An Observational Study to Develop Computable Algorithms for Identifying Opioid Abuse and Addiction Based on Administrative Claims Data

The most widely available information about patient care and conditions is that contained in medical claims data. If such data can be used to develop a model for identifying patients experiencing prescription opioid abuse/addiction it could be widely applied to patient populations throughout the United States.

A study recently conducted at Group Health comparing International Classification of Disease, Ninth edition (ICD-9) coding for opioid abuse/addiction to textual mentions in clinical notes describing abuse/addiction found that ICD-9 codes were 64% sensitive and 96% specific in their ability to identify patients experiencing opioid abuse/addiction (compared to evidence from clinical notes). This Group Health study considered codes for abuse (305.x) and addiction (304.x) equivalent because clinicians' usage of these codes did not differentiate well between abuse and addiction.

Needed are methods that can accurately identify patients experiencing opioid abuse/addiction based on widely available claims data.

This study will not evaluate opioid misuse because this will be captured by instruments in a prospective study of pain patients (Study 1A) using a combination of adapted validated instruments, and other new instruments that will be evaluated in post-marketing requirement (PMR) Study 2, plus medical record review to supplement questionnaire-based measurement of misuse, abuse and addiction with aberrant behaviors and physician text entries in the medical records.