CLINICIAN’S CAPSULE
What is known about the topic?
Not all patients at risk of opioid overdose accept emergency department (ED)-based take-home naloxone (THN).
What did this study ask?
Why do at-risk ED patients refuse or accept THN?
What did this study find?
Those refusing THN felt: 1) not at risk of overdose; or 2) their ED visit was not the right time or place for THN. Those accepting THN wanted to save the lives of others.
Why does this study matter to clinicians?
Those refusing ED THN may accept THN elsewhere if referred to appropriate community services for overdose risk education and THN distribution.
INTRODUCTION
In 2016, an unprecedented 2,800 Canadians died of an opioid overdose (OD), 1 while the hospitalization rates and emergency department (ED) visits for opioid OD have increased correspondingly. 2 Take-home naloxone (THN), consisting of several doses of naloxone, single-use instructions, simple rescue breathing masks, and educational brochures, is a harm reduction intervention for bystander opioid OD reversal. Dispensing THN to at-risk users has reversed over 26,000 ODs in the United StatesReference Wheeler, Jones, Gilbert and Davidson 3 and decreased opioid-related community death rates.Reference Walley, Xuan and Hackman 4 Seven of the 13 Canadian provinces and territories have initiatives to expand THN access and coverage. 2 , 5
As ED visits and non-fatal ODs are markers for subsequent OD mortality,Reference Wheeler, Jones, Gilbert and Davidson 3 , Reference Brady, DiMaggio and Keyes 6 , Reference Caudarella, Dong and Milloy 7 EDs become an opportune setting to reach those at highest risk. In North America, EDs increasingly dispense THN,Reference Kestler, Buxton and Meckling 8 - Reference Ellison, Walley and Feldman 13 and 70% of ED patients accept THN.Reference Kestler, Buxton and Meckling 8 While physician barriers to ED-based THN dispensation have been reported,Reference Samuels, Dwyer and Mello 9 , Reference Drainoni, Koppelman and Feldman 12 , Reference Lacroix, Thurgur, Orkin, Perry and Stiell 14 patient-specific reasons for ED THN acceptance or refusal have not. In non-ED settings, factors contributing to refusals include the reluctance to administer injections, fear of triggering additional drug use, and inappropriate minimization of OD risk.Reference Sondhi, Ryan and Day 15 - Reference Tzemis, Al-Qutub and Amlani 19 Conversely, patients are motivated to accept THN if they feel they can help others,Reference Wright, Oldham, Francis and Jones 17 , Reference Tzemis, Al-Qutub and Amlani 19 , Reference Wagner, Davidson and Iverson 20 if they have had an OD themselves,Reference Lagu, Anderson and Stein 18 or if they know a peer with THN.Reference Wagner, Iverson and Wong 21
To improve the quality and reach of ED THN distribution and education, it is crucial to gain a greater understanding of the reasons influencing THN refusal or acceptance for ED patients. We surveyed high-risk ED patients to explore reasons for refusing or accepting THN.
METHODS
Design and philosophy
The study was a content analysis of open-ended question responses in a researcher-administered survey of ED patients with both qualitative and quantitative aspects. Quantitative analysis of other survey questions has been previously reported.Reference Kestler, Buxton and Meckling 8 Descriptive qualitative tools borrowed from grounded theory informed the analysis of open-ended responses, allowing themes to emerge from the respondents’ own words.Reference Lingard, Albert and Levinson 22 Because use of qualitative tools confers some of the risks and benefits of qualitative research, we will report the methods and results according to the Standards for Reporting Qualitative Research (SRQR) format.Reference O'Brien, Harris, Beckman, Reed and Cook 23 Our research philosophy is consistent with post-positivism or pragmatism,Reference Yvonne Feilzer 24 , Reference Winit-Watjana 25 for which we acknowledge and address the subjectivity of the research team. Our design was dictated by the practical need for the THN program implementers to quickly solicit “real-world” (during an actual ED visit rather than a planned focus group) patient input that could be used for program evaluation and improvement. Because of time and space constraints of approaching candidates during their ED stay, the survey adopted sampling and analysis strategies that were predetermined and not iterative. For similar reasons, the survey did not contain follow-up questions to open-ended responses. As most open-ended responses were short, a count-based quantitative approach supplemented the use of qualitative tools. To facilitate this quantitative analysis, consensus building and inter-coder agreement were deemed desirable (contrary to qualitative approaches in which presenting interpretive disagreements between researchers might enhance the analysis). Finally, we included some quantitative analysis of previously unreported data from closed-ended questions on feeling respected in the ED and preferred locations for THN dispensing, as we felt these data were complimentary to the open-ended questions.
