Drug overdoses are a major public health issue in the United States (Paulozzi, 2012). Prescription opioid pain relievers (OPR) alone or in combination with benzodiazepines or other drugs account for nearly half of all drug overdose deaths (Jones, Mack, & Paulozzi, 2013). The percentage increase in prescription OPR deaths since 1999 is startling—a 265% increase among men and a 415% increase among women (Mack, Jones, & Paulozzi, 2013).

The location of drug use, the presence of others at the scene, and the actions of bystanders (e.g., calling for help) could determine whether an overdose becomes a fatality (Darke & Zador, 1996). Certain location characteristics, such as homelessness (Fischer et al., 2004) or residential isolation, have been extensively characterized among cases of drug overdoses (Cooper, Friedman, Tempalski, & Friedman, 2007), although primarily among injection heroin users in poor urban areas. Among specific inner-city populations, the following characteristics predict fatal overdoses: a lack of permanent housing (Fischer et al., 2004; Jenkins et al., 2011), housing in poor physical condition (Hembree et al., 2005), residence in a low-income hotel (Davidson et al., 2003), using heroin while alone, (Davidson et al., 2003), and using drugs in public (Dietze, Jolley, Fry, Bammer, & Moore, 2006) or in abandoned buildings (Bohnert, Tracy, & Galea, 2009).

There are differences in the population at highest risk of a prescription OPR overdose compared with those at highest risk of an illicit drug overdose (Mack et al., 2013; Paulozzi, 2012). Further, the location of a fatal overdose varies by the type of drug involved. Illicit drug deaths, such as those involving heroin and cocaine, occurred more often in nonresidential, nonmedical locations. People dying of illicit drug overdoses appeared to be more likely than those dying of prescription OPR overdoses to die in a medical facility (Paulozzi, 2012). A study of overdose deaths in Connecticut found that 8.2% of OPR deaths occurred in public places compared to 19.7% of heroin-related deaths (Green, Grau, Carver, Kinzly, & Heimer, 2011). Describing situations in which drug users survive overdoses is important; understanding why others do not is critical. We know little about whether some determinants of survival (e.g., being alone, being inexperienced, concealing use from bystanders) differ between prescription OPR and illicit drug users. A closer examination of the location of overdose deaths by type of drug (prescription OPR vs. illicit) for various risk groups may reveal important OPR-specific conditions that can guide overdose prevention activities.

During a 3-year period (2008–2010), 14,091 prescription OPR or illicit drug-related deaths occurred in 12 states at home (55.7%), at emergency room or health care facilities (27.8%), or in other or missing (16.5%) locations. The majority of the deaths (78.9%) were prescription OPR drug-related, either without illicit drugs (63.4%), or in combination with illicit drugs (15.4%).

There were similar patterns for location of death among white and Hispanic decedents in all three drug categories (Table 1). The majority of white OPR-related deaths occurred at home (62.0%); compared to 64.0% of Hispanic decedents; among white illicit drug drug-related decedents, the largest percentage was at home (45.2%); similar to Hispanic decedents (43.6%). Black OPR related deaths were equally distributed between a home location (44.0%) and an ER (44.3%); the largest percentage for illicit drug-related death was an ER location (46.3%). Among those aged 20 years or older, more than half of the prescription OPR related deaths occurred at home, and although the percentages are lower, the highest category for illicit and combination drug-related deaths was also at home. Among those younger than 20 years, the highest percentage category for OPR-related deaths was home (48.9%); and ER for illicit (36.1%) and combination drug-related deaths (44.2%). The highest percentage category was at home for each manner of death and drug type.

Overall, 61.3% of the prescription OPR-related deaths occurred at home, 25.3% in an ER or health care facility, and 13.4% in another location (Table 2). More than half of OPR-related deaths occurred at home in every state, except MD (49.0%). Although in most states, the highest percentage of illicit drug-related deaths occurred at home, the percentages were lower in each state for illicit drug-related deaths compared with prescription OPR-related deaths. Maryland and Massachusetts had a higher percentage of illicit drug-related deaths occurring at an ER (41.5%) than at home (MD = 35.3%; MA = 41.3%), and Oregon's highest category was other or missing (42.3%; 4.4% of which was a public place and 37.8% was other). For deaths where both prescription OPR and illicit drugs were involved, the highest category for location of death was at home, except in Maryland (ER was highest, 44.4%) and Vermont (ER; 40.0%; other 40%).

This study characterized drug-related deaths by state, age, race or ethnicity, manner of death, and location of death to help guide overdose prevention efforts. Among whites and Hispanics, for all three drug categories, the largest percentage of decedents died at home. The overall percent-age of deaths occurring in public or unknown places was low. In every state except Utah, a higher percentage of illicit drug-related deaths than prescription OPR-related deaths occurred in an ER or health care facility. In all but one state (Maryland), the majority (>50%) of prescription OPR-related deaths happened at home. Black decedents (for prescription OPR and illicit drug categories) and those aged younger than 20 years (illicit and combination drugs) were also more likely to die within an ER or other health care facility.

