5 page literature review

Knowledge of the 911 Good Samaritan Law and 911-calling behavior of overdose witnesses

Andrea Jakubowski , MD, Hillary V. Kunins, MD, MPH, Zina Huxley-Reicher, BA, and Anne Siegler, DrPH

New York City Department of Health and Mental Hygiene, Long Island City, New York, USA

ABSTRACT Background: Overdose deaths tripled between 1999 and 2014. Most fatal overdoses are witnessed, offering an opportunity for bystanders to call 911. However, fear of arrest may prevent them from calling authorities. Many states have passed 911 Good Samaritan laws that protects the 911 caller and overdose victim from prosecution for drug possession. Little is known, however, about whether the law affects 911- calling behavior of overdose witnesses. This study investigated the relationship between knowledge of a 911 Good Samaritan Law (GSL) and 911-calling behavior of study participants trained in opioid overdose rescue. Methods: 351 individuals (N D 351) trained in overdose rescue and educated about the New York State GSL were enrolled in a prospective longitudinal study. Trained researchers conducted baseline and 3-, 6-, and 12-month follow-up surveys with study participants to assess participant knowledge of the GSL and responses to witnessed overdoses. Results: At the 12-month follow-up, participants had witnessed 326 overdoses. In the overdose events where the participant had correct knowledge of the GSL at the time of the event, the odds of a bystander calling 911 were over 3 times greater than when the witness had incorrect knowledge of the GSL (odds ratio [OR] D 3.3, 95% confidence interval [CI]: 1.4–7.5). This association remained significant after adjusting for age, gender, race of the witness, and overdose setting (adjusted OR [AOR] D 3.6, 95% CI: 1.4–9.4). Conclusions: This study shows a clear association between knowledge of the GSL and 911-calling behavior. Legislation that protects overdose responders along with public awareness of the law may be an effective strategy to increase rates of 911-calling in response to overdose events and decrease overdose-related mortality.

KEYWORDS 911-calling; Good Samaritan Law; overdose prevention; naloxone

Introduction

Drug overdose deaths in the United States continue to rise, nearly tripling between 1999 and 2014.1 In 2014, 61% of over- dose deaths involved opioids.1 Most fatal overdoses are wit- nessed, presenting an opportunity for life-saving medical intervention, such as calling 911 for emergency medical ser- vice.2 However, barriers to calling 911 are well documented among users of illicit drugs, with fear of police involvement and arrest of primary concern.2–8 Rates of 911-calling for fatal overdose have been reported to range from 15% to 72%, indi- cating that emergency medical services are underutilized.9,10

In response to the opioid overdose epidemic, 36 states in the United States have passed 911 Good Samaritan laws (GSLs).11

These laws provide legal protection for overdose victims and bystanders who call 911, although vary in their specific criminal protections for drug possession, drug paraphernalia, and parole or probation violation.

Although the intent of GSLs is to increase 911 calls in overdose events, little is known about the effect of these laws on 911-calling behaviors of overdose witnesses. One survey showed that after receiving GSL information, drug users reported they would be more likely to call 911 if they were to witness an overdose,12 but no studies have documented an association between 911-calling behavior and knowledge of GSLs.

Educating individuals about GSLs is a key component of opioid overdose prevention training (OPT) curricula. OPTs are designed to teach people who use drugs and those in their social networks to recognize the signs of an overdose and respond by calling 911 and administering rescue breathing and naloxone, a medication that reverses the effects of opioid overdose.

In this prospective longitudinal study of the impact of OPT on behaviors associated with witnessed overdoses, we sought to determine the relationship between participant knowledge of the New York State GSL and 911-calling in response to over- dose events. We hypothesized that correct knowledge of the GSL would be associated with 911-calling. Based on prior stud- ies and a priori hypotheses, we also examined whether other event characteristics are associated with 911-calling, including overdose event location and participant receipt of public bene- fits or residence in public housing.

