Opioid-related morbidity substantially contributes to the extensive opioid-related disease and economic burden observed across North America [33-35]. Strong increases in opioid-related hospitalizations post-2000 have been observed in the US, as well as other countries with rising opioid availability [10, 12, 21, 36, 37].
We examined possible correlations between opioid dispensing and opioid-related hospitalization rates across Canada in contexts of most provinces featuring inversion patterns for opioid dispensing during the study period (2007 – 2016; ). These patterns reflect recent reductions in opioid availability following a variety of restrictive measures (e.g., intensified prescription monitoring, de-scheduling of select opioid formulations, restrictive opioid prescription guidelines) in a wider context of continuously rising opioid-related mortality and morbidity outcomes [3, 38, 39].
We identified significant correlations between opioid dispensing and opioid-related hospitalization rates for three (QC, NB, NS) of the ten provinces, with borderline significance in a fourth (SK). These results, partially, affirm an extensive body of evidence identifying correlations between the volume of opioid availability and health harm outcomes on population levels [14, 15, 19, 40]. While these correlations were identified in select provinces, they were not identified in most others. These selective findings, however may not be coincidental when considering key ecological dynamics. Importantly, opioid-related hospitalizations data may comprise incidents related to both licit and illicit opioids, as the data used do not discriminate between opioids by legal status [8, 10, 41]. However, opioid availability data are limited to levels of medical opioid dispensing only. These circumstances, consequently, may involve a ‘contamination’ of hospitalization data through illicit opioid-related cases and a subsequent distortion of possible correlation effects.
In Canada, population-levels of illicit opioid use have been relatively limited, compared with high levels of (medical and non-medical) PO use estimated at >20% and >5%, respectively, in peak years around 2010 . However, there have been strong increases in the availability of – highly potent and toxic – illicit synthetic opioid products (e.g., fentanyl, carfentanyl products) in Canada in more recent years, linked to substantial increases in fatal poisonings due to acutely elevated risk properties of these substances [4, 43, 44]. In this context, we note that significant (including borderline) correlations between opioid dispensing and hospitalizations were found specifically in four of the five Canadian provinces (PEI, NS, QC, NB, SK) reporting the lowest levels of contribution (<25%) of (mostly illicit) fentanyl or fentanyl-analogue product involvement among opioid-related mortality in 2017 (i.e., ‘low contamination’ provinces); conversely, no correlations were observed in the provinces (MN, ON, AB, BC) with fentanyl products identified as a contributor to mortality in the majority (>50%) of fatalities in 2017 (i.e., ‘high contamination’ provinces; [7, 8, 41]). Similarly differential – but consistent with our results – patterns of illicit opioid involvement in opioid-related hospitalizations have been shown in individual-based analyses in select provinces . These – rather consistent – differentiation patterns in our ecological study results allow to plausibly speculate that, in the absence of illicit opioid-related ‘contamination’ effects, the strength of province-based correlation signals between opioid dispensing and hospitalizations likely would have been more pronounced.
In this context, our findings of select Canada-based correlations between opioid dispensing and opioid-related hospitalizations provide additional evidence on associations between population-level opioid availability and key adverse health outcomes [34, 46-48]. This association – at least while supply for non-medical opioid use mostly involved prescription opioid products – had simple but essential implications: The higher the volume of opioids dispensed into the population, the higher the levels of consequential morbidity or mortality harms [15, 49, 50]. This provided crucial insight for guiding opioid policy and medical practice control (e.g., prescription guidance) in the distinct contexts of traditionally high opioid availability in North America, combined with evidence of only limited effectiveness of opioids for chronic pain therapy [51-53]. Beyond these current contexts, where illicitly produced opioids have increasingly replaced pharmaceutical-grade opioids for non-medical use, these insights may be helpful for settings where opioid availability is still low and preventive restraints can facilitate appropriate balance between opioid availability and related adverse health outcomes [5, 38, 54, 55]. As a potentially relevant limitation for the present analyses, we note that the (general population-level) data examined did not include additional patient-level descriptors, or other data variables that could have been used to examine possibly confounding factors in the correlations assessed.