High-risk prescribing and opioid overdose: Prospects for prescription drug monitoring program-based proactive alerts

Peter Geissert, Sara Hallvik, Joshua Van Otterloo, Nicole O'Kane, Lindsey Alley, Jody Carson, Gillian Leichtling, Christi Hildebran, Wayne Wakeland, Richard (Rick) Deyo

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

To develop a simple, valid model to identify patients at high risk of opioid overdose-related hospitalization and mortality, Oregon prescription drug monitoring program, Vital Records, and Hospital Discharge data were linked to estimate 2 logistic models; a first model that included a broad range of risk factors from the literature and a second simplified model. Receiver operating characteristic curves, sensitivity, and specificity of the models were analyzed. Variables retained in the final model were categories such as older than 35 years, number of prescribers, number of pharmacies, and prescriptions for long-acting opioids, benzodiazepines or sedatives, or carisoprodol. The ability of the model to discriminate between patients who did and did not overdose was reasonably good (area under the receiver operating characteristic curve = 0.82, Nagelkerke R 2 = 0.11). The positive predictive value of the model was low. Computationally simple models can identify high-risk patients based on prescription history alone, but improvement of the predictive value of models may require information from outside the prescription drug monitoring program. Patient or prescription features that predict opioid overdose may differ from those that predict diversion.

Original languageEnglish (US)
Pages (from-to)150-156
Number of pages7
JournalPain
Volume159
Issue number1
DOIs
StatePublished - Jan 1 2018

Fingerprint

Prescription Drugs
Drug Monitoring
Opioid Analgesics
Prescriptions
ROC Curve
Hospital Records
Pharmacies
Carisoprodol
Hypnotics and Sedatives
Benzodiazepines
Hospitalization
Logistic Models
History
Sensitivity and Specificity
Mortality

Keywords

  • Addiction medicine
  • Chronic pain management
  • Opioid overdose
  • Opioid prescription
  • Predictive modeling

ASJC Scopus subject areas

  • Neurology
  • Clinical Neurology
  • Anesthesiology and Pain Medicine

Cite this

High-risk prescribing and opioid overdose : Prospects for prescription drug monitoring program-based proactive alerts. / Geissert, Peter; Hallvik, Sara; Van Otterloo, Joshua; O'Kane, Nicole; Alley, Lindsey; Carson, Jody; Leichtling, Gillian; Hildebran, Christi; Wakeland, Wayne; Deyo, Richard (Rick).

In: Pain, Vol. 159, No. 1, 01.01.2018, p. 150-156.

Research output: Contribution to journalArticle

Geissert, P, Hallvik, S, Van Otterloo, J, O'Kane, N, Alley, L, Carson, J, Leichtling, G, Hildebran, C, Wakeland, W & Deyo, RR 2018, 'High-risk prescribing and opioid overdose: Prospects for prescription drug monitoring program-based proactive alerts', Pain, vol. 159, no. 1, pp. 150-156. https://doi.org/10.1097/j.pain.0000000000001078
Geissert, Peter ; Hallvik, Sara ; Van Otterloo, Joshua ; O'Kane, Nicole ; Alley, Lindsey ; Carson, Jody ; Leichtling, Gillian ; Hildebran, Christi ; Wakeland, Wayne ; Deyo, Richard (Rick). / High-risk prescribing and opioid overdose : Prospects for prescription drug monitoring program-based proactive alerts. In: Pain. 2018 ; Vol. 159, No. 1. pp. 150-156.
@article{3ae0a1ed4faa4f93a6b17d86bc8eb233,
title = "High-risk prescribing and opioid overdose: Prospects for prescription drug monitoring program-based proactive alerts",
abstract = "To develop a simple, valid model to identify patients at high risk of opioid overdose-related hospitalization and mortality, Oregon prescription drug monitoring program, Vital Records, and Hospital Discharge data were linked to estimate 2 logistic models; a first model that included a broad range of risk factors from the literature and a second simplified model. Receiver operating characteristic curves, sensitivity, and specificity of the models were analyzed. Variables retained in the final model were categories such as older than 35 years, number of prescribers, number of pharmacies, and prescriptions for long-acting opioids, benzodiazepines or sedatives, or carisoprodol. The ability of the model to discriminate between patients who did and did not overdose was reasonably good (area under the receiver operating characteristic curve = 0.82, Nagelkerke R 2 = 0.11). The positive predictive value of the model was low. Computationally simple models can identify high-risk patients based on prescription history alone, but improvement of the predictive value of models may require information from outside the prescription drug monitoring program. Patient or prescription features that predict opioid overdose may differ from those that predict diversion.",
keywords = "Addiction medicine, Chronic pain management, Opioid overdose, Opioid prescription, Predictive modeling",
author = "Peter Geissert and Sara Hallvik and {Van Otterloo}, Joshua and Nicole O'Kane and Lindsey Alley and Jody Carson and Gillian Leichtling and Christi Hildebran and Wayne Wakeland and Deyo, {Richard (Rick)}",
year = "2018",
month = "1",
day = "1",
doi = "10.1097/j.pain.0000000000001078",
language = "English (US)",
volume = "159",
pages = "150--156",
journal = "Pain",
issn = "0304-3959",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - High-risk prescribing and opioid overdose