Context
The study took place at St. Paul’s Hospital ED in Vancouver, British Columbia. Many of the 85,000 annual patients have unstable housing, mental illnesses, and substance use disorders including injection drug use (IDU). The province of British Columbia is home to the first officially sanctioned supervised drug injection site in North AmericaReference Kerr 26 and first provincial THN program in Canada.Reference Tzemis, Al-Qutub and Amlani 19 Shortly before the study, the ED became an approved dispensing site for the provincial THN program; ED staff offered THN to at-risk patients and provided standardized THN training lasting approximately five minutes. During the study, research assistants (RAs) offered eligible patients THN (as part of the provincial program) and survey participation simultaneously. RAs were non-physicians who had no direct patient care involvement. All RAs had an ED orientation, an introduction to harm reduction principles, and an opportunity to pilot the study before beginning study work. Regular ED staff continued to train patients and dispense THN to those interested, regardless of study participation.
Sampling strategy
RAs screened candidates using triage notes and printouts for the last six months from the provincial pharmacy database. The following patients were approached: 1) self-reported opioid use at triage or during pre-hospital care; 2) clinical opioid OD or withdrawal; 3) prescriptions of methadone, buprenorphine/naloxone, or other opioids exceeding 100 morphine equivalents daily; 4) injection-related infections; and 5) narcotic requests (defined as the chief complaint, including a request for a refill of a previous opioid prescription). ED staff could refer additional study candidates to the RAs. RAs invited consecutive eligible patients to participate and excluded patients too ill, agitated, or sedated to answer study questions or undergo THN training. Those who already had a THN kit were ineligible. The rationale for selecting a target sample size of 200 was based on the previously reported quantitative outcome-based component of the study.Reference Kestler, Buxton and Meckling 8
Ethics, consent, and permissions
All participants provided written informed consent, could choose a suitably private ED interview location, and could receive THN regardless of survey participation. Age range and gender, as well as eligibility criteria that were met, were recorded without personal identifiers for study candidates who declined participation or who met the exclusion criteria. The University of British-Columbia-Providence Health Care Research Ethics Board (H14-03430) and the London School of Hygiene & Tropical Medicine (MSc Ethics 8673) approved the study.
Data collection
RAs covered pre-specified four- to eight-hour shifts in a 2:1 daytime to evening ratio including weekends from May to August 2015 and administered the survey (see Online Appendix)Reference Kestler, Buxton and Meckling 8 to consenting patients after medical stabilization. RAs asked patients about demographics, substance use, and THN-related preferences, as well as an open-ended question on the reason for accepting or refusing THN. Participants who had difficulty answering open-ended questions were provided with “prompts,” a menu of possible answers (see Online Appendix) containing themes previously cited in the harm reduction literature and by content experts including patients and addiction specialists. Patients completing the standardized 15–20 minute survey received $10 compensation. RAs entered verbatim (in the patient’s voice) responses into anonymous electronic survey forms (Surveymonkey.com, Palo Alto, CA). Responses were later transferred to Excel (version 14.7.1; Microsoft, Redmond, WA) and STATA (version 11.0; StatCorp, College Station, TX) for further analysis. During data collection, RAs regularly communicated with each other and the study leads to clarify uncertain points or conflicting data.
