Evidence guiding the selection of opioids for continuous infusion in critically ill patients receiving mechanical ventilation is limited, and contemporary guidelines commonly treat opioids as interchangeable. This study aimed to describe temporal trends and identify factors associated with the choice of the first continuous opioid infusion in ICU patients, and to quantify variability in prescribing between ICUs.
The investigators performed a retrospective historical cohort study including patients who received mechanical ventilation in intensive care units in Alberta, Canada, between 2013 and 2019. The analysis included admissions in which the first continuous opioid infusion after intubation was one of three agents: fentanyl, hydromorphone, or morphine. Temporal trends in initial opioid selection were assessed across the study period. To identify patient-level factors associated with opioid choice and to evaluate between-ICU variability, the authors used a multilevel multinomial logistic regression model and reported adjusted median odds ratios (aMOR) to quantify ICU-level variation.
Among 17,898 admissions that met inclusion criteria across 20 ICUs, fentanyl was the most frequent initial infusion (78.6%; n = 14,060). Hydromorphone was the second most common (16.3%; n = 2,917) and morphine the least common (5.1%; n = 921).
Use of fentanyl declined over the study period from 89.0% of initial opioid infusions in 2013 (n = 1,468 of 1,650) to 75.0% in 2019 (n = 1,763 of 2,351), with P < 0.001 for trend. Morphine use also decreased (7.7% in 2013 to 3.2% in 2019; P < 0.001). By contrast, hydromorphone use rose substantially from 3.3% in 2013 (n = 55 of 1,650) to 21.8% in 2019 (n = 513 of 2,351); P < 0.001.
In adjusted analyses using multilevel multinomial logistic regression, surgical patients were more likely to receive hydromorphone rather than fentanyl (adjusted odds ratio [aOR] 1.52; 95% confidence interval [CI] 1.32–1.74). Surgical patients were less likely to receive morphine compared with fentanyl (aOR 0.81; 95% CI 0.68–0.96). The abstract reports these specific associations; further details on other patient-level covariates included in the model and their effect estimates were not reported in the abstract.
The study documented substantial variation across the 20 participating ICUs in choice of initial opioid. Variation was quantified with the adjusted median odds ratio (aMOR): for choosing hydromorphone versus fentanyl the aMOR was 14.5 (95% CI 7.3–25.7), and for morphine versus fentanyl the aMOR was 5.2 (95% CI 3.11–9.2). These aMOR values indicate large between-ICU differences in the likelihood of using hydromorphone or morphine as the first infusion compared with fentanyl, after adjustment for measured patient-level factors.
Across 20 ICUs in Alberta from 2013 to 2019, prescribing practices for the initial continuous opioid infusion in mechanically ventilated patients changed over time and varied markedly between ICUs. Fentanyl remained the most commonly used initial agent overall, but hydromorphone use increased substantially while morphine use declined. The authors highlight the need for further research to explore drivers of the observed inter-ICU variability and to examine whether choice of initial opioid infusion is associated with clinically important outcomes, with the goal of informing evidence-based guidance.
The abstract provides summary results for opioid distribution, temporal trends, and two adjusted associations (surgical status with hydromorphone and morphine use) and reports measures of between-ICU variability (aMORs). Details not reported in the abstract include the full list of covariates included in the adjusted model, model diagnostics, patient demographic and illness-severity characteristics, specific clinical outcomes related to opioid choice, and sensitivity analyses. Those details would need to be consulted in the full article for a complete appraisal.