Impact of prehospital olanzapine for methamphetamine intoxication on patient and system-level outcomes: a population-based retrospective cohort study

Abstract

Introduction

Acute methamphetamine intoxication is a common presentation in emergency departments (EDs). Empowering paramedics to treat methamphetamine intoxication in the prehospital setting could help to improve patient and system outcomes. This study evaluated the impact of a 2018 policy allowing paramedics to administer olanzapine for methamphetamine intoxication in the prehospital setting on patient and system outcomes.

Methods

Retrospective pre-post analysis of population-based health administrative data and qualitative paramedic reports. Individuals with a methamphetamine-related health system contact between 2016–2020 in Winnipeg, Canada, who were eligible for pre-hospital olanzapine were included. The exposure was the administration of olanzapine in the pre-hospital setting by paramedics. Individuals who received prehospital olanzapine were compared with those who did not, prior to (2016–2018) and after (2018–2020) the policy change. Primary outcomes included physical or chemical restraint prior to ED, transport to ED and leaving from the ED prior to being seen or against medical advice. Outcomes were modelled using logistic (binary outcomes) and Cox proportional hazard (temporal outcomes) models adjusted for covariates using inverse probability of treatment weighting.

Results

The cohort included 1041 individuals, of which 222 received the intervention. Among individuals receiving olanzapine, the median age was 31 years, 58.1% were male, 56.8% were in the lowest income quintile and a high prevalence of mental health disorders and legal system involvement were observed. After adjusting for covariates, individuals receiving olanzapine in the prehospital setting were more likely to be transported to ED by paramedics (adjusted OR (aOR): 11.83, 95% CI 3.19 to 43.80) and less likely to leave the ED without being seen by a doctor (aOR: 0.56, 95% CI 0.32 to 0.97) compared with individuals who did not.

Conclusions

Prehospital administration of olanzapine among individuals experiencing methamphetamine intoxication was associated with a higher likelihood of transport to the ED and being seen by a physician.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The use of linked paramedic field reports and population-level health services data is unique and offers insights into clinical trajectories from the community through to the hospital.

• The inclusion of both patient and health-system-level outcomes provides a comprehensive picture of the impact of prehospital olanzapine administration on both patients and providers.

• The use of health administrative data does not capture the more nuanced subjective experiences of individuals who require care for methamphetamine intoxication.

• The quantitative and retrospective nature of the study design means that there may be differences between those do and do not receive olanzapine, which could bias the results. Efforts were made to control for this potential bias in the study design.

Introduction

Methamphetamine use is a significant and growing public health concern. Rates of methamphetamine use are rising across many jurisdictions, as have associated rates of emergency department (ED) visits, hospital admissions and mortality.^1–10 Psychiatric diagnoses frequently accompany methamphetamine use, and presenting complaints range from anxiety and agitation to psychosis and aggression, which comes with the potential for violence and need for restraint.^{4 7 8 11} Cases of methamphetamine intoxication are further associated with varied and serious medical complications,^{9 12} high rates of repeat visits to EDs¹³ and wider reliance on social services compared with use of other substances.¹⁴ In many cases, these outcomes are disproportionately experienced by marginalised population groups, such as those experiencing extreme poverty, mental illness and housing instability.^{3 4 11–13}

Management of acute methamphetamine intoxication in the ED primarily aims to reduce immediate symptoms. General treatment relies on benzodiazepines (such as midazolam or lorazepam) and antipsychotic medications (such as haloperidol, olanzapine or risperidone), but exact medications, doses and indications vary by jurisdiction.^{5 12 15–20} To date, few studies have comprehensively evaluated different treatment options; among published analyses, several compare medications in terms of symptom resolution and show qualified support for olanzapine either alone or in combination for treating methamphetamine-induced agitation or psychosis.^17–20 Other outcomes, such as ED length of stay, restraint use and hospital admissions, have not been consistently compared across different regions and patient groups.¹⁹ Despite general agreement on broad approaches to managing methamphetamine intoxication, treatment for optimal patient and system outcomes is still unclear.