Methods

Study design

We conducted a prospective cohort study of trained overdose responders. Study participants were recruited following OPTs at 6 syringe exchange programs in New York City between

CONTACT Andrea Jakubowski, MD [email protected] Montefiore Medical Group Comprehensive Health Care Center, 305 E 161 Street, Bronx, NY 10451, USA. © 2018 Taylor & Francis Group, LLC

https://doi.org/10.1080/08897077.2017.1387213

SUBSTANCE ABUSE, 2018 VOL. 39, NO. 2, 233–238

ORIGINAL RESEARCH

June and September of 2013. OPT recipients were recruited immediately following completion of training. Exclusion crite- ria were age less than 18 years, inability to complete an oral sur- vey in Spanish or English, residence outside of the city in which the study was conducted, and no contact information for fol- low-up interviews. The study was approved by the New York City Department of Health and Mental Hygiene institutional review board. Participants provided written informed consent.

Trained researchers conducted in-person or phone surveys immediately after OPT (baseline) and at 3, 6, and 12 months according to standardized protocols using a closed-ended ques- tionnaire. We attempted to reach participants in person, by mail, phone, and text message up to 3 months after the 3- month follow-up date, up to 6 months after the 6-month fol- low-up date, and up to 3 months after the 12-month date of the last participant enrolled for the 12-month follow-up. For sur- veys administered after 12 months, outcomes were included only if they occurred within 2 weeks of the participant’s 12- month follow-up date.

Measures

Characteristics of study participants All participant characteristics were self-reported at baseline. Demographic characteristics included race/ethnicity (non- Hispanic white, non-Hispanic black, Hispanic, and other), gender (male or female, with transgender categorized according to the participant’s self-identified gender), and education (less than high school, grade 12 or General Edu- cational Development [GED], at least some college). Partici- pant age was collected as a continuous variable and then collapsed into 5 age groups (21–24, 25–34, 35–44, 45–54, 55C). Criminal justice involvement was defined as any criminal justice involvement in the last 12 months (having been arrested, on probation or on parole) or none. Current receipt of public benefits was defined as any federal or state benefits, including social security, public assistance, Supple- mental Nutrition Assistance Program (SNAP), Home Energy Assistance Program (HEAP), and HIV/AIDS Service Administration (HASA), or none. Housing status was defined as stable permanent, temporary, or unstable, based on United States Department of Housing and Urban Devel- opment definitions.13 Stable permanent housing included living in one’s own home or apartment. Temporary housing included living in a single room occupancy hotel, transi- tional housing, treatment facility, or family member or friend’s home. Unstable housing was defined as living on the street or in a shelter. Substances used 30 days prior to baseline included licit substances (prescription painkillers, benzodiazepines, methadone, buprenorphine, and alcohol) and illicit substances (heroin and crack/cocaine). No dis- tinction was made between prescribed and nonprescribed use of benzodiazepines, methadone, or buprenorphine. Injecting in the last year (always, sometimes, rarely, and never) was collapsed into a dichotomous variable (any or none). Participants reported current participation in an opi- oid treatment program (OTP) and participation in a syringe exchange program (SEP). We also measured any lifetime

personal experience of an overdose event and total number of lifetime witnessed overdose events.

Independent variables

Good Samaritan Law (GSL) knowledge at time of event. New York State’s GSL, passed in 2011, protects the 911 caller and the overdose victim from charge and prosecution for possession of up to 8 ounces of a controlled substance, alco- hol consumption for underage drinkers, any amount of marijuana, paraphernalia offenses, and sharing drugs.14 To develop the question about knowledge of the GSL, we con- sulted with policy experts and then performed cognitive pretesting on the question. Immediately following OPT and at 3-, 6-, and 12-month follow-ups, participants were asked, “Let’s say you have drugs on you, you see someone overdose, and you call 911. Do you think it would be legal or illegal for the police to prosecute you for your drugs, or are you not sure?” Participants responded “legal,” “illegal,” or “don’t know/not sure.” Participant responses were categorized as correct, incorrect, or unknown. “Illegal” was categorized as correct knowledge of the GSL, and “legal” was categorized as incorrect knowledge of the law. The correct answer to the question was explained to the participant, regardless of response, so participants were reeducated about the GSL at 3-, 6-, and 12-month follow-ups.

