Short-term mortality after opioid initiation among opioid-naïve and non-naïve patients with dementia: a retrospective cohort study.

Abstract

Background: In the ongoing opioid epidemic, the mortality risk of opioid initiation in patients with dementia or mild cognitive impairment (MCI) remains understudied despite their vulnerability. This study evaluates mortality risks associated with opioid exposure in patients diagnosed with dementia or MCI by comparing outcomes between the initiation and continuation groups.

Methods: We conducted a retrospective cohort study using data from a Northern California academic healthcare system (Stanford Health Care Alliance; 2015/01/01-2024/07/31), including 27,757 patients aged 50-100 with dementia or MCI. Of these, 14,105 received opioids after diagnosis and were classified as initiation (opioid-naïve; n=9443) or continuation (non-naïve; n=4662) groups. Cox regression assessed 14-day mortality risk. Aalen's additive model examined time-varying impact up to 180 days. Potential causes of death were extracted from clinical notes using GPT-3.5-Turbo. We also analyzed an independent community healthcare system cohort (Providence Health & Service; n=208,306) from western US states (2015/01/01-2023/05/31) as a replication cohort.

Results: In the primary cohort, 4.1% (572/14,105) of patients died within 14 days of opioid exposure. The initiation group had a significantly higher 14-day mortality risk than the continuation group (adjusted hazard ratio (aHR), 2.00 (1.59-2.52); P<0.0001). The replication cohort had a 14-day mortality rate of 6.2% (7022/113,343) with a smaller difference between the initiation (n=77,168) and continuation (n=36,175) groups (aHR 1.22 (1.16-1.30); P<0.0001). In both cohorts, elevated risk stabilized after day 30. In the primary cohort, respiratory conditions (62% vs. 48%, P<0.1), particularly pneumonia (38% vs. 19%, P<0.05), were more prevalent among the initiation group who died early.

Conclusions: Starting opioids in patients with dementia or MCI is associated with elevated short-term mortality risks, with the initiation group having twice the 14-day mortality risk in academic settings and a smaller but significant increase in community healthcare systems. The first 30 days after initiation represent a critical risk window, likely due to a lack of tolerance to opioid adverse effects. These findings underscore the need for cautious initiation, tailored follow-up protocols accounting for healthcare setting characteristics, and close monitoring during the first month in this vulnerable population.