A limited body of recent research has described the impact of acute methamphetamine intoxication on emergency medical services. Similar to what is seen in the ED, rates of prehospital methamphetamine cases are rising across many jurisdictions, and individual encounters are frequently complex, challenging and may require police or additional personnel for safety.^21–23 In contrast to cases of opioid use, where treatment with naloxone is well established and additional treatment options and referral pathways are rapidly evolving,^{24 25} there are limited treatments for methamphetamine intoxication in the prehospital setting beyond general protocols for patient agitation.^26–28 In the mid-size city of Winnipeg, Canada, where increasing methamphetamine use and the uncertainty surrounding optimal treatment of acute methamphetamine intoxication have placed growing strain on EDs and emergency medical services, local healthcare leaders sought to provide a prehospital treatment option that would address both patient care and system pressures.

In December 2018, paramedics in Winnipeg were authorised to administer olanzapine in cases of methamphetamine intoxication. The rationale for this policy change was that paramedics are often the first to encounter individuals experiencing the negative effects of methamphetamine intoxication, such as agitation, and this would provide a tool for paramedics to mitigate the risk of harms to patients, care providers and community members exposed to these situations in the prehospital setting. While this was a novel policy change in Canada, similar approaches had previously been employed in Australia.²⁹ Our team investigated the effectiveness of this intervention across a range of patient- and system-oriented outcomes as part of a larger study on reducing methamphetamine-related harms and improving access to services.³⁰ While a range of outcomes were considered, the primary goal of this analysis was to evaluate the impact of this policy on patient care variables such as the need for chemical and physical restraint, transportation to the ED and being seen by a physician in the ED.

Methods

Study design

We investigated the effectiveness of prehospital administration of olanzapine (10 mg, oral disintegrating tablet) by paramedics with the Winnipeg Fire Paramedic Service (WFPS) for individuals experiencing amphetamine-type toxicity. This is worded as ‘amphetamine-type toxicity’ in the policy documentation (online supplemental file 1) to provide paramedics the ability to treat the general toxidrome of stimulant intoxication with olanzapine. That said, almost all these cases relate to methamphetamine use, specifically, so for clarity we use the term ‘methamphetamine intoxication’ throughout this manuscript. WFPS implemented olanzapine for this use on 8 December 2018. The pre-intervention period was 1 January 2016 – 15 November 2018, and the post-intervention period 15 December 2018–31 December 2020. Data collected during the 2 weeks before and after implementation were censored. Among patients eligible to receive olanzapine (based on a clinical algorithm used by paramedics, see online supplemental file 1) we compared treated and untreated individuals before and after the implementation of the intervention. This included two comparisons: (1) between individuals who did versus did not receive olanzapine in the post-intervention period and (2) between individuals who received olanzapine in the post-intervention period and individuals treated in the pre-intervention period. The first comparison aimed to provide an estimate of the association between olanzapine administration and patient outcomes in the post-intervention period whereas the second comparison aimed to compare outcomes among patients who received prehospital olanzapine relative to patients managed under the previous standard of care, when prehospital olanzapine was not available.

Data sources

The study used whole-population administrative health and social data from the Manitoba Population Research Data Repository housed at the University of Manitoba.³¹ The Repository contains more than 100 datasets of individual-level records created during the routine administration of a range of provincial government services, including health, justice and social programmes. All individual data records are de-identified and linkable across all datasets. A dataset from the WFPS was a key asset for this study; it contains written narratives of paramedic encounters in the prehospital setting. All other datasets used in this study are detailed in online supplemental file 2).

Cohort construction

Figure 1 displays the cohort construction. This study included all Winnipeg residents who had a health system record related to methamphetamine use from 1 January 2011–31 December 2020. We identified each individual’s first methamphetamine-related health system contact within the study period (2016–2020). This contact was defined as the ‘index date’ for that individual. We excluded individuals with a record of a methamphetamine-related health system contact in the 5 years prior to the index date (to isolate individuals with a first methamphetamine-related health system contact), those who did not have continuous healthcare coverage in the 5 years prior to the index date (required to assess health system usage), those who did not live in Winnipeg continuously in the year prior to the index date (required to assess health system usage) and those who only had a methamphetamine-related health system contact during the washout period (as described above).