Overdose event characteristics. All characteristics of overdose events were reported by study participants who witnessed an overdose. Overdose event characteristics included victim gen- der (male, female, and transgender); relationship to overdose victim (relative, friend, acquaintance and/or drug-using asso- ciate, and stranger); and overdose location (by New York City borough determined by ZIP code or cross street of reported overdose). Overdose setting was categorized as public (aban- doned building, public bathroom, roof, stairway/lobby/eleva- tor, street/park/outside, and subway/bus/car), semiprivate (drug treatment program, hotel, syringe exchange program, shelter, and shooting gallery), and private (participant’s home, overdose victim’s home, someone else’s home, support- ive housing, and single room occupancy [SRO]). Naloxone administration at overdose event was categorized dichoto- mously; an affirmative response included either participant or another bystander administration of naloxone.

Dependent variables

Witnessing an overdose. Three, 6, and 12 months after OPT, participants were asked how many overdoses they had wit- nessed since their last interview, followed by a series of ques- tions about their response to each overdose. An overdose event was defined as witnessing an individual who “is unresponsive or cannot be woken up, collapses, has blue skin color, difficulty breathing, loses consciousness or dies while using drugs.” This analysis includes only participants who witnessed at least 1 overdose after OPT.

911-calling. For each witnessed overdose, participants were asked, “Were you able to call 911 during the overdose?” If they

234 A. JAKUBOWSKI ET AL.

were unable to call, they were asked “What’s the main reason you didn’t call 911?” Predefined field-coded responses included someone else called 911, I didn’t have a phone on me, fear of police/arrest, fear of losing my housing, fear of violating proba- tion/parole, fear of losing my kids, phone was dead/out of minutes, no phone signal, and the person who overdosed woke up. After asking why the participant did not call 911, partici- pants were asked if anyone else called. Events were classified into “911 called,” defined as either the participant or another bystander calling 911, and “911 not called,” defined as neither the participant nor another bystander calling 911.

Statistical analysis

We describe baseline demographic characteristics of study par- ticipants who witnessed 1 or more overdose within 12 months of OPT, and knowledge of the GSL at baseline and 3-, 6-, and 12-month follow-ups. Participants who completed a survey but did not respond to the Good Samaritan question and partici- pants who did not complete a follow-up survey were excluded from this analysis.

Bivariable frequencies were used to compare characteris- tics of overdose events in which 911 was called with events in which 911 was not called. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to assess for asso- ciations between overdose characteristics and 911-calling. Generalized linear models were used to test association between GSL knowledge of participant and whether 911 was called by any bystander. Logistic regression was per- formed using the PROC GENMOD model statement in SAS (SAS Institute, Cary, NC) to account for nonindepen- dence among overdose events witnessed by the same indi- vidual. We included in multivariable analysis variables significant at the P < .05 level in bivariable analysis and excluded variables that met criteria for confounding.15 We also adjusted the final model for witness age, gender, and race. We used a first-order autoregressive covariance struc- ture due to correlation of model residuals in repeated meas- ures. Analyses were carried out using SPSS 22 (IBM, Armonk, NY) and SAS 9.2.

Results

Sample description

Of the 675 individuals trained in overdose prevention at selected programs during the study period, 429 (64%) were approached to participate in the study, and 351 (52%) agreed to enroll. Overall, 299 (85%) completed at least 1 follow-up sur- vey in the 12-month period. Of the 299, 128 (43%) had wit- nessed 1 or more overdoses since baseline, constituting our analytic sample for this report. See Table 1 for descriptive sta- tistics of the 128 participants.

Number of overdoses witnessed at 12 months

The total number of overdose events witnessed by 128 partici- pants was 326. Two thirds (66%) witnessed more than 1 over- dose (range: 1–14): 31% witnessed 2, 18% witnessed 3, and 16%

witnessed 4 or more. In 5 (1.5%) events, the witness reported that the victim did not survive.16

Knowledge of Good Samaritan Law over time

The proportion of participants with correct knowledge of the GSL increased over time. The proportion with correct knowl- edge was 43% (n D 55) immediately after completing OPT, 55% (n D 61) at 3 months, 75% (n D 85) at 6 months, and 78% (n D 94) at 12 months. Incorrect knowledge declined over time, with 40% (n D 51) having incorrect knowledge immedi- ately after OPT, 31% (n D 34) at 3 months, 17% (n D 19) at 6 months, and 12% (n D 15) at 12-month follow-up. The per- centage that reported “don’t know/not sure” did not change over time, with 9% (n D 12) of participants responding “don’t know/not sure” at baseline and 11% (n D 13) responding “don’t know/not sure” at 12-month follow-up.

