Abstract
Aims
To describe characteristics of participants and overdose reversals associated with a community-based naloxone distribution program and identify predictors of obtaining naloxone refills and using naloxone for overdose reversal.
Design
Bivariate statistical tests were used to compare characteristics of participants who obtained refills and reported overdose reversals, versus those who did not. We fitted multiple logistic regression models to identify predictors of refills and reversals; zero-inflated multiple Poisson regression models were used to identify predictors of number of refills and reversals.
Setting
San Francisco, California, U.S.A.
Participants
Naloxone program participants registered and reversals reported from 2010-2013.
Measurements
Baseline characteristics of participants and reported characteristics of reversals.
Findings
2500 participants were registered and 702 reversals were reported from 2010-2013. Participants who had witnessed an overdose [AOR=2.02(1.53-2.66); AOR=2.73(1.73-4.30)] or used heroin [AOR=1.85(1.44-2.37); AOR=2.19(1.54-3.13)], or methamphetamine [AOR=1.71(1.37-2.15); AOR=1.61(1.18-2.19)] had higher odds of obtaining a refill and reporting a reversal, respectively. African American [Adjusted Odds Ratio=0.63(95%CI=0.45-0.88)] and Latino [AOR=0.65(0.43-1.00)] participants had lower odds of obtaining a naloxone refill whereas Latino participants who obtained at least one refill reported a higher number of refills [Incidence Rate Ratio=1.33(1.05-1.69)].
Conclusions
Community naloxone distribution programs are capable of reaching sizeable populations of high-risk individuals and facilitating large numbers of overdose reversals. Community members most likely to engage with a naloxone program and use naloxone to reverse an overdose are active drug users.
INTRODUCTION
Drug related deaths increased 92.7% worldwide between 1990 and 2013 (1), and drug overdose has been the largest cause of injury-related death among U.S. adults since 2008 (2-4), a trend driven by increased opioid overdose mortality (5). Naloxone, a short-acting mu-opioid antagonist with no abuse potential, is recommended by the World Health Organization and others as a key strategy to reduce mortality related to opioid overdose (6-13). Naloxone has been distributed to heroin users to facilitate lay reversal of opioid overdoses since the mid-1990s (6, 9, 14). As of 2010, there were at least 188 naloxone distribution programs in the U.S. but no programs in 19 of the 25 states with drug overdose death rates above the 2008 median, indicating a critical need to expand community access to naloxone. (6)
Extant literature demonstrates that drug users accept lay naloxone provision and frequently utilize naloxone successfully to reverse opioid overdoses (9). Multiple studies also demonstrate reductions in opioid overdose temporally associated with introduction and expansion of naloxone programming, including an interrupted time-series analysis demonstrating a relative reduction in opioid overdose in communities that distributed naloxone compared to those that did not distribute naloxone. (15, 16) Cohort and programmatic data from naloxone programs have described demographics as well as rates and characteristics of reversals (9). Generally, 10-20% of substance users given naloxone will passively report a reversal event (17-19); however, it is unclear what predicts reversals and what happens to the majority of naloxone kits. Understanding who is continuing to access naloxone programs and who effectively utilizes naloxone in the community would identify programming strengths and gaps and inform the cost-effective expansion of community naloxone distribution.
To address these gaps, we analyzed data from the Drug Overdose Prevention Education Project (DOPE), a San Francisco County overdose prevention and naloxone distribution program in operation since 2003. DOPE was the first naloxone distribution program in the U.S. to be sanctioned and supported by a health department, the San Francisco Department of Public Health, and a prior study reported on participant demographics and refill and reversal characteristics between 2003 and 2009 (18). Utilizing more comprehensive data collection from 2010 onward to understand who engages with a community naloxone program, we attempted to (1) describe DOPE participants and reversal events between 2010 and 2013, (2) compare demographic characteristics among subgroups based on whether or not they had returned to a DOPE site to obtain a refill of naloxone (henceforth referred to as a refill) or used naloxone to reverse an overdose (henceforth referred to as a reversal), (3) identify predictors of obtaining refills and reporting reversals, and (4) identify predictors of the numbers of refills obtained and reversals reported.
METHODS
Data Collection and Measures
DOPE Participants
DOPE provides brief (5-10 minute) training for anyone who might witness or experience an opioid overdose in how to recognize and respond to overdose and dispenses 2-dose intranasal or injectable naloxone kits. Services are provided at needle exchange sites, re-entry programs, pain management clinics, opioid substitution treatment programs, and single room occupancy hotels (SROs). All participants complete a brief questionnaire during enrollment, linked to a unique identifier based on personal information that can be easily recalled on return to the program. Data collected include birthdate, race/ethnicity, gender, housing status, use of specific substances in the preceding 30 days, and history of prior overdose, witnessing of an overdose, naloxone administration, witnessing of naloxone administration.
Naloxone Refill and Reversal Events
Participants returning to obtain a new naloxone kit completed a separate questionnaire in which they indicated whether the refill was due to use or loss of their previous kit. If the participant had used the naloxone, they were asked about the individual to whom the naloxone was administered (e.g., relationship to recipient, substances used) as well as the setting and result of the naloxone administration.
Analysis
Due to more comprehensive data collection since 2010, we limited analysis to participants registered since 2010 and reversals performed since 2010. All data were analyzed using STATA version 12 (College Station, TX). This study was approved and deemed exempt from review by the University of California San Francisco Committee on Human Research (IRB study ID 12-09877).
DOPE Participants
All participants were categorized into subgroups according to whether or not they had obtained a refill or reported an overdose reversal using naloxone. We calculated the mean age and frequencies of demographic and behavioral measures for all participants and those in each subgroup.