Figure 1. Note: Patients not eligible to receive olanzapine were defined by the protocol outlined in online supplemental file 1. Those ineligible include those with: no acute intoxication; use beyond 48 hours; acute medical complaints or injuries; severe agitation. WFPS, Winnipeg Fire Paramedic Service.

Finally, we also excluded patients who would not have been eligible to receive olanzapine based on the WFPS criteria. Two experienced paramedics reviewed selections of the linked written narrative of a methamphetamine-related contact to determine whether the case would have been eligible for olanzapine. This was done in an independent fashion, whereby the reviewers were blinded to each other’s adjudications, the intervention stage (pre-post), and receipt of olanzapine. Cases of disagreement were resolved by an experienced paramedic educator. Eligibility criteria are specified in an established protocol (online supplemental file 1), which states that patients may be offered olanzapine if they have used methamphetamine within the last 48 hours and either have a mental health complaint or are showing mild or moderate agitation. This protocol does allow for some subjective judgement given that it is not always possible to establish the exact timeframe of methamphetamine use; however, the 48-hour benchmark serves as a guide for paramedics to apply the protocol in the field according to their interpretation of clinical presentation and information from patients and bystanders. The protocol excludes patients with other acute medical complaints or who are severely agitated.

Once eligibility was established in the WFPS narratives, we extracted final counts for each group: pre-intervention (n=578), post-intervention treated with olanzapine (n=222) and post-intervention not treated (n=241). Of note, the many individuals excluded at this stage reflect the specific protocol criteria applied to undifferentiated methamphetamine-related health system contacts: no acute intoxication, use beyond 48 hours, acute medical complaints or injuries, or severe agitation.

Variables

As adjustment variables for statistical models, we used the following pre-defined covariates as part of a calculation of a high-dimensional propensity score (HDPS): a diagnosis of psychotic, personality, mood/anxiety or non-amphetamine substance use disorder in the 5 years prior to first olanzapine-eligible health system contact (see Nickel et al for a detailed description of how these diagnoses were coded³²; criminal charges for violent or non-violent incidents in the year prior to contact; receipt of Employment Income Assistance (welfare payments) in the year prior to contact; total Elixhauser comorbidity group diagnoses in the year prior to contact; neighbourhood income quintile and Socioeconomic Factor Index – Version 2 (SEFI-2) score; sex; age at contact; and First Nations or Red River Métis identity. The HDPS also draws covariates from health databases such as medical service tariff codes, physician diagnostic codes, hospital procedure codes and hospital diagnostic codes to describe the health status of the study cohort. The HDPS calculation automatically selected the top 400 health-status related variables in the year before contact, ranked by their association with the outcome variable under study.^{33 34} After truncation at the 1.5th and 98.5th percentiles, the propensity scores were used to calculate Inverse-Probability-of-Treatment Weights (IPTWs). Please see the online supplemental methods for more information on the HPDS and IPTW procedures.

Outcomes

10 outcome measures were selected based on available data in the Repository and derived in consultation with clinicians from relevant specialities (paramedicine, emergency medicine and psychiatry). Hypothesising that treatment would reduce patient agitation overall, we examined both ‘patient intervention’ and ‘ED process’ outcomes in prehospital and ED settings. The primary outcomes of interest were the four patient intervention outcomes, including documented physical restraint (cloth restraints or handcuffs) or chemical restraint (intramuscular or intravenous haloperidol and/or midazolam); transport to ED (compared with refusing transport); and leaving the ED prior to being seen or against medical advice. Secondary outcomes were six ED process outcomes including high-acuity (compared with low-acuity) ED bed assignment; hospital admission; and time within the ED (registration to triage, triage to bed, bed to discharge and bed to admission/transfer).