911-calling

Information is available on 316 overdose events witnessed over the course of 12 months by 128 individuals (Table 2). In 272 events (86% of events with information; 83% of all overdose events), 911 was called by either the participant or another wit- ness as previously reported.16 The most common reasons why participants reported not calling 911 (n D 139) were someone else called 911 (n D 89), the victim woke up (n D 31), and fear of arrest/police (n D 9).

In the events where the overdose witness had correct knowl- edge of the GSL at the time of the event, the unadjusted odds of 911 being called were over 3 times greater than when the wit- ness had incorrect knowledge of the GSL (OR D 3.3, 95% CI: 1.4–7.5). In events where the overdose witness responded they were unsure of the GSL, the unadjusted odds of 911 being called were 7.5 times greater than when the witness had incor- rect knowledge of the GSL (95% CI: 1.5–37.8). The odds of 911 being called for a relative were significantly lower than for a stranger (OR D 0.2, 95% CI: 0.1–0.6). There was no association between naloxone administration (OR D 1.8, 95% CI: 0.9–3.6) and 911-calling.

In the multivariable model, overdoses witnesses with cor- rect knowledge or who were unsure remained more likely to call 911 than witnesses with incorrect knowledge (adjusted OR [AOR] D 3.6 correct knowledge, 95% CI: 1.4–9.4; AOR unsure D 5.9, 95% CI: 1.8–20.1). Overdose setting was also independently associated with 911-calling: witnesses were less likely to call 911 for overdoses that took place in pri- vate settings compared with public settings (AOR D 0.2, 95% CI: 0.1–0.6). We could not include witness relationship to the victim in the multivariable model due to its collinear- ity with overdose setting.

Discussion

In our study, both correct GSL knowledge and no knowl- edge as compared with incorrect knowledge were associated with 911-calling during an overdose event, independent of race, age and gender of the witness, and overdose setting. To our knowledge, this is the first study to investigate the

SUBSTANCE ABUSE 235

relationship between GSL knowledge and 911-calling and highlights the potential impact of a GSL to reduce overdose mortality.

Furthermore, our finding that knowledge of the GSL increased over time has important implications for overdose prevention training. Our study protocol of reinforcing correct knowledge and correcting misinformation at each follow-up suggests that refresher trainings and repeated exposure to information about the GSL can help OPT participants retain and increase their knowledge of the law and, importantly, dis- pel misconceptions that can prevent overdose witnesses from calling 911.

Our findings, however, demonstrate that there are still seri- ous barriers preventing witnesses from calling 911. Overdose witnesses were less likely to call 911 when the overdose hap- pened in a private setting, consistent with other reports.8 This finding is particularly concerning given that previous studies have shown that between 28% and 83% of unintentional opioid overdose deaths occur in private settings.3,10,17,18

We hypothesize that a common reason witnesses were less likely to call 911 in a private setting was fear of police involve- ment and/or arrest at someone’s home. Participants may be reluctant to endanger friends and family members in their home with arrest and fear losing housing if drugs are found on their property by law enforcement.

Similarly, the finding that 911 was less likely to be called when the overdose victim was a witness’ relative than when the victim was a stranger may be the result of fear of exposing a rel- ative to legal risks. Both findings suggest that, despite the pro- tections afforded by the GSL, participants with personal connections to the setting or victim are reluctant to engage help. Additional work is needed to address the real and per- ceived consequences for Good Samaritans in order to further encourage help-seeking behaviors.

We report rates of 911-calling that were at the high end of the range previously reported for fatal and nonfatal overdoses (15%–72%).2,7,8,10,16 One explanation for our finding of high 911-calling rates may be that most other reports studied partici- pants who had not completed OPT. OPT typically includes 911-calling as part of the training. We are only aware of one small study that assessed 911-calling specifically among partici- pants trained in overdose prevention. In that study, the overall 911-calling rate was 43%.4 Another study showed that injection drug users who received information on how to respond to overdose solely from lay sources were less likely to call 911 than those who had received no information about how to respond to an overdose.5 Those who received information from medical or social services were less likely to delay calling 911 and less likely to use ineffective methods to try to resuscitate the overdose victim.5 Given these prior findings, it is plausible that completion of OPT contributed to our participants calling 911 at such high rates.