Demographic and behavioral characteristic comparisons between those who obtained a refill and those who did not, as well as between those who reported a reversal and those who did not, were conducted using unpaired Student's t-tests and Fisher's exact tests. We excluded any records of clinical registration (n=9) and refills (n=10) where a single unique identifier was given to multiple individuals at initial registration and that unique identifier was associated with at least one refill. Additionally, we excluded from analysis any records of refills (n=183) where no unique identifier could be linked to an existing clinical registration.
We also conducted a subgroup analysis among participants who obtained refills, in which we compared characteristics of those who did and did not report a reversal (please see Supplemental Online Appendix, Table S1)
Naloxone Administration Events
We calculated frequencies of circumstantial measures for all naloxone administration events occurring between January 1, 2010, and December 31, 2013, regardless of when the participant underwent initial clinical registration or whether their unique identifier could be linked to an existing clinical registration.
Multiple Regression Analyses
We used multiple logistic regression models to examine the relationship between demographic (e.g., age, gender, race, housing status) and behavioral characteristics (e.g., prior experience of overdose, prior witnessing of overdose, substance use) of participants trained from 2010-2013 and whether or not participants obtained naloxone refills or reported reversals in the same timeframe (see Table 3 for complete list of predictors). In sensitivity analyses, we fitted logistic regression models described above, but with measures on baseline history of naloxone administration and witnessing of naloxone administration, instead of baseline history of witnessing an overdose; these predictors were not fitted simultaneously to avoid collinearity. We also conducted a subgroup analysis examining predictors of refills and reversals restricted to opioid users.
We conducted an exploratory analysis assessing potential synergistic effects between demographic characteristics (age, race, and gender) and specific behavioral characteristics (heroin use, meth use, and history of witnessing an overdose), in separate multiple logistic regression models predicting odds of obtaining a naloxone refill. We used likelihood ratio tests to assess the joint statistical significance of the independent contributions of each set of interaction terms. Jointly significant interaction terms (p < 0.05) were summarized qualitatively in the results due to the exploratory nature of the analyses.
We used zero-inflated multiple Poisson regression models to examine the relationship between demographic and behavioral characteristics of participants trained from 2010-2013 and counts of both refills and reversals in the same timeframe. Models included the same covariates as the multiple logistic regression models as well as an offset for time of follow-up for each participant, calculated from date of initial registration to December 31, 2013. The zero-inflated Poisson model rests on the assumption that in an unobservable subset of the population, the count outcome is always (structurally) zero, while in a second, complementary subset, it arises from a standard Poisson distribution (including the expected proportion of random zeroes). These models estimate covariate effects on both sub-population membership as well as the mean of the Poisson distribution in the second population subset.
We conducted additional subgroup analyses limiting the logistic regression and Poisson regression models to those who had obtained refills, which is included in the appendix (Tables S2 and S3).
RESULTS
DOPE Participants
DOPE trained and prescribed naloxone to 2500 participants from 2010 to 2013. The majority of participants were male (60.5%), of European background / white (58.8%) and homeless or unstably housed (56.1%), with a mean age of 38.6 (Table 1). Most participants (73.9%) engaged in some illicit substance use in the 30 days prior to initial registration; 50.7% used opioids. Roughly one-third of participants had overdosed (32.5%) and nearly twice as many (63.7%) had witnessed at least one overdose. About one-third (32.0%) had witnessed naloxone administration and 10.8% had used naloxone on another individual prior to their initial training.
Table 1.
Baseline characteristics of DOPE Project participants registered 2010-2013 by refill and reversal status
| Demographic and Behavioral Characteristics* | All Participants | Participants Who Obtained a Refill† | Participants Who Did Not Obtain a Refill | Participants Who Reported a Reversal† | Participants Who Did Not Report a Reversal | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | (%) | N | (%) | N | (%) | N | (%) | N | (%) | |
| N | 2500 | 613 | (24.5) | 1887 | (75.5) | 257 | (10.3) | 2243 | (89.7) | |
| Mean Ageb (SD) | 38.6 | (12.7) | 38.0 | (12.1) | 38.8 | (12.9) | 37.1 | (12.3) | 38.8 | (12.7) |
| Gendera | ||||||||||
| Male | 1513 | (60.5) | 403 | (65.7) | 1110 | (58.8) | 170 | (66.1) | 1343 | (59.9) |
| Female | 929 | (37.2) | 199 | (32.5) | 730 | (38.7) | 81 | (31.5) | 848 | (37.8) |
| Transgender/Other gender | 55 | (2.2) | 11 | (1.8) | 44 | (2.3) | 6 | (2.3) | 49 | (2.2) |
| Missing | 3 | (0.1) | 0 | (0.0) | 3 | (0.2) | 0 | (0.0) | 3 | (0.1) |
