Factors Associated with Patient Perceived Suboptimal Dosing of In-hospital Opioid Agonist Therapy Among People Who Use Illicit Drugs in Vancouver, Canada

Abstract

Background:

Optimal dosing of opioid agonist therapy (OAT) is essential for treatment success. However, initiation and maintenance of OAT in hospital settings can be challenging given differing levels of opioid tolerance, withdrawal, and intoxication among patients.

Objective:

The objective of this study was to characterize the prevalence and factors associated with in-hospital patient perceived suboptimal OAT dosing among people who use illicit drugs (PWUD) in Vancouver, Canada.

Methods:

Data were derived from three prospective cohorts of PWUD in Vancouver, Canada. Bivariable and multivariable logistic regression models were used to examine factors associated with patient perceived suboptimal in-hospital OAT dose.

Results:

273 study participants were prescribed OAT while in hospital: 83 (30.4%) participants perceived their OAT dose to be suboptimal. In a multivariable model, factors positively associated with a perceived suboptimal OAT dose included: homelessness (adjusted odds ratio [AOR] = 2.85; 95% CI: 1.53 – 5.28), daily stimulant use (AOR = 2.03; 95% CI: 1.14 – 3.63) and illicit drug use while in hospital (AOR = 2.33; 95% CI: 1.31 – 4.16).

Conclusions:

Almost one third of participants perceived receiving a suboptimal OAT dose while in hospital. These observed correlations indicate that a patient’s perception of suboptimal OAT dosing in hospital may be more prevalent for patients who are homeless, report polysubstance use with stimulants and opioids and who obtain illicit drugs while hospitalized. While cautious prescribing of OAT in patients experiencing hospitalization is important, these findings demonstrate a high prevalence of and apparent risk factors for perceived suboptimal OAT dosing.

Introduction

Long-term opioid agonist therapy (OAT) has been shown to significantly reduce rates of relapse, morbidity and mortality related to opioid use disorder (OUD).^1–5 Hospitals have been identified as an important setting for pharmacological OAT intervention given that individuals with an OUD regularly frequent acute care settings.^6–8 However, initiation and maintenance of OAT in hospitals may pose challenges for both the patient and physician given the need to individually tailor doses to minimize subjective withdrawal, in the context of (often) concurrent pain and/or intoxication.^8–10 OAT also warrants cautious dosing and titration in patients who are clinically unstable and in those using alcohol or benzodiazepines concurrently, which can significantly delay optimization of opioid-related symptoms.^8–10 Moreover, challenges remain regarding transitions in treatment and care pre-admission and post-discharge, which can compromise the continuity of OUD care.^3,11–15 As per Canadian and United States guidelines methadone dosing is typically initiated at 10–30 mg and titrated to a dose of 60–120 mg daily. Buprenorphine-Naloxone dosing is typically initiated at a 2–4 mg dose and increased to 8–24 mg daily dosing. Achieving an optimal stable dose is crucial to reducing withdrawal symptoms. ^10,16 Logistical challenges such as coordinating transitions in care between community and hospital settings may lead to delays in starting OAT while in hospital and subsequently, lower doses than necessary being initiated.¹⁷ Health care provider-level barriers such as concerns about patient compliance, drug diversion, misuse and overall stigmatization of individuals with OUD are well documented and may also contribute to suboptimal dosing.^17–20

The aforementioned studies have explored and identified some of the challenges associated with successful inpatient induction and long-term maintenance of OAT in hospital settings. Suboptimal in-hospital OAT in the context of pain and withdrawal treatment has been associated with increased risk of using drugs while in hospital or leaving against medical advice, both of which are associated with increased risk of overdose, disease transmission and mortality. ^21,22 While prescribers may modify doses of OAT based on subjective factors such as how intense they believe the patient’s withdrawal or level of pain to be, patients with more comorbid pain, psychiatric disease or more intense withdrawal may be more likely to perceive their doses as suboptimal. ^9,23,24 Accordingly, the objective of this study was to identify factors associated with patient perceived suboptimal OAT dosing in a hospital setting among people who use illicit drugs (PWUD) in Vancouver, Canada.