Finally, we found no association between naloxone adminis- tration and 911-calling, which may allay concerns that nalox- one distribution could inadvertently cause harm by decreasing 911-calling. We found that in the great majority of overdose events, both naloxone is administered and 911 is called. This finding, along with other research showing that drug users do not engage in riskier drug use behavior after receiving take- home naloxone, shows that naloxone’s benefits far outweigh its theoretical risks.19

A strength of our study is that, to our knowledge, it is one of the largest prospective studies of overdose prevention training recipients and the only study to assess the associa- tion between knowledge of GSLs and 911-calling behavior. Our high rate of follow-up achieved with an often transient and difficult to reach population is another strength of our study.

Our findings are subject to several limitations. Since our study is based on self-report, participants could overreport positive responses to overdose, including naloxone adminis- tration and 911-calling. In order to address this limitation, we administered surveys in relative privacy and did not offer incentives for witnessing overdoses or responding to them in any particular way. Another limitation is that our protocol of reeducating participants about the GSL at each follow-up did not allow us to assess knowledge retention over longer periods of time. However, we felt that it would

Table 1. Baseline characteristics of participants who witnessed an overdose within 12 months of overdose (OD) prevention training.

Total n (%)

Total participants who witnessed OD 128 (100) Race White, non-Hispanic 15 (11.7) Black, non-Hispanic 44 (34.4) Hispanic 63 (49.2) Other 6 (4.7)

Gender Male 86 (67.2) Female 42 (32.8)

Age 21-24 2 (1.6) 25-34 16 (12.5) 35-44 24 (18.8) 45-54 56 (43.8) 55C 30 (23.4)

Education Less than high school 41 (32.0) Grade 12 or GED 45 (35.2) Some College 42 (32.8)

Criminal justice involvement 47 (36.7) Receives public benefits 108 (84.4) Housing status Unstablea 24 (19.1) Temporaryb 63 (50.0) Stable Permanentc 39 (31.0)

Substances used in study period Licitd

Prescription Painkillers 75 (58.6) Benzodiazepines 72 (56.3) Methadone 99 (77.3) Buprenorphine 15 (11.7) Alcohol 70 (54.7) Illicit Heroin 65 (50.8) Crack/Cocaine 54 (42.2)

Injected drugs in year prior to baseline 52 (40.6) Participant in opioid treatment program 84 (65.6) Participant in syringe exchange program 82 (64.1) Participant experienced OD in lifetime 55 (43.0) Lifetime mean number of ODs witnessed (SD) 11.4 (25.1)

aUnstable was defined as living on the street or in a shelter. bTemporary collapsed single room occupancy hotel, transitional housing, treat- ment facility, and family member or friend’s home.

cStable permanent was defined as participant’s own home or apartment. dNo distinction was made between prescribed and non-prescribed use.

236 A. JAKUBOWSKI ET AL.

be unethical not to supply the participant with the correct answer.

In conclusion, this is the first study to demonstrate a rela- tionship between knowledge of the GSL and 911-calling during overdose events. Our work highlights the importance of educat- ing people at risk of witnessing overdose about GSLs. Having correct knowledge about the GSL was associated with use of emergency medical care for individuals who experience an overdose, and may be an important strategy to reduce overdose fatalities nationally. For states that have not yet passed GSLs, our research provides evidence that passage of GSLs may pro- mote seeking help for an overdose victim and help reduce opi- oid overdose–related mortality.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors declare that they have no conflicts of interest.

Author contributions

Andrea Jakubowski was involved in collection of data, analysis, interpreta- tion of the results, and writing the manuscript. Hillary Kunins was

involved in research conception and design, interpretation of results, and revision of the manuscript. Zina Huxley-Reicher was involved in collection of data, analysis, interpretation of the results, and revising the manuscript. Anne Siegler was involved in research conception and design and revision of the manuscript.