The pharmacokinetics of opioids vary by drug and can vary even within a drug depending on route of administration (e.g., injection vs. oral administration). Death may occur rapidly depending upon potency and route of administration. Because users of illicit drugs often inject where they buy drugs, and signs of distress may appear rapidly, other users or strangers may notice that a victim is in distress. These other users, however, often do not call 911 for help, perhaps fearing police intervention (Dietze et al., 2006; Tobin, Davey, & Latkin, 2005; Wagner et al., 2015), but instead they may slap or pinch the victim, give them coffee, try to make them walk, inject them with a different substance (e.g., cocaine, salt, water), administer naloxone, or attempt CPR (Baca & Grant, 2005; Bohnert, Tracy, & Galea, 2012; Wagner et al., 2015). Researchers studying inner city illicit drug use have suggested that efforts to prevent overdose (i.e., good Samaritan laws or increased access to naloxone) could focus on those persons using drugs in public or in abandoned buildings (Bohnert et al., 2009). Prescription OPRs, however, are most often taken orally, absorbed slowly and variably, and with toxicity escalating gradually, perhaps after potential bystanders depart (Baca & Grant, 2005). In private homes, overdose prevention may fall to different groups of bystanders, friends, relatives, or caregivers, or their responses may differ. For instance, in private locations, which would include homes, bystanders may not call for emergency services after giving naloxone (Lankenau et al., 2013). Given the short duration of action of naloxone, a return of opioid intoxication from an OPR with a longer duration of action, might then lead to fatal rebound respiratory distress.

The percentage of prescription OPR-related deaths occurring at home generally increased with age, rising to more than two-thirds among those aged 50 years or older. Age and residential status are closely related, perhaps decreasing the chance of overdose witnesses as age increases. In the United States in 2013, 15% of those younger than 20 years and 21.5% of those aged 45–54 years or older lived alone (U.S. Census Bureau, 2013). As people age, they are more likely to live alone and spend more time alone, which may increase their chances of having an unwitnessed drug overdose.

This study has several limitations. We were unable to distinguish between decedents who had a prescription for an OPR and those who acquired them in other ways. Further, the data do not indicate the location where the drug was taken, the amount consumed, whether bystanders were present in the home, or whether bystanders (e.g., family or housemates) knew the drug(s) were taken. These five pieces of information would greatly enhance the ability to understand the circumstances of the fatal event. Given increasing public awareness about naloxone and increased access to it in the United States, it would also be useful to know whether anyone in a household had access to it.

In addition, assigning responsibility to one drug in multidrug deaths is a challenge. Non-opioid prescription drugs were also reported by ME/Cs (e.g., alprazolam, diazepam, benzodiazepines), and they were sometimes taken in combination with prescription opioids, but not in sufficient quantity to present as a separate analytic category. Thus, any of the cases coded into the three drug categories used in this analysis (prescription opioids, illicit drugs, prescription opioids, and illicit drugs) could have involved other prescription, illicit, or over the counter drugs. The drugs involved in deaths might not all have been identified and documented. DAWN ME/C does not rely on a statistical sampling of MEs, and findings cannot be considered representative of ME/Cs who did not participate, and results cannot be extrapolated to the entire United States (Substance Abuse and Mental Health Services Administration, 2012). State laws dictate which deaths are subject to ME review, and these laws vary by state. Finally, toxicology testing practices can vary depending on local concerns, funding, and testing technology, which affects the number of deaths determined to be DAWN ME/C cases and the number of deaths attributed to particular drugs. It is likely that our results undercount the number of drug-related deaths.

Consideration of home-based interventions for prescription OPR pain reliever overdoses is a prudent public health strategy. On-site interventions, such as naloxone administration, rescue breathing, or calling 911, are clearly most useful when someone is there to administer them (Clark, Wilder, & Winstanley, 2014; Enteen et al., 2010). Naloxone given by first responders can reverse the effects of the opioid taken with high efficacy (Vilke, Sloane, Smith, & Chan, 2003), but even the administration of CPR by a bystander before an ambulance arrives can decrease hospitalizations among nonfatal cases (Dietze, Cantwell, & Burgess, 2002). SAMHSA's Opioid Overdose Toolkit (http://store.samhsa.gov/shin/content//SMA14-4742/Overdose_Toolkit.pdf) provides individuals and communities with information about preventing opioid-related overdose deaths and the U.S. Department of Health & Human Services offers a federal resource (www.hhs.gov/opioids) of tools and information for families, health care providers, law enforcement, and other stakeholders about prescription drug abuse and heroin use prevention, treatment, and response.

Patients and physicians can also take steps to reduce prescription OPR overdose deaths. Patients need to discuss all medications they are taking with their health care provider, use prescription drugs only as directed, store them in a secure place, and get help for substance abuse problems (National Treatment Referral Helpline 1-800-662-HELP [4357]). Physicians can discuss all pain treatment options, including ones that do not involve prescription drugs, as well as the risks and benefits of taking prescription painkillers. Physicians can follow guidelines for prescribing painkillers responsibly (http://www.cdc.gov/drugoverdose/prescribing/guideline.html), including the following: screening and monitoring for substance abuse and mental health problems; prescribing only the quantity needed on the basis of appropriate pain diagnosis; using patient–provider agreements combined with urine drug tests for people who use prescription painkillers long-term; and avoiding combinations of prescription painkillers and benzodiazepines unless there is a specific medical indication.