Methods

Study Overview and Population:

We used data originating from three prospective cohort studies in Vancouver, Canada: The AIDS Care Cohort to evaluate Exposure to Survival Services (ACCESS), The Vancouver Injection Drug Users Study (VIDUS) and the At-Risk Youth Study (ARYS). Specific details of these cohorts have been discussed elsewhere.^25–28 Participants in all three cohorts are recruited via word of mouth, street outreach and self-referral. The ACCESS cohort consists of HIV-infected adults (i.e. ≥18 years of age) who have used illicit drugs other than or in addition to cannabis (which was an illicit substance during the study period) in the month prior to enrollment.^26,29,30 Individuals enrolled in the VIDUS study must be HIV-negative adults who have injected illicit drugs in the past month prior to enrollment in the study.^26,29,30 The ARYS cohort enrolls street-involved youth age 14 to 26 years that have used drugs other than cannabis in the month prior to enrolment.³¹ Individuals in the cohorts must have resided in the greater Vancouver area at the time of enrollment and have provided written informed consent to participate in the study. Participants in all three cohorts were administered a harmonized baseline questionnaire at their initial visit. Following this, participants complete follow-up questionnaires on a semi-annual basis. These comprehensive questionnaires elicit a range of information, including socio-demographics, drug use patterns, drug-related treatments, and access to healthcare services. The ACCESS (Vancouver, June 2016, #H05–50233), VIDUS (Vancouver, June 2016, #H14–01396) and ARYS (Vancouver, June 2016, #H04–50160) studies have all been granted approval from the Providence Health Care and University of British Columbia Research Ethics Board.

Study Sample Inclusion Criteria:

We restricted the study sample to individuals who reported having been hospitalized in the last six months and having received OAT while in hospital during that time. The study period was restricted between June 2016 to May 2018 based on the availability of questionnaire data. The majority of participants (79.1%) only contributed one study visit during the entire study period; thus, we applied a cross-sectional study design. For participants who had more than one study visit, we used the most recent study visit for each participant.

Measures:

The main outcome measure was suboptimal in-hospital OAT dose in the last six months as perceived by the participant. This was measured by their response (yes versus no) to the question: ‘the last time you were prescribed Methadone or other opioid substitution therapy while in hospital, was your dose high enough?’. We subsequently considered covariates that we hypothesised may be associated with the main outcome measure, including: sex (female vs. male); age at study visit reporting in-hospital OAT (per one-year increase); White ethnicity (yes vs. no); HIV-positive; homelessness (yes vs. no); OAT dosing interrupted > 24 hours in last 6 months (yes vs. no); daily heroin use (≥ daily vs. < daily); daily illicit prescription opioid use (≥ daily vs. < daily); daily injection drug use (≥ daily vs. < daily); heavy alcohol use defined as >4 drinks on any day for men or >3 drinks on any day for women (yes vs. no); benzodiazepine use (≥ daily vs. < daily); stimulant use (≥ daily vs. < daily); drug overdose (yes vs. no); illicit drug use while in hospital (yes vs. no); pain scale average over the past week (per one unit increase on the Brief Pain Inventory, where 0 is no pain and 10 is the worst possible pain); and mental illness diagnosis in the last six months (yes vs. no). We considered having a history of being on OAT as a potential covariate; however, upon further analysis, we decided to exclude this from regression modelling given that the counts were too low (only 6 (2.2%) participants had never been on OAT before, and only 1 of those 6 felt that their OAT dose was not high enough (1.2%). Unless otherwise indicated, all variables were dichotomized as yes vs. no and referred to the six months prior to interview.

Statistical Analyses:

To begin, we assessed descriptive characteristics of the study sample, stratified by whether participants perceived to have received suboptimal OAT dose or not. We also examined what medical conditions participants were hospitalized for. We constructed bivariable and multivariable logistic regression models to identify which factors were independently associated with perceived suboptimal in-hospital OAT. We applied an Akaike Information Criterion (AIC) protocol, where we calculated and noted the AIC for a full model that included all variables significant at P<0.10 in bivariable analyses. Next, we removed the variable with the largest P-value and re-calculated the AIC for this reduced model. This iterative process was continued until no variables remained. We selected the model with the lowest AIC score. All P-values were two sided.

As a secondary analysis, we calculated the frequency and proportion of participants who used street drugs while in hospital specifically because their OAT dose was not high enough. This was examined by the following question: “Did you use street drugs in hospital in the last six months? Why? Which drugs did use?”. Additionally, among those who used street drugs while in hospital, we examined the frequency and proportion of different types of drugs used in hospital.