ORCID

Andrea Jakubowski http://orcid.org/0000-0001-5651-5589

References

Rudd, RA, Aleshire N, Zibbell JE, Gladden RM. Increases in drug and opioid overdose deaths—United States, 2000–2014. MMWR Morb Mortal Wkly Rep. 2016;64:1378–1382. doi: 10.15585/mmwr. mm6450a3. PMID: 26720857 Baca CT, Grant KJ. What heroin users tell us about overdose. J Addict Dis. 2007;26:63–68. doi: 10.1300/J069v26n04_08. PMID: 18032233 Davidson PJ, McLean RL, Kral AH, Gleghorn AA, Edlin BR, Moss AR. Fatal heroin-related overdose in San Francisco, 1997–2000: a case for targeted intervention. J Urban Health. 2003;80:261–273. doi: 10.1093/ jurban/jtg029. PMID: 12791802 Lankenau SE, Wagner KD, Silva K, et al. Injection drug users trained by overdose prevention programs: responses to witnessed overdoses. J Community Health. 2013;38:133–141. doi: 10.1007/s10900-012-9591- 7. PMID: 22847602

Table 2. Factors associated with 911-calling at witnessed overdose (OD) events.

Total n (%) 911 called, n (%) OR (95% CI)a AORb

Total witnessed OD events 316 (100) 272 (100) Gender of OD victim

Male 230 (72.8) 201 (73.9) ref Female 84 (26.6) 69 (25.4) 0.80 (0.5-1.4) Transgender 2 (0.6) 2 (0.7) —

OD locationc

Area of city 1d 180 (58.4) 155 (57.0) ref Area of city 2 92 (29.9) 79 (29.0) 1.1 (0.3-3.9) Area of city 3 36 (11.7) 31 (11.4) 1.25 (0.2-8.6)

OD setting Public 170 (54.0) 158 (58.1) ref ref Semi-Private 53 (16.8) 49 (18.0) 0.6 (0.2-1.9) 0.5 (0.2-1.4) Private 92 (29.2) 64 (23.5) 0.2 (0.1-0.5) 0.2 (0.1-0.6)

Witness relationship to OD victim Relative 19 (6.0) 12 (4.4) 0.2 (0.1-0.6) Friend 115 (36.5) 96 (35.3) 0.9 (0.4-2.5) Acquaintance/Drug Using Associate 97 (30.8) 87 (32.0) 1.5 (0.4-5.1) Stranger 84 (26.7) 76 (27.9) ref

Naloxone given at event 241 (77.2) 210 (77.2) 1.8 (0.9-3.6) Witness Good Samaritan knowledge at time of evente

Correct 212 (67.5) 189 (69.5) 3.3 (1.4-7.5) 3.6 (1.4-9.4) Incorrect 55 (17.5) 41 (15.1) ref ref Don’t know/not sure 47 (15.0) 41 (15.1) 7.5 (1.5-37.8) 5.9 (1.8-20.1)

Witness participant in syringe exchange program 226 (71.5) 189 (69.5) 0.4 (0.1-1.1) Witness participant in opioid treatment program 182 (57.6) 153 (56.3) 0.8 (0.3-2.2) Witness race

White, non-Hispanic 33 (10.4) 25 (9.2) 0.6 (0.2-1.8) 0.6 (0.2-2.6) Black, non-Hispanic 107 (33.9) 102 (37.5) 3.6 (1.0-12.7) 3.1 (1.0-9.8) Hispanic 156 (49.4) 126 (46.3) ref ref Other 20 (6.3) 19 (7.0) — —

Witness gender Male 234 (74.1) 199 (73.2) ref ref Female 82 (25.9) 73 (26.8) 0.72 (0.3-1.9) 1.9 (0.6-5.9)

Witness age (Mean, SD) 47.8 (9.6) 47.9 (9.4) 1.0 (1.0-1.1) 1.0 (1.0-1.1)

aOdds Ratio. 95% CI: 95% Confidence Interval. ORs model events where 911 was called vs. those where 911 was not. OR >1 indicates greater odds of 911 being called. bFinal model adjusts for GSL knowledge, age, gender, race of witness, and OD setting. cn D 308. dArea of city is blinded. en D 314 Two events where GSL knowledge was missing were excluded from this analysis.