T2 - Prospects for prescription drug monitoring program-based proactive alerts

AU - Geissert, Peter

AU - Hallvik, Sara

AU - Van Otterloo, Joshua

AU - O'Kane, Nicole

AU - Alley, Lindsey

AU - Carson, Jody

AU - Leichtling, Gillian

AU - Hildebran, Christi

AU - Wakeland, Wayne

AU - Deyo, Richard (Rick)

PY - 2018/1/1

Y1 - 2018/1/1

N2 - To develop a simple, valid model to identify patients at high risk of opioid overdose-related hospitalization and mortality, Oregon prescription drug monitoring program, Vital Records, and Hospital Discharge data were linked to estimate 2 logistic models; a first model that included a broad range of risk factors from the literature and a second simplified model. Receiver operating characteristic curves, sensitivity, and specificity of the models were analyzed. Variables retained in the final model were categories such as older than 35 years, number of prescribers, number of pharmacies, and prescriptions for long-acting opioids, benzodiazepines or sedatives, or carisoprodol. The ability of the model to discriminate between patients who did and did not overdose was reasonably good (area under the receiver operating characteristic curve = 0.82, Nagelkerke R 2 = 0.11). The positive predictive value of the model was low. Computationally simple models can identify high-risk patients based on prescription history alone, but improvement of the predictive value of models may require information from outside the prescription drug monitoring program. Patient or prescription features that predict opioid overdose may differ from those that predict diversion.

AB - To develop a simple, valid model to identify patients at high risk of opioid overdose-related hospitalization and mortality, Oregon prescription drug monitoring program, Vital Records, and Hospital Discharge data were linked to estimate 2 logistic models; a first model that included a broad range of risk factors from the literature and a second simplified model. Receiver operating characteristic curves, sensitivity, and specificity of the models were analyzed. Variables retained in the final model were categories such as older than 35 years, number of prescribers, number of pharmacies, and prescriptions for long-acting opioids, benzodiazepines or sedatives, or carisoprodol. The ability of the model to discriminate between patients who did and did not overdose was reasonably good (area under the receiver operating characteristic curve = 0.82, Nagelkerke R 2 = 0.11). The positive predictive value of the model was low. Computationally simple models can identify high-risk patients based on prescription history alone, but improvement of the predictive value of models may require information from outside the prescription drug monitoring program. Patient or prescription features that predict opioid overdose may differ from those that predict diversion.

KW - Addiction medicine

KW - Chronic pain management

KW - Opioid overdose

KW - Opioid prescription

KW - Predictive modeling

UR - http://www.scopus.com/inward/record.url?scp=85044204452&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044204452&partnerID=8YFLogxK

U2 - 10.1097/j.pain.0000000000001078

DO - 10.1097/j.pain.0000000000001078

M3 - Article

C2 - 28976421

AN - SCOPUS:85044204452

VL - 159

SP - 150

EP - 156

JO - Pain

JF - Pain

SN - 0304-3959

IS - 1

ER -