Data analysis
We analyzed open-ended question responses by developing a coding structure for reasons to accept or reject THN. The primary coder was an RA who analyzed the open-ended responses as part of her master’s thesis. The primary coder and one senior researcher initially independently coded the same random 20% subset of responses and then met to discuss their coding structure with a qualitative research methodologist to generate a consensus for a codebook. The primary coder applied the codebook to the complete data set, adding additional codes, as needed. A secondary coder not involved in the initial coding or interviews analyzed the data using the codebook. The two coders’ selections were compared for inter-coder reliability, with full code matching required if a response had more than one code assigned. A discussion between coders, moderated by a senior researcher, generated further consensus, leading to a measure of inter-coder agreement.Reference Campbell, Quincy, Osserman and Pedersen 27 If any disagreement between coders persisted, the senior moderator’s chosen code was used for further analysis.Reference Campbell, Quincy, Osserman and Pedersen 27 To test associations of patient characteristics with specific reasons for THN acceptance or refusal, we classified characteristics by the following pre-specified mutually exclusive subgroups: male/female, IDU/non-IDU, no fixed address (homeless)/fixed address, and older than 40/younger than 40. Reasons mentioned in 10% or more of the refusals or acceptances were tested for their association with the patient characteristics. To facilitate some quantitative analysis of the reasons mentioned by fewer than 10% of the respondents, we included these responses into larger, mutually exclusive, and binary category groupings by identifying overarching themes. We tested these categories for their association with the patient characteristics. Reasons for THN refusal were classified into the categories, absolute (not interested in THN at all) or conditional (potentially interested in THN but not desiring it during this ED visit). Reasons for THN acceptance were classified into the following categories: either mentioning a desire to help others or not mentioning this desire. To test for potential confounding, patient characteristics were checked for association with other patient characteristics. We used Fisher’s exact test for testing all associations, with a p equal to 0.05 as a threshold for significance. Responses to closed-ended questions on feeling respected in the ED and THN preferences were analyzed using descriptive statistics only.
RESULTS
We identified 417 consecutive THN candidates and offered 247 eligible patients THN and survey participation, of whom 193 (78.1%) completed the open-ended questions on THN acceptance and were analyzed (Figure 1). Participant characteristics, as well as those of patients who were excluded or refused to participate, are shown in Table 1. Interviewers used at least one prompt for approximately one-quarter of the patients. The coding process identified 10 reasons for refusal and 12 reasons for acceptance that were grouped into mutually exclusive binary response categories (Table 2). Associations between patient characteristics and types of responses are presented in Table 3. Inter-coder reliability was 74.9% (95% confidence interval [CI] 68.7–81.0%, N=191 because two originally misclassified responses were not included in the calculation), and inter-coder agreement after discussion was 95.3% (95% CI 92.3–98.3%, N=191 as above).
Study flow diagram.
Patient characteristics including those who declined study participation or were excluded
CI=confidence interval; ED=emergency department; IDU=injection drug use; THN=take-home naloxone.
* Difference among the three groups significant at a 95% level using Fisher’s exact test.
† “IDU from triage screening” refers to IDU reported in triage documents only.
‡ Prescription data obtained from print-out report of the last six months from the provincial pharmacy database.
§ Other opioid-related chief complaints included soft-tissue infection consistent with IDU, withdrawal, and opioid request.
¶ -- data unavailable for patients refusing or excluded.
ǁ n=155
Participants’ reasons codes and categories, with illustrative quotes and frequency
THN=take-home naloxone.
* Frequency sum is greater than 100% because some responses included more than one reason
† Response also counted as acceptance code 1.
‡ Response also counted as acceptance code 3.
§ Supervised drug injection site.
Association of participant’s responses with participant characteristics
IDU: Injection drug use; NFA: No fixed address (homeless)
* One participant identifying as “other” gender was not included in the analysis.
† Variables: age<40 years, IDU, and NFA were all associated with each other using Fisher’s exact test (p<0.05).
‡ p<0.05 using Fisher’s exact test.
§ Characteristics from Table 1 that were not associated with specific reasons for accepting or refusing are not shown: Indigenous, methadone or buprenorphine prescription, high-dose opioid analgesic prescription, and type of emergency department presentation.