[1]

[2]

[3]

[4]

SUBSTANCE ABUSE 237

Pollini RA, McCall L, Mehta SH, Celentano DD, Vlahov D, Strathdee SA. Response to overdose among injection drug users. Am J Prev Med. 2006;31:261–264. doi: 10.1016/j.amepre.2006.04.002. PMID: 16905039 Sherman SG, Gann DS, Scott G, Carlberg S, Bigg D, Heimer R. A qual- itative study of overdose responses among Chicago IDUs. Harm Reduct J. 2008;5:2. doi: 10.1186/1477-7517-5-2. PMID: 18218071 Tobin KE, Davey MA, Latkin CA. Calling emergency medical services during drug overdose: an examination of individual, social and setting correlates. Addiction. 2005;100:397–404. doi: 10.1111/j.1360- 0443.2005.00975.x. PMID: 15733253 Tracy M, Piper TM, Ompad D, et al. Circumstances of witnessed drug overdose in New York City: implications for intervention. Drug Alco- hol Depend. 2005;79:181–190. doi: 10.1016/j.drugalcdep.2005.01.010. PMID: 16002027 Levy B, Spelke B, Paulozzi LJ, et al. Recognition and response to opi- oid overdose deaths—New Mexico, 2012. Drug Alcohol Depend. 2016;167:29–35. doi: 10.1016/j.drugalcdep.2016.07.011. Darke S, Ross J, Zador D, Sunjic S. Heroin-related deaths in New South Wales, Australia, 1992–1996. Drug Alcohol Depend. 2000;60:141–150. doi: 10.1016/S0376-8716(99)00147-7. PMID: 10940541 Atlas Law. The Policy Surveillance Portal. Good Samaritan Overdose Prevention Laws Map. http://www.lawatlas.org/query?datasetDgood- samaritan-overdose-laws. Accessed December 29, 2016. Banta-Green, CJ, Kuszler PC, Coffin PO, Schoeppe JA. (2011). Wash- ington’s 911 Good Samaritan Drug Overdose Law—Initial Evaluation

Results. Seattle, WA: Alcohol & Drug Abuse Institute, University of Washington; 2011. http://adai.uw.edu/pubs/infobriefs/ADAI-IB- 2011-05.pdf. Accessed December 29, 2016. Homeless Emergency Assistance and Rapid Transition to Housing Act of 2009, Pub. L. No. 111-22, 1664 Stat. 123, codifed as amended at 42 U.S.C. x1003. Witness or victim of drug or alcohol overdose, N.Y. Penal Law x220.78 Hern�an MA, Hern�an, AM. Confounding—structure. In: Wiley Stats- Ref: Statistics Reference Online. Chichester, UK: John Wiley & Sons; 2014. doi: 10.1002/9781118445112.stat03729. Siegler A, Huxley-Reicher Maldjian L, Jordan R, Oliver C, Jakubowski A, Kunins HV. Naloxone use among opioid overdose rescue trainees in New York City: a longitudinal cohort study. Drug Alchol Depend. 2017;179:124–130. doi: 10.1016/j.drugalcdep.2017.06.029. Cerd�a M, Ransome Y, Keyes KM, et al. Prescription opioid mortality trends in New York City, 1990–2006: examining the emergence of an epidemic. Drug Alcohol Depend. 2013;132:53–62. doi: 10.1016/j. drugalcdep.2012.12.027. PMID: 23357743 Siegler A, Tuazon E, Bradley O’Brien D, Paone D. Unintentional opi- oid overdose deaths in New York City, 2005–2010: a place-based approach to reduce risk. Int J Drug Policy. 2014;25:569–574. doi: 10.1016/j.drugpo.2013.10.015. PMID: 24412006 Jones JD, Campbell A, Metz VE, Comer SD. No evidence of compen- satory drug use risk behavior among heroin users after receiving take- home naloxone. Addict Behav. 2017;71:104–106. doi: 10.1016/j. addbeh.2017.03.008. PMID: 28325710

[5]

[6]

[7]

[8]

[9]

[10]

1 ]1

1 ]2

[

[

[1 ]

[1 ]

3

4 [1 ]5

[1 ]6

[1 ]7

[1 ]8

[1 ]9

238 A. JAKUBOWSKI ET AL.

Copyright of Substance Abuse is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.