| Racea,b | ||||||||||
| European background / white | 1471 | (58.8) | 403 | (65.7) | 1068 | (56.6) | 181 | (70.4) | 1290 | (57.5) |
| African American | 505 | (20.2) | 79 | (12.9) | 426 | (22.6) | 26 | (10.1) | 479 | (21.4) |
| Latino | 224 | (9.0) | 41 | (6.7) | 183 | (9.7) | 16 | (6.2) | 208 | (9.3) |
| Mixed/Other race | 264 | (10.6) | 79 | (12.9) | 185 | (9.8) | 28 | (10.9) | 236 | (10.5) |
| Missing | 36 | (1.4) | 11 | (1.8) | 25 | (1.3) | 6 | (2.3) | 30 | (1.3) |
| Housing Statusa, b | ||||||||||
| Stably housed | 1007 | (40.3) | 197 | (32.1) | 810 | (42.9) | 86 | (33.5) | 921 | (41.1) |
| Homeless/Unstably housed | 1403 | (56.1) | 391 | (63.8) | 1012 | (53.6) | 156 | (60.7) | 1247 | (55.6) |
| Missing | 90 | (3.6) | 25 | (4.1) | 65 | (3.4) | 15 | (5.8) | 75 | (3.3) |
| Overdosea, b | ||||||||||
| Prior overdose | 813 | (32.5) | 256 | (41.8) | 557 | (29.5) | 120 | (46.7) | 693 | (30.9) |
| No Prior overdose | 1505 | (60.2) | 296 | (48.3) | 1209 | (64.1) | 111 | (43.2) | 1394 | (62.1) |
| Missing | 182 | (7.3) | 61 | (10.0) | 121 | (6.4) | 26 | (10.1) | 156 | (7.0) |
| Witness of Overdosea, b | ||||||||||
| Witnessed an overdose | 1592 | (63.7) | 464 | (75.7) | 1128 | (59.8) | 210 | (81.7) | 1382 | (61.6) |
| Never witnessed an overdose | 747 | (29.9) | 97 | (15.8) | 650 | (34.4) | 27 | (10.5) | 720 | (32.1) |
| Missing | 162 | (6.5) | 52 | (8.5) | 110 | (5.8) | 20 | (7.8) | 142 | (6.3) |
| Naloxone Administrationa, b | ||||||||||
| Administered naloxone | 270 | (10.8) | 100 | (16.3) | 170 | (9.0) | 68 | (26.5) | 202 | (9.0) |
| Never administered naloxone | 2016 | (80.6) | 451 | (73.6) | 1565 | (82.9) | 166 | (64.6) | 1850 | (82.5) |
| Missing | 214 | (8.6) | 62 | (10.1) | 152 | (8.1) | 23 | (8.9) | 191 | (8.5) |
| Witness of Naloxone Administrationa, b | ||||||||||
| Witnessed naloxone administration | 800 | (32.0) | 280 | (45.7) | 520 | (27.6) | 147 | (57.2) | 653 | (29.1) |
| Never witnessed naloxone administration | 1484 | (59.4) | 270 | (44.0) | 1214 | (64.3) | 87 | (33.9) | 1397 | (62.3) |
| Missing | 216 | (8.6) | 63 | (10.3) | 153 | (8.1) | 23 | (8.9) | 193 | (8.6) |
| General Substance Use in Last 30 Daysa, b | ||||||||||
| Any substance use | 1847 | (73.9) | 508 | (82.9) | 1339 | (71.0) | 216 | (84.0) | 1631 | (72.7) |
| No substance use | 265 | (10.6) | 19 | (3.1) | 246 | (13.0) | 9 | (3.5) | 256 | (11.4) |
| Missing | 388 | (15.5) | 86 | (14.0) | 302 | (16.0) | 32 | (12.5) | 356 | (15.9) |
| Polydrug Use in Last 30 Daysa, b | ||||||||||
| Polydrug use | 1144 | (45.8) | 371 | (60.5) | 773 | (41.0) | 171 | (66.5) | 973 | (43.4) |
| No polydrug use | 968 | (38.7) | 156 | (25.4) | 812 | (43.0) | 54 | (21.0) | 914 | (40.7) |
| Missing | 388 | (15.5) | 86 | (14.0) | 302 | (16.0) | 32 | (12.5) | 356 | (15.9) |
| Specific Substance Use in Last 30 Days | ||||||||||
| Any opioidsa, b | 1267 | (50.7) | 399 | (65.1) | 868 | (46.0) | 177 | (68.9) | 1090 | (48.6) |
| Heroina, b | 892 | (35.7) | 318 | (51.9) | 574 | (30.4) | 151 | (58.8) | 741 | (33.0) |
| Methadonea, b | 584 | (23.4) | 193 | (31.5) | 391 | (20.7) | 82 | (31.9) | 502 | (22.4) |
| Benzodiazepinesa, b | 497 | (19.9) | 174 | (28.4) | 323 | (17.1) | 87 | (33.9) | 410 | (18.3) |
| Other opioidsa, b | 612 | (24.5) | 206 | (33.6) | 406 | (21.5) | 95 | (37.0) | 517 | (23.0) |
| Cocaine/Cracka, b | 687 | (27.5) | 219 | (35.7) | 468 | (24.8) | 93 | (36.2) | 594 | (26.5) |
| Alcohol | 864 | (34.6) | 221 | (36.1) | 643 | (34.1) | 90 | (35.0) | 774 | (34.5) |
| Methamphetamine/Speeda, b | 776 | (31.0) | 268 | (43.7) | 508 | (26.9) | 122 | (47.5) | 654 | (29.2) |
| Other substances | 371 | (14.8) | 107 | (17.5) | 264 | (14.0) | 48 | (18.7) | 323 | (14.4) |
All characteristics were assessed at baseline during initial registration between January 1, 2010 and December 31, 2013.
All considered refills and reversals occurred between January 1, 2010 and December 31, 2013.
P < 0.05 for Fisher exact or t-test comparing participants who did and did not obtain refills
P < 0.05 for Fisher exact or t-test comparing participants who did and did not report reversals
Among participants who obtained at least 1 refill, the median number of refills was 1 (IQR 1-3). The differences between participants who obtained at least one refill compared to those who did not return for any refills are summarized in Table 1. Those who received refills were significantly more likely to be male, of European background, homeless or unstably housed or to have previously overdosed, witnessed an overdose, used naloxone on another individual, or witnessed the use of naloxone on another individual. Additionally, participants who obtained refills were more likely to use any drugs, multiple drugs, heroin, other opioids, or methamphetamine.
Correlates were similar for those who reported at least one reversal (median overdose reversals reported was 1 [IQR 1-2]). Compared to those who did not report any reversals, those who reported at least one reversal were younger, more likely to be of European background and homeless or unstably housed, or to have overdosed, witnessed an overdose, used naloxone on another individual, or witnessed the use of naloxone on another individual. Additionally, participants who reported reversals were more likely to use any drugs, multiple drugs, heroin, or methamphetamine.
Comparisons between participants who did and did not report reversals among those who obtained refills are presented in Table S1.