Statistical analysis

We used SAS V.9.4³⁵ for all analyses and data management. We produced a descriptive analysis (counts, percentages) of health and sociodemographic characteristics for each study group. For additional context, these characteristics were also calculated for Winnipeg residents with no methamphetamine-related record. We examined the distribution of outcomes across the three cohort groups by calculating crude proportions and median wait times with 95% CIs for all 10 outcomes. The patient-level outcome variables, the high-acuity ED bed assignment variable and hospital admission were all binary; we used log-binomial generalised linear models with treatment status as the independent variable to calculate proportions and 95% CI. For the remaining outcomes, we used a quantile regression with the same independent variable to estimate median wait times.

In the main analysis, we compared the post-intervention treated group (patients who received prehospital olanzapine) to pre- and post-intervention untreated groups using a series of statistical models, producing crude and IPTW-adjusted ORs and HRs with 95% CI. For the binary outcomes, we used a logistic regression to estimate ORs. Because of the small cohort size and sometimes rare outcome events, we used Firth’s penalised maximum likelihood estimation to reduce bias in the parameter estimates. For the ED wait time outcomes, we conducted time-to-event analyses using Cox proportional hazard regressions to estimate HRs.

Results

Table 1 presents descriptive characteristics for the three study groups and for Winnipeg residents without any record of a methamphetamine-related health system contact. The median age of individuals in the study groups was 29–33 years and the groups were roughly evenly divided between male and female individuals. Most individuals in each group (54–60%) were in the lowest income quintile (Q1). At least 75% of study individuals had an Elixhauser Comorbidity Index of 3 or more compared with 50% of other Winnipeg residents. The number of individuals diagnosed with a mental disorder or involved in the justice system was also higher in the study groups compared with other Winnipeg residents. Of note, many individuals in the cohort had multiple methamphetamine-related health system contacts during the study timeframe (of which, only the first was retained as the ‘index event’). Among all 1041 individuals with 3980 total methamphetamine-related contacts, 718 (69%) had a single contact, 280 (27%) had from 2 to 5 contacts and 43 (4%) had 6–29 contacts.

Table 1. Cohort characteristics. Winnipeg residents with and without a methamphetamine-related health system record and by olanzapine treatment, 2013–2020.

		No methamphetamine-related record		Pre-intervention (Untreated)		Post-Intervention (Treated)		Post-Intervention (Untreated)
		N	%	N	%	N	%	N	%
Total Individuals		632 716	100	578	100	222	100	241	100
Age group	17–24	99 162	15.7	180	31.2	53	23.9	47	19.5
	25–34	106 257	16.8	243	42.0	89	40.1	86	35.7
	35–44	103 435	16.4	100	17.3	53	23.9	78	32.4
	45+	323 862	51.19	55	9.52	27	12.16	30	12.45
Median age (IQR)		46 (30.79)		29 (11.68)		31 (13.11)		33 (13.76)
Sex	Male	308 175	48.7	248	42.9	129	58.1	128	53.1
	Female	324 541	51.3	330	57.1	93	41.9	113	46.9
Income quintile	Q1 (lowest)	114 006	18.0	347	60.0	126	56.8	130	53.9
	Q2	120 388	19.0	90	15.6	40	18.0	49	20.3
	Q3	123 629	19.5	62	10.7	25	11.3	30	12.5
	Q4	120 910	19.1	43	7.4	18	8.1	s	s
	Q5 (highest)	127 552	20.2	11	1.9	8	3.6	s	s
Elixhauser Comorbidity Index	0	153 413	24.3	51	8.8	18	8.1	25	10.4
	1–2	278 757	44.1	194	33.6	81	36.5	64	26.6
	3+	200 545	31.7	333	57.61	123	55.41	152	63.07
Mental disorder diagnosis	Mood/anxiety disorder	155 159	24.5	381	65.9	140	63.1	137	56.9
	Personality disorder	1182	0.2	62	10.7	21	9.5	25	10.4
	Psychosis	11 759	1.9	108	18.7	37	16.7	60	24.9
	Substance use disorder	26 796	4.2	333	57.6	113	50.1	143	59.3
Justice system involvement	Charged with violent incident	1180	0.2	22	3.8	12	5.4	9.0	3.7
	Charged with non-violent incident	8991	1.4	225	38.9	77	34.7	80	33.2

s, suppressed due to small cell size.