Results

Study Sample Characteristics:

In total, 277 PWUD received OAT while in hospital and were eligible for the present analysis; four (1.4%) participants were excluded due to missing outcome data. 144 (52.7%) of the 273 included participants were female and the median age of the sample at the study visit when they reported in-hospital OAT was 45.7 years (quartile[Q]1 – Q3: 33.8 – 53.3). The baseline characteristics of the included 273 participants in the cohorts are presented in Table 1.

TABLE 1.

Characteristics of Study Sample (n = 273)

Characteristic	Total (%) (n = 273)	Outcome Variable
		Perceived Suboptimal OAT dose (30.4%) (n = 83)	Perceived Optimal OAT dose (69.6%) (n = 190)
Sex
Male	144 (52.7)	38 (45.8)	106 (55.8)
Female	129 (47.3)	45 (54.2)	84 (44.2)
Age at baseline (median, Q1-Q3)	45.7 (33.8–53.3)	40.8 (31.1–49.8)	47.4 (34.6–54.0)
White Ethnicity
Yes	137 (50.2)	42 (50.6)	95 (50.0)
No	135 (49.5)	41 (49.4)	94 (49.5)
HIV Positive
Yes	106 (38.8)	30 (36.1)	76 (40.0)
No	166 (60.8)	52 (62.7)	114 (60.0)
Homelessness *
Yes	63 (23.1)	33 (39.8)	30 (15.8)
No	206 (75.5)	48 (57.8)	158 (83.2)
Missed OAT Dose *
Yes	80 (29.3)	19 (22.9)	61 (32.1)
No	192 (70.3)	64 (77.1)	128 (67.4)
Heroin Use *
≥daily	106 (38.8)	44 (53.0)	62 (32.6)
<daily	167 (61.2)	39 (47.0)	128 (67.4)
Prescription Opioid Use *
≥daily	8 (2.9)	3 (3.6)	5 (2.6)
<daily	265 (97.1)	80 (96.4)	185 (97.4)
Injection Drug Use *
≥daily	128 (46.9)	52 (62.7)	76 (40.0)
<daily	145 (53.1)	31 (37.3)	114 (60.0)
Heavy Alcohol Use *
Yes	34 (12.5)	8 (9.6)	26 (13.7)
No	239 (87.5)	75 (90.4)	164 (86.3)
Benzodiazepine Use *
≥daily	3 (1.1)	2 (2.4)	1 (0.5)
<daily	270 (98.9)	81 (97.6)	189 (99.5)
Stimulant Use *
≥daily	85 (31.1)	37 (44.6)	48 (25.3)
<daily	188 (68.9)	46 (55.4)	142 (74.7)
Non-fatal Overdose *
Yes	69 (25.3)	29 (34.9)	40 (21.1)
No	203 (74.4)	53 (63.9)	150 (78.9)
Hospital Drug Use *
Yes	89 (32.6)	42 (50.6)	47 (24.7)
No	184 (67.4)	41 (49.4)	143 (75.3)
Mental Illness Diagnosis *
Yes	15 (5.5)	7 (8.4)	8 (4.2)
No	232 (85.0)	63 (75.9)	169 (88.9)
Pain Scale Average (median, Q1-Q3)	3.0 (0.0–6.0)	4.0 (0.0–7.0)	2.0 (0.0–6.0)

Q: Quartile

^*Indicates behaviors/activities in the six months prior to the interview

Among the sample of 273 participants, 221(80.9%) reported receiving methadone in the past 6 months and 24 (8.8%) reported using buprenorphine. 28 (10.3%) participants reported receiving alternative forms of OAT including injectable opioids or slow-release oral morphine in the last 6 months. Among the 106 participants who reported being HIV positive, 73 reported being on ART in the last 6 months (68.9%).

The five leading causes of hospital admission were pneumonia (16.1%), mental health complications (9.9%), surgery (9.5%), abscess (7.0%) and overdose (5.1%). Seventy-three (26.7%) participants accessed the hospital once every six months, over half of the participants (53.9%) accessed the hospital once every three months, and 15 (5.5%) participants accessed the hospital at least once every month.

Factors associated with suboptimal in-hospital OAT dose:

Bivariable analysis of factors associated with suboptimal OAT are presented in Table 2. Factors positively associated with the outcome included: homelessness (odds ratio [OR] = 3.62; 95% confidence interval [CI]: 2.01 – 6.54), ≥daily heroin use (OR = 2.33; 95% CI: 1.38 – 3.95), ≥daily injection drug use (OR = 2.52; 95% CI: 1.48 – 4.28), ≥stimulant use (OR = 2.38; 95% CI: 1.38 – 4.09), drug overdose (OR = 2.05; 95% CI: 1.16 – 3.63) and illicit drug use while in hospital (OR = 3.12; 95% CI: 1.81 – 5.36), whereas older age (OR = 0.98; 95% CI: 0.95 – 1.00) was negatively associated with the outcome.