Reasons for refusing THN
Sixty-two patients provided reasons for refusing THN (Table 3 for all results below unless otherwise mentioned). “I’m not at risk” was the most common, mentioned by 30.7% (95% CI 19.2–42.2%) of the patients refusing. IDU patients expressed this reason significantly less frequently than non-IDU patients. Some participants taking prescription opioids provided additional details regarding risk perception: “I’m just not interested. I don’t believe I will truly OD because my doc has close eye on my meds.” Other reasons for absolute refusals included “Encourages abuse,” “Would not use,” “Uncomfortable with injections,” “Done with drugs,” and passive suicidality (Table 2). When refusing THN, women endorsed “Done with drugs” three times as frequently as men that was a significant difference. Of 62 patients refusing THN, 28 (45.2% [95% CI 32.8–57.6%]) endorsed the following conditional reasons: “Poor timing,” “In a rush,” “Too sick,” and preference for another location. Patient characteristics significantly associated with conditional refusals were younger than 40 years, homelessness, and IDU status.
Reasons for accepting THN
“Save someone else’s life” was the most common acceptance reason, mentioned by 65.6% (95% CI 57.4–73.9%) of patients accepting THN, followed by “Save my own life” with 35.1% (95% CI 26.8–43.4%,), and 22.1% (95% CI 15.0–29.2%) of the patients stated both reasons. Aged less than 40 years was significantly associated with the “Save my own life” reason for accepting THN. One accepting patient also mentioned feelings of potential self-harm: “Because I’ve thought about suicide by an overdose so it would be safe to have one around in case I change my mind.” The acceptance category “for self,” encompassing all acceptance responses not specifically mentioning the desire to help others, was not associated with any particular patient characteristic (Table 3).
Preference for other THN distribution locations and feeling respected
Overall, 155 patients (including those refusing and accepting THN) answered a closed-ended question on THN location preference. Of these, 117 (75.5% [95% CI 68.6–82.3%]) preferred non-ED sites. The supervised injection site was the most frequently preferred location for THN distribution, followed by the ED, and, then, in descending order of preference, community clinics, family doctors, methadone doctors, pharmacies, and their own apartment buildings. THN location preference was not associated with specific open-ended response reasons for refusing or accepting THN, other than “Prefers other location.” Nine patients expressed “Prefers other location” as a reason for refusal in open-ended responses, although only one mentioned a specific reason (“not a hospital fan”). Of the overall study cohort, 36.8% of the patients felt inadequately respected in the ED. This lack of respect was not significantly associated with a preference for a non-ED THN location, conditional refusals, or refusing THN overall.
DISCUSSION
In this survey of 193 ED patients at risk of opioid overdose, we identified the key reasons for refusal and acceptance of THN. While understanding motivators of THN acceptance is important, understanding barriers is critical; our findings may help physicians overcome the reluctance of vulnerable ED patients to accept a life-saving intervention.
The most common reason for refusal was feeling “not at risk” for OD. Some patients expressed confidence that they would not overdose if taking opioids only as prescribed by their physician, as previously noted in patients on opioids for chronic pain.Reference Tzemis, Al-Qutub and Amlani 19 Unfortunately, patients taking prescription opioids are at an increased overdose risk, proportional to increasing opioid doses.Reference Bohnert, Valenstein and Bair 28 Conversely, our IDU patients less frequently mentioned “not at risk” as a reason for refusing THN, suggesting they might be more aware of overdose risk.
The reasons “done with drugs” and “encourages abuse” arose in our study population, mirroring concerns elsewhere that THN kits with syringes could increase risky behaviour or trigger opioid use relapses.Reference Sondhi, Ryan and Day 15 - Reference Wright, Oldham, Francis and Jones 17 Nevertheless, participation in the THN program has been associated with decreased (rather than increased) high-risk opioid use.Reference McDonald and Strang 29
Both patients accepting and refusing THN mentioned thoughts of suicide or not caring whether they overdosed. This is not surprising given the high correlation of opioid use disorder and depressionReference Goldner, Lusted, Roerecke, Rehm and Fischer 30 and previously reported associations between opioid overdoses and suicidal ideation,Reference Richer, Bertrand, Vandermeerschen and Roy 31 further highlighting the need for integration of addiction and mental health care. Although “not being comfortable with injections” has been noted in other populations,Reference Black, Connolly and Getty 16 our patients, the majority of whom injected drugs, did not mention this barrier often.