In table 2, we present crude proportions and median times for the patient and system outcomes by study group. Prehospital physical or chemical restraint was used for 1–5% of individuals. Between 88% and 99% of individuals were transported to an ED by the WFPS. Among these, 31–41% were assigned to high-resource ED beds, 3–5% left against medical advice, 11–19% left without being seen by a physician and 7–8% were admitted to hospital or transferred to another hospital.

Table 2. Distribution of patient and health system outcomes by olanzapine treatment proportions and median ED wait times among Winnipeg residents with a methamphetamine-related health system record, 2013–2020.

	Pre-intervention (untreated, n=578)		Post-intervention (treated, n=222)		Post-intervention (untreated, n=241)
Outcome	Proportion or median time	95% CI	Proportion or median time	95% CI	Proportion or median time	95% CI
Prehospital physical or chemical restraint	0.01	0.01 to 0.03	0.05	0.02 to 0.08	0.01	0.00 to 0.04
Transported to ED by WFPS	0.92	0.90 to 0.94	0.99	0.98 to 1.00	0.88	0.83 to 0.92
Left ED against medical advice	0.03	0.02 to 0.05	0.05	0.03 to 0.09	0.04	0.02 to 0.08
Left ED without being seen	0.12	0.09 to 0.15	0.11	0.08 to 0.16	0.19	0.15 to 0.25
Assigned to high resource ED bed	0.41	0.37 to 0.45	0.31	0.25 to 0.38	0.34	0.28 to 0.41
Admitted/transferred after ED visit	0.08	0.06 to 0.11	0.07	0.04 to 0.11	0.07	0.05 to 0.12
Time from registration to triage (min)	3	3 to 3	2	2 to 2	3	3 to 4
Time from triage to ED bed (min)	27	20 to 34	35	24 to 46	31	21 to 41
Time from ED bed to discharge (min)*	369	312 to 453	520	430 to 610	361	227 to 495
Time from ED bed to admission/transfer (min)†	1246	924 to 1568	1599	919 to 2279	1567	884 to 2250

^*For those not admitted to hospital.

^†For those admitted or transferred.

ED, emergency department; WFPS, Winnipeg Fire Paramedic Service.

Table 3 shows unadjusted and IPTW-adjusted ORs and HRs for the outcomes, first for the post-intervention treated group vs the pre-intervention group, and then the post-intervention treated group vs the post-intervention untreated group. In the first unadjusted comparison, prehospital restraint was 3.32 (95% CI 1.32 to 8.32) times more likely and being transported to the ED by the WFPS was 7.70 (95% CI 2.13 to 27.82) times more likely in the post-intervention group. Conversely, the odds of being assigned to a high-resource ED bed were lower in the post-intervention versus pre-intervention group (OR 0.65, 95% CI 0.46 to 0.91). After adjusting with IPTWs, the statistical significance of these findings remained. Additionally, the adjusted HR for the median time from triage to a bed was lower in the post-intervention group: 0.79 (95% CI 0.66 to 0.94). In the second unadjusted comparison, the treated group was significantly more likely to be transported to the ED by the WFPS (OR 12.7, 95% CI 3.45 to 46.95) and less likely to leave the ED without being seen (adjusted OR 0.53, 95% CI 0.31 to 0.91). After applying IPTWs, these findings remained statistically significant.

Table 3. Comparison of patient and health system outcomes by olanzapine treatment crude and IPTW-adjusted ORs and HRs for Winnipeg residents with a methamphetamine-related health system record, 2013–2020.

