# Developing Patient-level Risk Prediction Models for Prescription Opioid Overdose

> **NIH NIH R01** · UNIVERSITY OF CALIFORNIA AT DAVIS · 2020 · $619,158

## Abstract

Developing Patient-level Risk Prediction Models for Prescription Opioid Overdose
Summary / Abstract
Morbidity and mortality related to prescription opioid use and abuse are major clinical and public health
problems. Reducing prescription opioid overdose rates requires efforts on multiple fronts aimed at reducing
both patients’ transition to long-term opioid use and their subsequent overdose risk. Prescription drug
monitoring programs (PDMPs)—statewide electronic databases containing all controlled substance
prescriptions and that clinicians can query in real time—are one promising tool for promoting safe opioid
prescribing, but their full potential remains untapped. One reason for this is that most prior research has
focused on patients’ mean opioid dose, and has used either aggregate data or data restricted to specific health
systems or insurers. Evidence derived from large, population-based, patient-level longitudinal data is needed
to better inform national and state efforts to reduce prescription opioid-related harms. For example, high-dose
opioid use is associated with greater overdose risk, but our preliminary data indicate that rate of opioid dose
escalation is also an important and under-studied predictor of overdose risk. This proposal’s long-term goal is
to lay the groundwork for multivariable PDMP-based risk prediction tools that clinicians and public health
officials can use to assess overdose risk in the same way that Framingham-type tools are currently used to
assess cardiac risk. The proposal’s overarching hypotheses are that rate of opioid dose escalation will be
associated with both transition to long-term opioid use and incident opioid overdose, and that overall overdose
risk will be concentrated in a relatively small group of high-risk patients. The objective of this proposal is to
identify longitudinal opioid prescribing patterns associated with a) new opioid users’ transition to long-term use
(i.e., continual opioid use for >90 days), b) patients’ incident fatal or nonfatal opioid-related overdose (including
heroin overdose), and c) repeat overdose by analyzing longitudinal, patient-level prescribing and overdose
data for all of California between 2008 and 2016. We will take the novel step of linking 3 statewide longitudinal
databases at the patient level: prescribing data from California’s PDMP, death certificate data, and statewide
hospital discharge and emergency department data. Mixed-effects regression methods for longitudinal data will
be used to analyze associations between opioid prescribing patterns and proposal outcomes. Results from
these analyses will be used to develop and prospectively validate clinical risk prediction models for each
outcome. This project will produce validated risk prediction models derived from population-based, patient-level
longitudinal data that will be used to build clinical risk prediction tools that can eventually be incorporated into
PDMPs in order to inform prescribing decisions at th...

## Key facts

- **NIH application ID:** 9982285
- **Project number:** 5R01DA044282-04
- **Recipient organization:** UNIVERSITY OF CALIFORNIA AT DAVIS
- **Principal Investigator:** Stephen G Henry
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2020
- **Award amount:** $619,158
- **Award type:** 5
- **Project period:** 2017-09-30 → 2023-07-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/9982285

## Citation

> US National Institutes of Health, RePORTER application 9982285, Developing Patient-level Risk Prediction Models for Prescription Opioid Overdose (5R01DA044282-04). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/9982285. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*