Naloxone Administration Events
DOPE recorded 702 naloxone administration events between 2010 and 2013 (405 reported by participants initially registered from 2010-2013, 270 by participants registered before 2010, and 27 by participants whose unique identifier was not linkable to a registration record). The most common setting was a private residence (40.5%), followed by a SRO (29.6%). Naloxone was mostly used on companions or acquaintances of participants (74.6%) and 95.7% of recipients were known to have survived. Of 10 reported deaths (1.4% of reversals), in 6 cases participants reported that they arrived too late but administered naloxone anyway. Heroin was the most commonly reported substance consumed by the person reversed by naloxone (90.3%). Roughly one-quarter (27.4%) of reversal attempts also involved a call to emergency medical services (Table 2).
Table 2.
Characteristics of reported reversal events reported 2010-2013 by DOPE Project participants
| Reversals by All Participants'
regardless of Initial Registration Date |
Reversals by Participants
Initially Registered 2010 - 2013 |
|||
|---|---|---|---|---|
| N | (%) | N | (%) | |
| N | 702 | 405 | ||
| Setting | ||||
| Private residence | 284 | (40.5) | 154 | (38.0) |
| Single room occupancy hotel (SRO) | 208 | (29.6) | 120 | (29.6) |
| Public park | 58 | (8.3) | 39 | (9.6) |
| Public toilet | 27 | (3.8) | 15 | (3.7) |
| Other | 102 | (14.5) | 70 | (17.3) |
| Missing | 23 | (3.3) | 7 | (1.7) |
| Recipient Relationship | ||||
| Companion (friend, partner, family member) | 524 | (74.6) | 295 | (72.8) |
| Stranger | 139 | (19.8) | 91 | (22.5) |
| Self | 34 | (4.8) | 17 | (4.2) |
| Missing | 5 | (0.7) | 2 | (0.5) |
| Result | ||||
| Reversed, all reasons | 673 | (95.9) | 390 | (96.3) |
| Reversed due to participant administering naloxone | 658 | (93.7) | 385 | (95.1) |
| Reversed following EMS response | 15 | (2.1) | 5 | (1.2) |
| Death | 10† | (1.4) | 9 | (2.2) |
| Unknown | 9 | (1.3) | 1 | (0.2) |
| Missing | 10 | (1.4) | 2 | (0.5) |
| General Substance Use Details | ||||
| Single substance | 416 | (59.3) | 246 | (60.7) |
| Multiple substances | 286 | (40.7) | 159 | (39.3) |
| Missing | 18 | (2.6) | 9 | (2.2) |
| Specific Substance | ||||
| Heroin | 634 | (90.3) | 363 | (89.6) |
| Heroin alone | 379 | (54.0) | 222 | (54.8) |
| Heroin with other substances | 255 | (36.3) | 141 | (34.8) |
| Benzodiazepine | 107 | (15.2) | 57 | (14.1) |
| Alcohol | 109 | (15.5) | 56 | (13.8) |
| Other Opioids | 90 | (12.8) | 58 | (14.3) |
| Meth | 86 | (12.3) | 53 | (13.1) |
| Cocaine/Crack | 54 | (7.7) | 30 | (7.4) |
| Methadone | 37 | (5.3) | 20 | (4.9) |
| Other Drugs | 14 | (2.0) | 8 | (2.0) |
| Other Measures Taken | ||||
| Sternum rub | 243 | (34.6) | 136 | (33.6) |
| Call 9-1-1 | 192 | (27.4) | 118 | (29.1) |
| Rescue breathing | 364 | (51.9) | 209 | (51.6) |
| Missing | 3 | (0.4) | 1 | (0.2) |
Of 10 deaths, victim was already deceased in 2, respondent reported arriving “too late” in 4 others.
Multiple Regression Analyses
The results of multiple logistic regression analyses evaluating predictors of obtaining at least one refill and reporting at least one reversal are summarized in Table 3. With regard to refills, African American and Latino participants had lower odds of obtaining a refill. Participants of mixed or other race and those who had witnessed an overdose, used heroin, or used methamphetamine had higher odds of obtaining a refill. In a subgroup analysis examining only opioid using participants, African American participants also had lower odds of obtaining a naloxone refill [AOR 0.59 (95% CI, 0.39-0.90)].
Table 3.
Multiple logistic regression models predicting naloxone refills and reversals among DOPE Project participants registered 2010-2013 (N = 1972)
| Refill† |
Reversal† |
|||||
|---|---|---|---|---|---|---|
| AOR | 95% CI | P-Value | AOR | 95% CI | P-Value | |
| Age | 1.01 | (1.00, 1.02) | 0.125 | 1.00 | (0.99, 1.02) | 0.620 |
| Gender | ||||||
| Male | - | - | - | - | - | - |
| Female | 0.81 | (0.64, 1.02) | 0.073 | 0.96 | (0.70, 1.33) | 0.825 |
| Transgender/other | 0.76 | (0.32, 1.81) | 0.533 | 1.27 | (0.42, 3.84) | 0.666 |
| Race | ||||||
| European background / white | - | - | - | - | - | - |
| African American | 0.63* | (0.45, 0.88) | 0.007 | 0.62 | (0.37, 1.03) | 0.063 |
| Latino | 0.65* | (0.43, 1.00) | 0.050 | 0.58 | (0.31, 1.09) | 0.091 |
| Mixed/other race | 1.51* | (1.07, 2.11) | 0.018 | 1.13 | (0.70, 1.82) | 0.614 |
| Homeless | 1.14 | (0.90, 1.44) | 0.280 | 0.95 | (0.69, 1.30) | 0.734 |
| Prior overdose | 1.03 | (0.82, 1.31) | 0.775 | 1.14 | (0.83, 1.57) | 0.411 |
| Witnessed overdose | 2.02* | (1.53, 2.66) | <0.001 | 2.73* | (1.73, 4.30) | <0.001 |
| Use heroin | 1.85* | (1.44, 2.37) | <0.001 | 2.19* | (1.54, 3.13) | <0.001 |
| Use methadone | 1.22 | (0.96, 1.56) | 0.109 | 0.99 | (0.71, 1.37) | 0.934 |
| Use benzodiazepines | 1.14 | (0.87, 1.51) | 0.340 | 1.35 | (0.94, 1.94) | 0.108 |
| Use other opioids | 1.18 | (0.92, 1.52) | 0.201 | 1.25 | (0.89, 1.75) | 0.203 |
| Use cocaine/crack | 0.15 | (0.92, 1.52) | 0.187 | 1.04 | (0.74, 1.46) | 0.833 |
| Use alcohol | 0.82 | (0.65, 1.03) | 0.094 | 0.72* | (0.52, 1.00) | 0.049 |
| Use methamphetamine | 1.71* | (1.37, 2.15) | <0.001 | 1.61* | (1.18, 2.19) | 0.003 |
| Use other drugs | 1.00 | (0.75, 1.32) | 0.978 | 0.97 | (0.66, 1.43) | 0.875 |
All considered refills and reversals occurred between January 1, 2010 and December 31, 2013.