Crude comparisons
	Post-intervention (treated) vs pre-intervention (untreated)		Post-intervention (treated) vs post-intervention (untreated)
Outcome	Ratio	95% CI	Ratio	95% CI
Prehospital physical or chemical restraint (OR)	3.32	1.32 to 8.32	3.37	0.99 to 11.48
Transported to ED by WFPS (OR)	7.70	2.13 to 27.82	12.72	3.45 to 46.95
Left ED against medical advice (OR)	1.85	0.85 to 4.03	1.32	0.53 to 3.28
Left ED without being seen (OR)	0.95	0.58 to 1.56	0.53	0.31 to 0.91
Assigned to high resource ED bed (OR)	0.65	0.46 to 0.91	0.89	0.59 to 1.33
Admitted/transferred after ED visit (OR)	0.80	0.44 to 1.45	0.90	0.45 to 1.82
Time from registration to triage (HR)	1.17	1.00 to 1.37	0.94	0.77 to 1.13
Time from triage to ED bed (HR)	0.89	0.75 to 1.06	0.97	0.78 to 1.20
Time from ED bed to discharge (HR)*	0.91	0.76 to 1.09	1.04	0.83 to 1.30
Time from ED bed to admission/transfer (HR)†	0.77	0.43 to 1.38	1.21	0.59 to 2.49

IPTW-adjusted comparisons
	Post-intervention (treated) vs pre-intervention (untreated)		Post-intervention (treated) vs post-intervention (untreated)
Outcome	Ratio	95% CI	Ratio	95% CI
Prehospital physical or chemical restraint (OR)	3.89	1.49 to 10.15	3.12	0.95 to 10.31
Transported to ED by WFPS (OR)	8.98	2.50 to 32.33	11.83	3.19 to 43.80
Left ED against medical advice (OR)	1.50	0.70 to 3.19	1.34	0.54 to 3.35
Left ED without being seen (OR)	0.77	0.47 to 1.26	0.56	0.32 to 0.97
Assigned to high resource ED bed (OR)	0.66	0.47 to 0.92	0.90	0.60 to 1.36
Admitted/transferred after ED visit (OR)	0.77	0.43 to 1.39	0.86	0.71 to 1.05
Time from registration to triage (HR)	1.15	0.98 to 1.34	0.86	0.71 to 1.05
Time from triage to ED bed (HR)	0.79	0.66 to 0.94	1.00	0.81 to 1.24
Time from ED bed to discharge (HR)*	0.90	0.75 to 1.08	0.90	0.72 to 1.12
Time from ED bed to admission/transfer (HR)†	0.61	0.34 to 1.09	0.87	0.43 to 1.76

Bolded associations indicate those with p<0.05.

^*For those not admitted to hospital.

^†For those admitted or transferred.

ED, emergency department; IPTW, inverse probability of treatment weight; WFPS, Winnipeg Fire Paramedic Service.

Discussion

This study examines the outcomes of patients who were eligible to receive prehospital olanzapine for acute methamphetamine intoxication. In the post-intervention period, patients who received olanzapine were more likely to be transported to hospital and less likely to leave hospital without being seen compared with those who did not receive olanzapine. Patients who received olanzapine were also less likely to be assigned a high-resource bed in the ED and more likely to be physically or chemically restrained after treatment compared with those who were untreated in the pre-intervention period. These results offer several important points of discussion and have implications for further study.

Given the unpredictable nature of methamphetamine intoxication and associated high-risk behaviours,³⁶ our findings showing fewer patients refusing care or leaving hospital against medical advice suggest that olanzapine is a promising prehospital treatment option. This view must be balanced against patients’ documented frustrations with ED experiences,^{37 38} as well as the possibility that pervasive ED overcrowding in some jurisdictions contributes to poor outcomes.³⁹ In contrast, the finding that patients might be subjected to more physical and chemical restraint after receiving olanzapine may appear both counterintuitive and undesirable. It is unlikely, however, that treatment with olanzapine escalates patient agitation, but instead that paramedics are inclined to attempt treatment in borderline cases of severe agitation before resorting to restraint when needed.

Our study offers some new insights, as there are few other studies examining similar outcomes of methamphetamine intoxication on comparable terms. Murphy et al tracked ED length of stay, chemical restraint and psychiatric hospitalisation for cases with confirmed methamphetamine use compared with those with other drugs or no drugs in San Francisco, California.⁴⁰ They confirmed longer ED stays and higher rates of chemical restraint for methamphetamine intoxication cases compared with those with no drug use, but did not evaluate different treatment strategies beyond comparing medications for severe agitation.⁴⁰ Another study, also from California, compared treatment pathways for acute methamphetamine intoxication before and after implementing a standard protocol that included olanzapine.¹⁹ The authors found no difference in ED length of stay or rates of restraint use after protocol implementation, but their patient group appears to have included more severely agitated patients than the current study.¹⁹ Work characterising prehospital methamphetamine cases in Victoria, Australia, has also documented increases in transports to an ED as well as police attendances.²² While not directly comparable, these studies document patterns of presentations and outcomes that are generally consistent with our results, while highlighting differences in case identification, treatments, and outcome measurement.

