What impact did the COVID-19 pandemic have on the variability of fentanyl concentrations in the Vancouver, Canada illicit drug supply? An interrupted time-series analysis

Introduction

Overdose deaths in North America increased dramatically after the onset of the COVID-19 pandemic in 2020.1 2 In the USA, the number of overdose deaths reached the highest levels ever recorded during the first half of 2020, coinciding with the onset of the pandemic.2 Other jurisdictions saw similar increases corresponding to the start of the pandemic; for instance, in British Columbia (BC), Canada’s westernmost province and the one hit hardest by the ongoing overdose crisis, monthly overdose death rates more than doubled in the months following the onset of the COVID-19 public health emergency.3 Hypothesised reasons for this increase include lockdown attributable service interruption and social isolation, as well as possible drug market disruption4; however, research is required to explore these hypotheses and better inform ongoing public health interventions. Unfortunately, explanations for increased overdose deaths coinciding with the COVID-19 pandemic remain largely underexplored.

The illicit opioid drug supply in Vancouver, BC has been dominated by fentanyl since 2016.5 It has been suggested that one of the reasons illicit fentanyl is so heavily associated with overdose mortality is because of the wide variation in unknown concentrations of it in final prepared products (eg, ‘dope’, ‘down’ or counterfeit pills).6 7 In an illicit drug market plagued by this unpredictability, one of the ways people who use drugs (PWUD) assess the potency of fentanyl before consuming it (thereby putting themselves at risk of overdose) is to check their drugs.8 Drug checking is a harm reduction intervention growing in popularity in North America as a way for PWUD to protect themselves in the context of a toxic drug supply.8 Different drug checking technologies have various advantages and disadvantages, but the favoured method for drug checking service delivery in harm reduction sites in BC includes Fourier-transform infrared (FTIR) spectroscopy.5 9 FTIR drug checking can determine up to six components in a drug sample, both active and inert cutting agents, and can offer estimates on concentration of identified components.5 8

A recent time-series analysis conducted in Vancouver, BC showed the concentration of fentanyl in illicit opioids rose and fell between 2017 and 2019 (monthly medians ranging from 4.5% to 10.4%), with absolute monthly variance decreasing by 0.1% per month over the 26-month period.5 That study ended prior to the onset of the COVID-19 pandemic, so the effect of the pandemic on fentanyl levels in illicit opioids has remained unknown, as to our knowledge, there are no other studies that examine these temporal changes. This study used data derived from the same community drug checking services in the Canadian setting of Vancouver, BC to determine if there were any specific changes to the fentanyl concentration of illicit opioids brought to drug checking services following the onset of the COVID-19 public health emergency. We hypothesised that a potential supply shock caused by COVID-19-related measures would result in increased volatility to the fentanyl supply, measured as variance of the fentanyl concentrations in community drug checking samples.

Methods

Drug checking services have been offered in Vancouver harm reduction sites since October 2017 using Bruker (Billerica, Massachusetts, USA) ALPHA FTIR spectrometers and BTNX (Markham, Ontario, Canada) fentanyl immunoassay strips.9 Samples are checked for participants at point of care with the results being available in as little as 5 min. Results include components detected by FTIR analysis (active and inert compounds present above the spectrometer detection limit) and paired harm reduction advice from trained technicians.9 Previous research indicates that drug checking with FTIR and fentanyl immunoassay strips results in a fentanyl detection sensitivity as high as 98% and specificity of 100%.10 Point-of-care drug checking results and corresponding FTIR spectrum files are uploaded contemporaneously to an electronic data capture system, facilitating rapid sharing of results and calls for assistance from other technicians if needed.

Using a validated quantification model for fentanyl hydrochloride,5 historical FTIR spectra from fentanyl-positive drug checking samples in Vancouver were analysed to determine individual fentanyl concentrations. Details of the model are described elsewhere,5 but briefly, using quantitative nuclear MR spectroscopy results from a subset of samples forwarded for confirmatory analysis at Health Canada’s Drug Analysis Service, a fentanyl quantification model was developed with the QUANT 2 functionality of the spectroscopy software OPUS (V.7.8.44; Bruker, Billerica, Massachusetts, USA). QUANT 2 uses a partial least squares fit method which correlates spectral intensity in specified wavelength regions with values that were generated from reference entries (ie, point-of-care FTIR scans later quantified by the Drug Analysis Service).5 The model outputs an estimation of fentanyl concentration as a percent-by-weight for each sample. Samples included in the present analysis were restricted to participant-expected opioids (fentanyl, heroin, both, or ‘down,’ a slang term to refer to illicit opioids that contain an unspecified opioid, typically fentanyl) except pharmaceutical opioids or pressed pills. Samples must have been checked in Vancouver, BC (five harm reduction sites included) between November 2017 and October 2020. Samples must have been determined to be fentanyl-positive by either FTIR or fentanyl immunoassay at point of care to be included in the analysis. Drug samples included any specimen (eg, powders, pebbles, crystals) meeting these criteria and may have been submitted by service users prior to or post consumption. Fentanyl-negative samples were excluded as the monthly proportion of fentanyl-negative opioid samples did not appreciably change over the study period and the quantification model cannot reliably return zero results, requiring extrapolating to 0.0% fentanyl based on negative fentanyl immunoassay and FTIR results.