P ≤ 0.05
In the exploratory models addressing the interaction effects between demographic and behavioral predictors, those with history of overdose had differential odds of obtaining a refill depending on the race of the participant, with a stronger effect among African Americans and participants of other / mixed race compared to participants of European background. There were no other significant interaction effects in these models.
With regard to reversals, participants who used alcohol had lower odds of reporting a reversal while participants who had witnessed an overdose, used heroin, or used methamphetamine had higher odds of reporting a reversal. In a subgroup analysis examining only opioid-using participants, African American participants had lower odds of reporting a reversal [AOR, 0.50 (0.26-0.95)].
In sensitivity analyses, both prior naloxone administration [AOR 1.37 (1.01-1.88); AOR 2.49 (1.73-3.59)] and previously witnessing naloxone being administered [AOR 1.85 (1.47-2.32); AOR 2.67 (1.94-3.67)] were associated with higher odds of obtaining a refill and reporting a reversal, respectively.
Table 4 summarizes the results of the zero-inflated multiple Poisson regression models evaluating predictors of the number of naloxone refills and number of overdose reversal events among those who obtained at least one refill and reported at least one reversal, respectively. With regard to refills, Latino participants who had obtained at least one refill had a greater average number of refills than participants of European background. Participants who were homeless, had experienced an overdose, used heroin, or used methamphetamine also had a greater average number of naloxone refills. Those who used methadone had a lower number of refills than those who did not use methadone. With regard to reversal events, participants of mixed or other race reported a greater average number of reversals than participants of European background, and those who used heroin or methamphetamine reported a greater average number of reversal events than those who did not use these substances.
Table 4.
Zero-inflated multiple Poisson regression models predicting naloxone refill and reversal counts among DOPE Project participants registered 2010-2013 (N = 1972)
| Refill Countt |
Refill Count Inflate
Componentt |
|||||
|---|---|---|---|---|---|---|
| IRR | 95% CI | P-Value | Coefficient | 95% CI | P-Value | |
| Age | 0.98* | (0.98, 0.99) | <0.001 | −0.03* | (−0.04, −0.01) | <0.001 |
| Gender | ||||||
| Male | - | - | - | - | - | - |
| Female | 1.11 | (0.97, 1.28) | 0.132 | 0.29* | (0.01, 0.57) | 0.040 |
| Transgender/other | 0.90 | (0.48, 1.68) | 0.745 | 0.32 | (−0.79, 1.43) | 0.577 |
| Race | ||||||
| European background / white | - | - | - | - | - | - |
| African American | 0.97 | (0.75, 1.26) | 0.831 | 0.62* | (0.20, 1.05) | 0.004 |
| Latino | 1.33* | (1.05, 1.69) | 0.019 | 0.67* | (0.19, 1.15) | 0.006 |
| Mixed/other race | 0.84 | (0.69, 1.03) | 0.095 | −0.60* | (−1.06, −0.15) | 0.010 |
| Homeless | 1.24* | (1.07, 1.45) | 0.005 | 0.06 | (−0.24, 0.35) | 0.711 |
| Prior overdose | 1.22* | (1.07, 1.41) | 0.004 | 0.05 | (−0.23, 0.33) | 0.732 |
| Witnessed overdose | 0.96 | (0.81, 1.15) | 0.683 | −0.70* | (−1.02, −0.38) | <0.001 |
| Use heroin | 1.30* | (1.09, 1.54) | 0.003 | −0.62* | (−0.93, −0.31) | <0.001 |
| Use methadone | 0.83* | (0.71, 0.96) | 0.013 | −0.30 | (−0.61, 0.01) | 0.055 |
| Use benzodiazepines | 0.93 | (0.80, 1.09) | 0.389 | −0.28 | (−0.62, 0.06) | 0.110 |
| Use other opioids | 1.05 | (0.90, 1.22) | 0.535 | −0.16 | (−0.47, 0.15) | 0.321 |
| Use cocaine/crack | 1.09 | (0.94, 1.27) | 0.240 | −0.08 | (−0.39, 0.23) | 0.601 |
| Use alcohol | 0.93 | (0.80, 1.07) | 0.298 | 0.21 | (−0.08, 0.49)_ | 0.158 |
| Use methamphetamine | 1.83* | (1.57, 2.13) | <0.001 | −0.36* | (−0.64, −0.08) | 0.011 |
| Use other drugs | 0.79* | (0.66, 0.95) | 0.012 | −0.10 | (−0.46, 0.25) | 0.563 |
| Reversal Count† |
Reversal Count Inflate
Component† |
|||||
|---|---|---|---|---|---|---|
| IRR | 95% CI | P-Value | Coefficient | 95% CI | P-Value | |
| Age | 0.99 | (0.98, 1.01) | 0.432 | −0.01 | (−0.03, 0.00) | 0.133 |
| Gender | ||||||
| Male | - | - | - | - | - | - |
| Female | 1.31 | (0.95, 1.81) | 0.105 | 0.20 | (−0.20, 0.61) | 0.328 |
| Transgender/other | 0.92 | (0.28, 2.97) | 0.884 | −0.20 | (−1.60, 1.19) | 0.776 |
| Race | ||||||
| European background / white | - | - | - | - | - | - |
| African American | 1.49 | (0.83, 2.65) | 0.180 | 0.88* | (0.23, 1.53) | 0.008 |
| Latino | 1.56 | (0.93, 2.61) | 0.094 | 0.91* | (0.19, 1.63) | 0.013 |
| Mixed/other race | 1.75* | (1.16, 2.65) | 0.008 | 0.25 | (−0.31, 0.81) | 0.383 |
| Homeless | 0.84 | (0.61, 1.16) | 0.297 | 0.00 | (−0.41, 0.41) | 0.994 |
| Prior overdose | 1.29 | (0.94, 1.76) | 0.114 | 0.01 | (−0.40, 0.42) | 0.955 |
| Witnessed overdose | 1.33 | (0.82, 2.15) | 0.248 | −0.86* | (−1.44, −0.29) | 0.003 |