TABLE 2.

Bivariable and multivariable logistic regression modeling of factors associated with perceived suboptimal OAT (n = 273)

Characteristic	Unadjusted OR (95% CI)	P-value	Adjusted OR (95% CI)	P-value
Sex (Male vs. Female)	0.67 (0.40 – 1.12)	0.129
Age at baseline (per one-year increase)	0.98 (0.95 – 1.00)	0.018*
White Ethnicity (yes vs. no)	1.01 (0.60 – 1.70)	0.959
HIV Positive (yes vs. no)	0.87 (0.51 – 1.48)	0.596
Homeless (yes vs. no) *	3.62 (2.01 – 6.54)	<0.001*	2.85 (1.53 – 5.28)	< 0.001*
Missed OAT Dose (yes vs. no) *	0.62 (0.34 – 1.13)	0.120
Heroin Use (≥ daily vs. < daily) *	2.33 (1.38 – 3.95)	0.002*
Prescription Opioid Use (≥ daily vs. < daily) *	1.39 (0.32 – 5.95)	0.659
Injection Drug Use (≥ daily vs. < daily) *	2.52 (1.48 – 4.28)	<0.001*
Heavy alcohol use (yes vs. no) *	0.67 (0.29, 1.56)	0.354
Benzodiazepine Use (≥ daily vs. < daily) *	4.67 (0.42, 52.19)	0.211
Stimulant Use (≥ daily vs. < daily) *	2.38 (1.38, 4.09)	0.002	2.03 (1.14 – 3.63)	0.017*
Drug Overdose (yes vs. no) *	2.05 (1.16 – 3.63)	0.014*
Hospital Drug Use (yes vs. no) *	3.12 (1.81 – 5.36)	<0.001*	2.33 (1.31 – 4.16)	0.004*
Mental Illness Diagnosis (yes vs. no) *	2.35 (0.82 – 6.74)	0.113
Pain Scale Average (per one-unit increase)	1.05 (0.97 – 1.13)	0.236

CI: confidence interval

^*Indicates behaviors/activities in the six months prior to the interview

In multivariable analysis, as specified in Table 2, factors that remained positively associated with perceived suboptimal OAT while in hospital after adjusting for other covariates included: homelessness (adjusted odds ratio [AOR] = 2.85; 95% CI: 1.53 – 5.28), stimulant use (AOR = 2.03; 95% CI: 1.14 – 3.63) and in-hospital illicit drug use (AOR = 2.33; 95% CI: 1.31 – 4.16).

Secondary Analyses:

In secondary analyses, we found that 89 (32.6%) of the 273 participants in the sample used illicit drugs while in hospital. Among those, 15 participants (16.9%) reported using street drugs while in hospital because their OAT dose was not high enough. Among the participants who reported using street drugs while in hospital, the top five drugs used were: heroin (66; 74.2%); crystal methamphetamine (16; 18.0%); crack cocaine (12; 13.5%); prescription opioids (8; 9.0%); and cocaine (6; 6.7%).

Discussion

In the present study, almost one-third of PWUD in the study sample that had received OAT while in hospital perceived their dose to be suboptimal. Multiple factors were found to be positively associated with perceived receipt of suboptimal OAT, including homelessness, frequent stimulant use, and drug use in hospital.

Although literature studying subjective ratings of in-hospital OAT efficacy is scarce, an available study suggests that among 55 patients who received OAT in hospital, all 55 reported significant improvements in withdrawal symptoms during the induction process based on subjective and objective opiate (SOWS and COWS) withdrawal scales. ⁶ Another study reported significant mean improvements in withdrawal symptoms during in-hospital OAT induction using the Clinical Institute Narcotic Assessment (CINA) scoring system which is based on objective and subjective withdrawal criteria. ¹⁴ In-hospital OAT induction/maintenance was shown to improve outpatient OAT continuation compared to 4-day detox with buprenorphine alone. ^13,14 Despite this, our study’s findings indicate that almost one-third of participants prescribed OAT in hospital perceived their dosing to be suboptimal. This may highlight key differences between patient’s subjective rating of OAT dosing compared to subjective/objective withdrawal scales used in hospital settings to assess the adequacy of OAT induction. Future studies should seek to examine the difference between these perspectives when evaluating the success of in-hospital OAT induction.

