03/18/2024
By Casey Leon

The Public Health Program invites you to attend a doctoral dissertation defense by Casey Leon on "Opioid-related prodromal symptom occurrence rates, long-term opioid therapy, and associates with opioid overdose among Veterans."

Candidate: Casey Leon
Degree: ScD in Public Health - Epidemiology
Defense Date: Monday, March 25, 2024
Time: 9 a.m.
Location: O'Leary Hall Room 500M

Dissertation Committee:

  • Chair: Wenjun Li
  • Leland Ackerson
  • Dan Berlowitz
  • Hong Yu

Abstract

OBJECTIVE: The aim of this research was to investigate the occurrence of prodromal opioid-related symptoms among patients on long-term opioid therapy (LTOT) in care at the Veterans Health Administration (VHA) and their association with opioid overdose events.

BACKGROUND: People on LTOT for chronic non-cancer pain are at risk of adverse events due to opioids. Identifying early potential predictors of adverse events is a high priority in combatting the effects of the opioid epidemic through preventive interventions. The prodromal symptoms of interest were neurocognitive symptoms (NCS) (e.g. cognitive impairment, sedation, light-headedness, altered mental status, and intoxication), and aberrant medication-related behaviors (AMRB) (i.e. requests for early refills, unauthorized dose escalation, doctor shopping, prescription forging, diversion, tampering, and stockpiling medications).

METHODS: In a cohort of patients prescribed LTOT in 2018 from the VHA, prodromal opioid-related symptoms were ascertained using natural language processing (NLP) of clinical notes. Occurrence rates for NCS and AMRB were examined for six months before, during, and one year after discontinuation of LTOT. Incidence rates, incidence rate ratios, and crude and adjusted hazard rations of each prodromal symptom were calculated by patient characteristics. Using both NLP and structured data extraction techniques, we calculated NCS prevalence and opioid overdose rates of patients with and without NCS during LTOT.

RESULTS: Of the 55,652 patients in the cohort, 91.7% were men, 62.4% were between the ages of 55 and 75 years, and 69.7% were White. Among the study cohort, 9.4% of patients on LTOT experienced AMRB and 3.1% experienced opioid-related NCS identified by NLP. In adjusted hazard ratio models alcohol use disorder (AUD), depression substance use disorder (SUD), chronic obstructive pulmonary disease (COPD), and traumatic brain injury (TBI) were significantly associated with increased occurrence of both AMRB and NCS. Additional factors associated with increased occurrence of AMRB were age below 75 years, Black identified, Hispanic ethnicity, service-connected condition or disability, anxiety, and hepatitis C infection. Additional factors associated with increased occurrence of NCS were being Asian and having diagnosed cirrhosis, chronic kidney disease, and pain. When identified with both NLP and structured data, 13.3% of the cohort experienced an opioid-related NCS. Patients experiencing NCS compared to those who did not experience NCS had an increased risk of opioid overdose event with risk ratios of 5.80 during LTOT and 9.09 after LTOT. There was a dose-response relationship between NCS experiences and opioid overdose events.

CONCLUSIONS: Opioid-related AMRB and NCS, prodromal of opioid overdose, could inform opportunities for intervention and prevention of harms due to opioids, and specific subpopulations of patients and time-periods warrant tailored clinical intervention. This research highlights the discriminant validity of NLP-derived prodromal opioid symptoms. Opioid-related NCS captured via structured and NLP-derived data are associated with increased risk for opioid overdose both during and after LTOT episodes identifying NCS as an important prodrome of opioid overdose.