Conditional refusals are critical, as they imply that patients might be ready for THN acceptance, although not during that particular ED encounter. Conditional refusals were associated with younger patients and those experiencing homelessness, although neither characteristic was associated with a lower THN acceptance overall.Reference Kestler, Buxton and Meckling 8 Among specific reasons for conditional refusals, patients cited “too sick,” “in a rush,” and “poor timing,” all previously unreported in community settings. These reasons may not be surprising in an ED setting where patients are acutely ill and do not present specifically seeking harm reduction services. Some previously identified barriers to accepting THN were not directly mentioned by our study population, notably fear of interacting with police and drug-related stigma.Reference Black, Connolly and Getty 16 , Reference Tzemis, Al-Qutub and Amlani 19
The most common motivator for THN acceptance was helping others, reflected in responses about saving and educating others. This desire is also common in non-ED settings.Reference Wright, Oldham, Francis and Jones 17 , Reference Tzemis, Al-Qutub and Amlani 19 , Reference Wagner, Davidson and Iverson 20 Wanting to protect oneself from an OD or to generally “feel safer” corroborates with a previous quantitative analysis involving our study cohort; those concerned about their own OD risk were more likely to accept THN.Reference Kestler, Buxton and Meckling 8 One participant mentioned peer influence, a strong driver of THN acceptance elsewhere.Reference Wagner, Iverson and Wong 21 The “feels good” reason mentioned by another is reminiscent and echoes the feeling of personal empowerment cited by other THN research.Reference Tzemis, Al-Qutub and Amlani 19 , Reference Wagner, Davidson and Iverson 20
The THN acceptance reasons, “did not know where to get one,” “willing to try,” “never had one,” “lost other kit,” and “supervised injection site crowded,” highlight the convenience of ED dispensing. Prior analysis of our study population indicated that 80% of the patients felt the ED was indeed a convenient setting for THN.Reference Kestler, Buxton and Meckling 8 Taken together, these findings further support the concept of the ED as a 24-hour location for patients who might not otherwise have easy THN access.
Preferences for THN dispensing sites have not been previously reported. Even among accepting patients, the supervised injection site of the community was the preferred location. Many factors may play into this preference, including the dislike of hospitals mentioned by one participant. Nevertheless, a lack of feeling respect in the ED, previously noted in another marginalized ED population,Reference McCormack, Hoffman, Norman, Goldfrank and Norman 32 did not appear to play a significant role in patients’ THN location preferences or their reasons for refusing THN.
Importantly, our findings may encourage ED THN programs to adopt strategies to enhance uptake. ED staff should gently emphasize true OD risk in a non-judgmental manner. Second, staff should empower patients by encouraging THN acceptance to help other at-risk peers. For patient convenience, THN needs can be addressed early in the ED stay in parallel with medical management. Even if patients refuse THN in the ED, providers should emphasize the importance of THN kits and encourage patients to discuss it in the community with family, peers, and trusted care providers. The presence of a dedicated nurse, a social worker, or another care provider specially trained in addiction treatment in the ED would likely enhance patient receptivity to both THN and other forms of addiction care and harm reduction.
LIMITATIONS
Our findings at a single inner-city centre with a high proportion of IDU may be challenging to extrapolate to other settings, but the themes found in our population are congruent with those in non-ED environments. The analysis of open-ended question responses is necessarily limited because of a sampling strategy based on the quantitative design of the survey. Traditional qualitative approaches such as iterative interviewing and theory development could have helped clarify responses, explore additional reasons, and assess theme saturation. Social desirability bias might have occurred in the context of compensation for participation, although the candour of responses suggests that such bias is minimal. A lack of interview recording could have allowed RAs to inject their own words into transcriptions. Interviewer prompting could have directed participants’ responses in some cases by providing a quick and easy answer rather than one in an authentic patient voice. Some subgroups, especially those involving patients refusing THN, are small and could draw spurious associations or fail to identify correct ones.
CONCLUSION
ED patients refusing THN either did not feel at risk of OD or did not feel their ED encounter was the right time or place for THN. Most ED patients accepting THN wanted to save others from an OD.
Acknowledgements
We gratefully acknowledge Alexis Crabtree, MD, MPH, PhD, post-doctoral student at the British Columbia Centre for Disease Control, for her critical revisions of the final manuscript.
Competing interests
None declared.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/cem.2018.368