The intersection between substance use and homelessness has shown to impact a variety of financial, medical and social difficulties experienced by PWUD.^32,33 In the present study, we found a positive association between homelessness and perceived receipt of suboptimal OAT while in hospital. There are many possible explanations for why PWUD experiencing homelessness may be vulnerable to perceiving suboptimal OAT while in hospital. It may be that these individuals represent a subset of patients with more severe OUD despite the present study’s attempt to control for a number of opioid use behaviours. Subsequently, higher tolerance, more severe withdrawal and a lengthier OAT induction process in-hospital may explain why PWUD experiencing homelessness were more likely to perceive their OAT dose as inadequate.^23,24,34,35 Additionally, compared to their housed counterparts, homeless PWUD may experience higher rates of comorbid substance use, psychiatric complications, physical health complications, overdose related mortality and stigmatization from healthcare providers, which may have resulted in delays in OAT initiation or more cautious induction dosing.^{1,10,42–45,32,33,36–41}

We also found a positive association between daily stimulant use and perceiving receipt of suboptimal OAT dosing in-hospital. People who frequently use stimulants and engage in polysubstance use represent a group of individuals with a more severe form of substance dependence. ^46,47 Studies have shown that polysubstance use is associated with increased risk of drug overdose and decreased likelihood of overdose training. ^48–50 In the context of OAT dosing, some physicians may have undertreated patients’ pain or withdrawal due to biases such as fear of drug misuse/diversion, overdose and not believing the patients stated level of withdrawal or pain.^18–20 Due to the short half-life of stimulants and the potential need to frequently leave the hospital to maintain active drug use, OAT induction and maintenance among those who frequently use stimulants can be challenging. ^51,52 Our results highlight the concurrent challenges associated with optimizing OAT among this population of high intensity PWUD, as these individuals may be at increased risk of experiencing stimulant withdrawal symptoms and overdose due to polysubstance use.^23,46,50

Finally, we found a positive association between illicit drug use while in hospital and patients perceiving a suboptimal OAT dose. While temporality cannot be assessed due to the nature of the study, it may be that health care providers are reluctant to prescribe higher doses of OAT if they observe or suspect the use of street drugs while in hospital given potential drug interactions, contraindications and risk for overdose.^1,10 Alternatively, these individuals may use street drugs while in hospital as a means of maintaining their ongoing substance dependence or self-medicating as a result of inadequate OAT dosing and a high level of withdrawal. Our secondary analyses support our primary findings showing that among the participants who used street drugs while in hospital, a substantial proportion reported that this was specifically because their OAT dose was not high enough, and that heroin was the most common drug used while in hospital settings. This emphasizes the need for closer attention to adequate dosing during the OAT induction period.

Our study has limitations. First, our study is observational in nature and despite controlling for variety of confounders upon multivariable analysis, there may be unmeasured confounders that we have failed to account for. As a result, we are unable to establish any causal relationships. Moreover, our main outcome was having perceived a suboptimal OAT dose while in hospital; no information on what the participant was actually prescribed was recorded. Given that this self-reported data was not linked to administrative data, we were also unable to validate participants perceptions of suboptimal dosing with objective measures such as the Clinical Opiate Withdrawal Scale or admission length. Additionally, there may be unmeasured confounding as non-reported variables such as the altered metabolic demands of pregnancy and various drug-drug interactions that alter opioid pharmacokinetics, and thereby the adequacy of dosing, were not considered in the questionnaire. ^53,54 Finally, our findings may not be generalizable to all PWUD patients receiving in-hospital OAT in our study settings or other regions with varying demographics, healthcare systems and OAT prescribing patterns. Despite these limitations this study still provides unique and exploratory insights into reportable factors that affect perceived adequacy of in-hospital OAT. These limitations also highlight future avenues for research that would involve linking questionnaire, admission, and prescription databases to correlate patient perceived factors with more objective measures as described above.

In summary, one third of our study sample perceived their in-hospital OAT dose to be suboptimal. We found a positive association between patient perceived suboptimal dosing of OAT in-hospital and homelessness, frequent stimulant drug use and illicit drug use while in hospital. This study highlights the need to better understand the barriers to optimizing OAT prescribing in hospital settings for a subpopulation of PWUD who may be more dependent on opioids or be at greater risk for opioid overdose but may be predisposed to receiving suboptimal therapy while in hospital.