| Use heroin | 1.61* | (1.05, 2.47) | 0.028 | −0.62* | (−1.11, −0.13) | 0.013 |
| Use methadone | 0.96 | (0.70, 1.30) | 0.789 | 0.08 | (−0.35, 0.50) | 0.727 |
| Use benzodiazepines | 1.00 | (0.68, 1.47) | 0.996 | −0.40 | (−0.89, 0.08) | 0.105 |
| Use other opioids | 1.07 | (0.76, 1.51) | 0.701 | −0.23 | (−0.68, 0.22) | 0.312 |
| Use cocaine/crack | 0.81 | (0.58, 1.14) | 0.230 | −0.12 | (−0.57, 0.32) | 0.589 |
| Use alcohol | 0.84 | (0.61, 1.16) | 0.295 | 0.27 | (−0.15, 0.69) | 0.203 |
| Use methamphetamine | 1.61* | (1.18, 2.19) | 0.003 | −0.31 | (−0.71, 0.09) | 0.132 |
| Use other drugs | 0.85 | (0.57, 1.25) | 0.412 | −0.06 | (−0.57, 0.45) | 0.816 |
All considered refills and reversals occurred between January 1, 2010 and December 31, 2013.
P ≤ 0.05
Results of subgroup analyses assessing the odds of reporting a reversal and the number of reversals among participants who had obtained refills are reported in the appendix. For the multiple logistic regression analysis among this subgroup, those who had witnessed an overdose or used heroin had higher odds of reporting a reversal (Table S2). For the zero-inflated multiple Poisson regression analysis among this subgroup, female participants reported a greater average number of reversals than male participants and African American participants and those of mixed / other race reported a greater average number of reversals than participants of European background. Participants who had a prior overdose, used heroin, or used methamphetamine also reported greater average numbers of reversals. Participants who used cocaine or crack reported a lower average number of reversals (Table S3).
DISCUSSION
We identified unique demographic and behavioral characteristics associated with continuous engagement in a naloxone distribution program. Specifically, we observed that participants who were of European background, had prior experiences with overdoses, and used heroin or methamphetamine were more likely to return for refills and those who had prior experiences with overdoses and used heroin or methamphetamine were more likely to report reversals. To our knowledge, these findings are the first to analyze predictors of obtaining naloxone refills and of reporting reversals. These results emphasize the impact, with regard to utilization of this medication, of directly reaching drug users with lay naloxone programming.
DOPE participants were mostly of European background, homeless or unstably housed, and used multiple substances, while about one-third drank alcohol and one-third reported a prior personal overdose. These findings are highly consistent with other research on overdose identifying European background, homelessness, use of multiple substances, use of alcohol, and having experienced a prior overdose as risk factors for a subsequent overdose (20-23). These results suggest that naloxone distributed through DOPE is reaching very high risk individuals and, based on the multiple logistic regression analyses, that these high-risk individuals are also the most likely to utilize this intervention in an overdose scenario.
Additionally, multiple logistic regression analyses revealed that witnessing an overdose was significantly associated with greater odds of both obtaining a refill and reporting a reversal. This relationship is supported by the Information-Motivation-Behavior Skills (IMB) model of behavior change, which posits that behavior change occurs when individuals have information about a target behavior, motivation to prevent future outcomes and skills to prevent or reverse outcomes (24, 25). DOPE offers information on overdose management, having witnessed an overdose may provide the motivation to prepare for future overdoses, and DOPE provides participants with tools to address overdose. Our findings suggest that those who are at high risk for overdose are also likely to witness an overdose, consistent with other data, and that the most efficient approach to getting naloxone at the scene of overdose events is to put it in the hands of those at greatest risk for overdose themselves (26).
We also reported several unique findings that merit further exploration. Participants who used methadone obtained fewer refills; if methadone use was in a maintenance program, participants may be using illicit opioids less frequently and thus be less likely to visit a DOPE site or witness an overdose (27). Participants reporting methamphetamine use had higher odds of obtaining a refill and reporting a reversal. We assessed and found a significant bivariate association between methamphetamine use and polydrug use, suggesting that methamphetamine users may be more likely to use multiple substances, highlighting the importance of engaging drug users broadly, and not just opioid users, as part of naloxone distribution programs.
