeted and effective overdose response programs. Methods Sample The sample included all unintentional opioid poisoning deaths among NYC residents aged 15–84 from January 1, 2005 to December 31, 2010, using linked death certificates and medical examiner files. Unintentional drug poisoning death was defined as a death for which the death certificate recorded (i) the manner of death as "accidental," and (ii) the codes for underlying causes of death as "poisoning by a psychoactive substance (excluding alco- hol or tobacco)" (ICD-10 codes X40-X44) or a "mental or behavioral disorder due to a psychoactive substance" (ICD-10 codes F11-16, F18-19). The sample was limited to decedents with toxicology results positive for one or more opioids. Methadone was recorded for toxicologies including methadone or methadone metabolite. Heroin was recorded if any of the following were present in toxicology: morphine, 6-monoacetylmorphine (6-MAM), diacetyl- morphine, acetylcodeine, morphine and codeine, 6-MAM and codeine, and diacetylmorphine and codeine. Opioid analgesics were recorded if any of the following were present in toxico- logy: codeine (without heroin), alfentanyl, fentanyl, carfentanyl, sufentanil, hydrocodone, hydromorphone, meperidine, oxycodone, oxymorphone, papavirine, pentazocine, propoxyphene, thebaine, tramadol, or phenacetin. All manners of death other than unintentional, i.e. inten- tional, undetermined, or homicide were not included. The sample excluded non-NYC residents and decedents whose borough of resi- dence was unknown or missing. The final analytic sample excluded decedents with missing overdose setting. Variables and definitions Demographic variables included gender, race/ethnicity, age, education, borough of residence, and neighborhood poverty. Race/ethnicity included non-Hispanic white, non-Hispanic black, Hispanic, and Non-Hispanic other, which collapsed other race/ethnicities and missing. Age was categorized as 15–24, 25–34, 35–44, 45–54, 55–64, and 65–84 years. Education level was deter- mined from the death certificate and defined as less than high school, completed high school or General Education Development exam, some college, and college or more. Neighborhood poverty was defined from United Hospital Fund area (UHF, an aggregation of zip codes averaging nearly 200,000 population) of residence, as the percent of residents with incomes below 100% of the Federal Poverty Level per American Community Survey and Census 2000. Neighborhood poverty was categorized into four groups: low (<10% of residents below poverty), medium (10% to <20% below poverty), high (20% to <30% below poverty), and very high (≥30% below poverty). Settings of overdose were abstracted from medical examiner files as a text field and categorized into five groups. The decedent's home or others' homes were collapsed into the category home. Home was defined as a non-staffed residential address, including independent apartments, houses, and public housing. Staffed resi- dences such as homeless shelters and supportive housing facilities were not included in the 'home' category. Other locations were classified as non-home, and stratified into four sub-groups: institu- tional residences, public indoors, outdoors, and other. Institutional residences included homeless shelters, single room occupancies, supportive housing, nursing homes, and assisted living facilities. Public indoor settings included bars, restaurants, hotels, public bathrooms, offices, building lobbies, elevators, and stairways. Out- doors included parks, streets, roofs, cars, buses, and subways. The setting category other included prison, drug treatment programs, and hospitals. Drugs and drug metabolites were abstracted from toxicology reports of medical examiner files, and included alcohol, benzodi- azepines, cocaine, methadone, heroin, and opioid analgesics. Given the large number of patients in methadone maintenance in NYC, methadone is reported separately from other opioid analgesics. Drugs were not mutually exclusive; a decedent's toxicology could be positive for more than one drug. In addition to single drug classifications, four drug combinations were analyzed: heroin and opioid analgesics, heroin and methadone, methadone and opioid analgesics, and opioid analgesics and benzodiazepines. Data analysis A descriptive analysis of demographics and drug types was con- ducted for the total sample and for subsamples by setting. All drugs and drug combinations were analyzed as dummy variables. Age- adjusted rates were calculated for demographics and drug types, while age-standardized rates were calculated for each age group. Rates were age-adjusted to Census 2000. To obtain aggregate rates across six years, age-adjusted rates were averaged. Logistic regression was used to compare characteristics of those who overdosed at home to those who overdosed elsewhere. Inter- actions were tested between demographics and between each drug. Multivariate logistic regression using backward selection was con- ducted to determine predictors of overdose settings. All variables significant at 0.05 were kept in the final model. Adjusted odds ratios and corresponding 95% confidence intervals were computed, and model fit was evaluated. All analyses were conducted using SAS 9.2 (Cary, NC). Results From 2005 to 2010, there were 4083 unintentional drug over- dose deaths in NYC. The greatest number of unintentional drug poisoning deaths occurred in 2006 (n = 838) and decreased each subsequent year to a low of 541 in 2010 (data not presented). A total of 1434 decedents did not meet inclusion criteria (criteria were not mutually exclusive and could overlap): 1126 decedents did not test positive for any opioids, 179 were not NYC residents, 127 were missing borough of residence, and 53 were missing set- ting of overdose. The final analytic sample was comprised of 2649 decedents. Proportions of overdoses by setting did not change sig- nificantly over the study period, so all years were collapsed. There

Articles in this issue

view archives of International Journal of Drug Policy - 2014 - Volume 25 Issue 3 May 2014