All statistical analyses were conducted using SAS (V.9.4; SAS Institute). Descriptive statistics of fentanyl concentrations were calculated for each calendar month and plotted in a graph. We conducted an interrupted time-series analysis using an autoregressive model on the monthly variance values to detect the presence of residual autocorrelation. We chose to use variance as a measure of variability as increased dispersion of fentanyl concentrations indicates volatility and inconsistency of the fentanyl supply. On identifying significant autocorrelation up to a lag of 4, we proceeded with ordinary least squares regression, employing Newey-West autocorrelation adjusted standard errors.11 We chose this method for its straightforward interpretability. Our interruption date of April 2020 was used to account for the delayed impact of pandemic measures. For reference, the provincial health officer of BC declared a public health emergency on 17 March 2020 and a provincial state of emergency was declared the following day.12 During that week of March 2020, schools and dine-in restaurants were ordered closed, elective surgeries were postponed, harm reduction services were limited,13 and events with more than 50 people were banned province-wide.14

The public was not involved in the design, conduct or reporting plans of this research.

Results

During the 36-month study period, 4713 samples met the criteria for inclusion and were quantified using the fentanyl quantification model. Monthly median fentanyl concentrations ranged from 4.5% in August 2019 to 10.4% in August 2018 and the median of all samples (n=4713) was 7.3% (quartiles 1–3: 5.3%–10.1%). Monthly variance values ranged from a low of 7.9% in December 2017 to a high of 159.2% in September 2020. Data from 2020 are presented in table 1. An interrupted time-series analysis showed that the variance in monthly fentanyl concentrations increased significantly following the declaration of the COVID-19 public health emergency (figure 1), with a level change of 26.1 (95% CI 7.2 to 45.0, p=0.011) and a slope change of 15.8 (95% CI 10.2 to 21.4, p<0.001) postinterruption (table 2).

Request reuse permission

Discussion

Coinciding with the public health measures following the declaration of the COVID-19 public health emergency in this setting, the variance among fentanyl concentrations of opioid drug checking samples increased significantly and dramatically. Prior to the onset of the public health emergency, the variance among fentanyl concentration was rather consistent, even noted to decrease over the preinterruption period.5 It was suggested that this decrease in variance may have been due to the unregulated fentanyl market stabilising on a desired concentration.5 Alternatively, postemergency declaration, the variance in concentration of fentanyl samples saw an immediate increase (level change) with a new trend of increasing variability (slope change).

It is not immediately clear what the source of the drastic increase in variability was postdeclaration. A suggested source of drug market destabilisation has been that the restrictive pandemic-related trade and travel measures may have impacted the flow of drugs (including fentanyl) around the world4; however, fentanyl imports into Canada are estimated to have already been decreasing over recent years.15 16 A reduction in the overall potency of drugs, potentially due to trade restrictions accounting for difficulties in procuring them, was not observed in this study; rather, illicit opioids became more potent (monthly medians increased). Importantly, authorities have concluded that a significant proportion of fentanyl in the BC drug market is believed to be produced domestically, as evidenced by the growth in seizures of fentanyl precursor molecules in concurrence with the decrease in synthesised fentanyl seizures at the Canadian border.16 17 Thus, factors other than trade interruption likely explain the increased variability. One possible explanation is that COVID-19 restrictions and related disruptions destabilised drug preparation activities in settings where drugs are prepared into units that are then sold in the street drug market. Given the known role of potent fentanyl being likely the largest contributor to fatal overdoses, this factor may exceed other concerns that have been voiced such as service interruptions or social isolation as contributors.18 It is important to recognise, however, that these factors may differ from other drugs and drug markets.

This study has some limitations to consider. First, pandemic-related physical distancing restrictions in Vancouver led to a drop in visits to supervised consumption facilities and drug checking services.13 As a result, during the initial months of the COVID-19 public health emergency, we observed a decline in the number of drug checking samples in our study. This reduction in sample size raised potential concerns about the reliability of variance estimates as smaller sample sizes can often lead to inflated variances. However, we found that even as the sample sizes rebounded in subsequent months, the variance in fentanyl concentrations did not revert to prepandemic levels (table 1). Relatedly, it is not clear how the drop in drug checking visits may have affected the representation of the unregulated opioid drug supply in our study, but it is worthy of mention as a limitation of the study. As well, the study only includes samples collected in one part of the Vancouver metropolitan area, so the generalisability to other settings may be limited, though we know of no reason why there would be heterogeneity across the province. Finally, we acknowledge that the lack of a comparison group for our study. The significant impact of the COVID-19 pandemic presented unique challenges for identifying an appropriate comparison time-series and we hope future research can help distinguish aspects of the pandemic responsible for the changes to the drug supply.

Fentanyl and its analogues, being such potent opioids, likely pose a markedly increased risk when their concentration fluctuates in street samples.6 7 Even in a setting such as the one described in this study where fentanyl is ubiquitous in the illicit opioid supply, variation in potency places PWUD who rely on illicit opioids at increased risk of both fatal and non-fatal opioid toxicity due to the unpredictability on drug make-up.7 19 It is likely that the dramatic increase in overdose mortality in BC after the declaration of the COVID-19 public health emergency was caused by a multitude of interconnected reasons.18 Physical distancing measures led to isolation, driving people to use drugs alone, a major risk factor for fatal overdose.20 21 However, following relaxation of physical distancing measures, overdose rates remained high,3 so the role of the drug supply must be seriously considered. In the already precarious risk environment, external shocks to the unregulated drug supply destabilise the market and only enhance the existing risk of harm and death.