With regard to race/ethnicity, African Americans constitute only 6.1% of San Francisco County residents yet made up 23.9% of accidental drug-related deaths in 2010-2011, including those related to opioid overdose (28, 29), and 20% of DOPE participants, again suggesting that DOPE is effective in reaching at-risk populations. Moreover, Latino participants who reported obtaining refills on average obtained greater number of refills and, among participants who obtained refills, African Americans who reported at least one reversal on average reported more reversals (Table S3), suggesting that some participants of racial/ethnic minority groups were heavily engaged in naloxone programming. This is further supported by our assessment of statistical interaction between race and prior witnessing of an overdose, in which the association between witnessing of an overdose and odds of obtaining a refill was stronger in African Americans and participants of mixed / other race compared to participants of European background. Regardless of these findings on number of refills and number of reversals, both African American and Latino participants had lower odds of any obtaining refills. Even among the subgroup of only-opioid using participants, African Americans were less likely to both obtain refills and report any reversals, suggesting that this finding is robust and not mediated by drug of choice. The discrepancy between lower early engagement with naloxone programming and the potential for higher subsequent engagement among racial/ethnic minority participants reveals an important opportunity to further engage these populations through outreach efforts and services that are both culturally sensitive and relevant. These findings also highlight the need for further research, including the use of qualitative methods, to better understand the underlying causes of these racial/ethnic trends.
DOPE has documented a substantial increase in reported annual reversals with naloxone since 2003 (from 5 in 2003 to 252 in 2013) which has paralleled a substantial decline in heroin-related overdose mortality in San Francisco (18, 28, 30, 31). Moreover, our results show that 90.3% of the naloxone administrations have targeted overdoses that involved heroin. The tendency of participants to use naloxone primarily to reverse heroin-related overdoses, the growing number of reversals, and the declining number of deaths from heroin overdose suggest that the DOPE Project and naloxone distribution may be having an effect on population-level overdose mortality among heroin users. As heroin-related overdose mortality has decreased in San Francisco, however, there has been a corresponding rise in deaths and emergency room visits related to opioid analgesics (28, 30-32). Opioids besides heroin were used by a minority of DOPE participants and were involved in only a small minority of reversed overdoses, suggesting that DOPE activities are not sufficiently reaching the population of opioid analgesic users at risk for overdose. The finding that the use of heroin is associated with greater odds of obtaining a refill of naloxone or reporting a reversal, as shown in our multiple logistic regression analyses, is likely because most DOPE activities are based at needle exchange program sites, highlighting the need to explore alternative naloxone distribution streams to reach other populations at risk. Efforts are underway in San Francisco to achieve this by identifying patients receiving opioid analgesics through primary care for naloxone prescription (NIDA R21 DA036776).
This study has several limitations. All information is self-reported by participants and may be subject to social desirability and/or recall bias. Additionally, reversals were only reported by participants who visited a DOPE site to obtain a refill of naloxone (i.e., “passive surveillance”), which excludes reversals that may have been performed by participants who did not obtain refills. Thus, the reversal analyses may be subject to collider-stratification bias, if obtaining a refill is a common effect of our predictors and the reversals analyzed. Behavioral characteristics of participants, many of which are subject to changes over time, are only collected during initial registration, precluding the analyses from accounting for the effects of recent changes in characteristics such as housing status, substance use, and overdose experience.
In conclusion, we have identified demographic and behavioral characteristics associated with subsequent engagement with naloxone programming and utilization of naloxone to reverse overdoses. Our results reaffirm the success of community-based naloxone distribution in reaching the most at-risk populations and identify opportunities for service expansion to reach or re-engage other populations in need.
Supplementary Material
ACKGNOWLEDGEMENTS
This study was supported by funding from the National Institutes of Health (NIDA R03 DA038084). Authors would like to acknowledge those who participated in the establishment and early conduct of the DOPE Project, including Pete Morse, Rachel McClean, Emalie Huriaux, Lauren Enteen, and Alex Kral, as well as medical director Josh Bamberger.
Footnotes
Declarations of Interest
Authors declare no conflicts.
REFERENCES
- 1.Mortality GBD, Causes of Death C Global, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 2015;385:117–171. doi: 10.1016/S0140-6736(14)61682-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention . Prescription Drug Overdose. 2014. [Google Scholar]
- 3.Centers for DISEASE Control and Prevention . Policy Impact: Prescription Painkiller Overdoses. 2014. [Google Scholar]
- 4.Warner M, Chen LH, Makuc DM, Anderson RN, Minino AM. Drug poisoning deaths in the United States, 1980-2008. NCHS data brief. 2011:1–8. [PubMed] [Google Scholar]
- 5.Calcaterra S, Glanz J, Binswanger IA. National trends in pharmaceutical opioid related overdose deaths compared to other substance related overdose deaths: 1999-2009. Drug and alcohol dependence. 2013;131:263–270. doi: 10.1016/j.drugalcdep.2012.11.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Centers for Disease C. Prevention. Community-based opioid overdose prevention programs providing naloxone - United States, 2010. MMWR Morbidity and mortality weekly report. 2012;61:101–105. [PMC free article] [PubMed] [Google Scholar]