Studies on methamphetamine use in the prehospital and ED settings commonly report and analyse cases of intoxication, not individual people, and only sometimes account for repeat visits by individuals in the analysis.⁴⁰ However, Richards et al documented ED recidivism in patients with confirmed use of amphetamine or derivatives in California.¹³ They categorised patients into groups by amphetamine-related visits per year: 1, 2–5, 6–11 and 12+.¹³ The two highest use groups represented 2% of all individuals but accounted for 10% of visits and had statistically significant differences in their demographic and medical history compared with the rest of the cohort.¹³ Although the current study analyses outcomes related to individuals’ index visit, the results show that each unique patient averaged between three and four contacts where they would have been eligible for olanzapine, and likely more that were simply related to methamphetamine. The addition of case-based analysis to characterise patterns of visits, their change over time, and the effects of treatment would capture important perspectives on the phenomenon as experienced by both patients and systems of care and is planned for future analyses by our team.

Our study has several important strengths. Use of the comprehensive administrative health data in the Manitoba Repository linked to prehospital data collected by paramedics allows for an investigation of key patient and system-oriented outcomes from multiple angles and supports a longitudinal cohort design with opportunities to update the cohort for future analyses. Our decision to examine index visits only avoids bias that would occur with the inclusion of multiple cases from the same individuals. Our approach also allows for a detailed description of the patient group. In keeping with previous reports from Manitoba,³² individuals eligible for olanzapine experience marked burdens of not only medical co-morbidity but also poverty and mental illness. We included First Nations and Red River Métis identifiers in our analysis in recognition of how Canada’s colonial history continues to shape health outcomes for Indigenous peoples, including outcomes related to methamphetamine.^{41 42} The social and structural determinants of health also impact First Nations, Red River Métis and other underserved populations, driving higher rates of substance use, legal system involvement and healthcare service use.^43–46 While not reported on to the same extent in similar studies,^{13 19 40} these circumstances are likely shared among patient groups in other jurisdictions. The complex and often marginalising factors associated with methamphetamine use and intoxication resist simple solutions: in 2019, Winnipeg’s Illicit Drug Task Force noted the importance of the social determinants of health in a multi-level response to address the use of illicit drugs and their associated harms.⁴⁷

Some important limitations of our study are that prehospital olanzapine in Winnipeg is available to a variety of patients and is administered at the discretion of the attending paramedics. The WFPS database used to identify those eligible for olanzapine treatment is free-form narrative, and it is difficult to evaluate indications retrospectively without standardised documentation of indications, although this method has been used previously for case identification in the paramedic record.²² Because of the discretionary use of olanzapine by paramedics, there is also the possibility of confounding by indication, whereby individuals who received olanzapine were more agitated than those who did not, resulting in differences in clinical trajectories once they arrived at the ED. Finally, it would have been ideal to include paramedic characteristics in the propensity score given the possibility of variability in olanzapine administration between paramedics. Unfortunately, however, these data were not available for the present analysis and will have to be addressed in future work.

Conclusions

Paramedic administration of olanzapine in the prehospital setting for methamphetamine intoxication demonstrates some promise in connecting patients to ED care and improving ED outcomes. Despite this promise, treating methamphetamine intoxication is complex and shaped by the many social determinants of health that intersect with methamphetamine use, including psychiatric comorbidity, legal involvement and financial/housing instability. In turn, there remains a need for comprehensive policy solutions that move beyond single, episodic treatments to focus on upstream policy changes that include improved access to prevention, harm reduction and treatment services; stronger coordination across healthcare, social services and justice systems; and the use of linked data to inform ongoing policy and programme development.

Data availability statement

Data are available upon reasonable request.