- 7.Davis C, Webb D, Burris S. Changing law from barrier to facilitator of opioid overdose prevention. The Journal of law, medicine & ethics : a journal of the American Society of Law, Medicine & Ethics. 2013;41(Suppl 1):33–36. doi: 10.1111/jlme.12035. [DOI] [PubMed] [Google Scholar]
- 8.COffin PO, Sullivan SD. Cost-effectiveness of distributing naloxone to heroin users for lay overdose reversal. Annals of internal medicine. 2013;158:1–9. doi: 10.7326/0003-4819-158-1-201301010-00003. [DOI] [PubMed] [Google Scholar]
- 9.Clark AK, Wilder CM, Winstanley EL. A systematic review of community opioid overdose prevention and naloxone distribution programs. Journal of addiction medicine. 2014;8:153–163. doi: 10.1097/ADM.0000000000000034. [DOI] [PubMed] [Google Scholar]
- 10.Chamberlain JM, Klein BL. A comprehensive review of naloxone for the emergency physician. The American journal of emergency medicine. 1994;12:650–660. doi: 10.1016/0735-6757(94)90033-7. [DOI] [PubMed] [Google Scholar]
- 11.Dahan A, Aarts L, Smith TW. Incidence, Reversal, and Prevention of Opioid-induced Respiratory Depression. Anesthesiology. 2010;112:226–238. doi: 10.1097/ALN.0b013e3181c38c25. [DOI] [PubMed] [Google Scholar]
- 12.World Health Organization . WHO Model List of Essential Medicines. 18th Edition 2013. [Google Scholar]
- 13.World Health Organization . Consolidated guidelines on HIV prevention, treatment and care for key populations. World Health Organization; Geneva, Switzerland: 2014. [PubMed] [Google Scholar]
- 14.Sporer KA, Kral AH. Prescription naloxone: a novel approach to heroin overdose prevention. Ann Emerg Med. 2007;49:172–177. doi: 10.1016/j.annemergmed.2006.05.025. [DOI] [PubMed] [Google Scholar]
- 15.Walley AY, Xuan Z, Hackman HH, Quinn E, Doe-Simkins M, Sorensen-Alawad A, et al. Opioid overdose rates and implementation of overdose education and nasal naloxone distribution in Massachusetts: interrupted time series analysis. Bmj. 2013;346:f174. doi: 10.1136/bmj.f174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Maxwell S, Bigg D, Stanczykiewicz K, Carlberg-Racich S. Prescribing naloxone to actively injecting heroin users: a program to reduce heroin overdose deaths. Journal of addictive diseases. 2006;25:89–96. doi: 10.1300/J069v25n03_11. [DOI] [PubMed] [Google Scholar]
- 17.Doe-Simkins M, Walley AY, Epstein A, Moyer P. Saved by the nose: bystander-administered intranasal naloxone hydrochloride for opioid overdose. American journal of public health. 2009;99:788–791. doi: 10.2105/AJPH.2008.146647. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Enteen L, Bauer J, McLean R, Wheeler E, Huriaux E, Kral AH, et al. Overdose prevention and naloxone prescription for opioid users in San Francisco. Journal of urban health : bulletin of the New York Academy of Medicine. 2010;87:931–941. doi: 10.1007/s11524-010-9495-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bennett AS, Bell A, Tomedi L, Hulsey EG, Kral AH. Characteristics of an overdose prevention, response, and naloxone distribution program in Pittsburgh and Allegheny County. Pennsylvania, Journal of urban health : bulletin of the New York Academy of Medicine. 2011;88:1020–1030. doi: 10.1007/s11524-011-9600-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Binswanger IA, Blatchford PJ, Mueller SR, Stern MF. Mortality after prison release: opioid overdose and other causes of death, risk factors, and time trends from 1999 to 2009. Annals of internal medicine. 2013159:592–600. doi: 10.7326/0003-4819-159-9-201311050-00005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Coffin PO, Tracy M, Bucciarelli A, Ompad D, Vlahov D, Galea S. Identifying injection drug users at risk of nonfatal overdose. Acad Emerg Med. 2007;14:616–623. doi: 10.1197/j.aem.2007.04.005. [DOI] [PubMed] [Google Scholar]
- 22.Warner M, Chen LH, Makuc DM. Increase in fatal poisonings involving opioid analgesics in the United States, 1999-2006. NCHS data brief. 2009:1–8. [PubMed] [Google Scholar]
- 23.Jenkins LM, Banta-Green CJ, Maynard C, Kingston S, Hanrahan M, Merrill JO, et al. Risk factors for nonfatal overdose at Seattle-area syringe exchanges. Journal of urban health : bulletin of the New York Academy of Medicine. 2011;88:118–128. doi: 10.1007/s11524-010-9525-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Fisher JD, Fisher WA. Changing AIDS-risk behavior. Psychological bulletin. 1992;111:455–474. doi: 10.1037/0033-2909.111.3.455. [DOI] [PubMed] [Google Scholar]
- 25.Fisher WA, Fisher JD, Harman J. The Information-Motivation-Behavioral Skills model: A general social psychological approach to understanding and promoting health behavior. In: J. S., K.A. W., editors. Social Psychological Foundations of Health and Illness. Blackwell; Malden, MA: 2003. pp. 82–106. [Google Scholar]
- 26.Bohnert AS, Tracy M, Galea S. Circumstances and witness characteristics associated with overdose fatality. Annals of emergency medicine. 2009;54:618–624. doi: 10.1016/j.annemergmed.2009.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Marsch LA. The efficacy of methadone maintenance interventions in reducing illicit opiate use, HIV risk behavior and criminality: a meta-analysis. Addiction. 1998;93:515–532. doi: 10.1046/j.1360-0443.1998.9345157.x. [DOI] [PubMed] [Google Scholar]
- 28.Hart A. Office of the Chief Medical Examiner City and County of San Francisco: Fiscal Year 2013-2014 Annual Report for FY 2010-2011. San Francisco, CA: 2013. [Google Scholar]
- 29.U.S. Census Bureau . State and County QuickFacts. U.S. Census Bureau; 2014. [Google Scholar]
- 30.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. Journal of urban health : bulletin of the New York Academy of Medicine. 200380:261–273. doi: 10.1093/jurban/jtg029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Hart A. Office of the Chief Medical Examiner City and County of San Francisco: Annual Report for July 2002 to June 2003. San Francisco, CA: 2004. [Google Scholar]
- 32.Substance Abuse and Mental Health Services Administration (SAMHSA) Drug Abuse Warning Network, 2011 Emergency Department Excel Files, Metro Division, San Francisco. Div_2011_AllMA S. S. Office of Applied Studies, SAMHSA; Rockville, MD: 2011. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.