Results Summary
PCORI funded the development of PCORnet®, the National Patient-Centered Clinical Research Network, to make research faster, easier, and less costly to conduct. PCORnet is made up of Partner Networks of healthcare systems, patients and communities, and health plans that harness the power of large amounts of health data.
PCORI supports brief, descriptive projects to assess the feasibility of conducting research using data gathered and shared securely through PCORnet. This project is one of several designed to test the network while addressing priorities identified by PCORI and its stakeholders.
What was the project about?
Opioid dependence is a public health crisis. Opioids can treat cancer pain effectively and safely. But people have a higher risk of becoming dependent on opioids when they get prescriptions for opioids that they don’t need, when they have other health problems, or when they take other medicines that affect how opioids work. PCORnet created a shared database system that includes information about patients and opioid prescriptions. The database includes information from patients’ electronic health records, or EHRs. It also includes data from insurance claims, state information about prescription medicines, results of drug tests, and information about people’s cause of death. The project team looked at the database to answer two questions:
- Can researchers use the database to identify patients at risk for becoming dependent on opioids?
- Can researchers use the database to identify risk factors, such as age, diagnosis, or treatment, related to opioid dependence?
What were the results?
Can researchers use the database to identify patients at risk for becoming dependent on opioids? The team collected data about patients, their diagnoses, and what kinds of treatment they received. They also looked at how doctors prescribed opioids and tests and treatments for opioid dependence. Using the database, the team determined that about 19 percent of all patients had opioids in some form in 2011. In 2017, this number dropped to 16 percent. Among patients without cancer, the rate was 18 percent in 2011 and 15 percent in 2017.
Can researchers use the database to identify risk factors related to opioid dependence? The database had enough data for the team to assess how likely patients were to get opioids based on factors such as age, race, and diagnosis. They also compared patients with fatal and non-fatal overdoses based on factors such as long-term opioid use.
Who was in the project?
The project team gathered data for more than 15 million patients from nine hospitals and clinics. They separated data for patients who had cancer and those who didn’t.
What did the project team do?
The team used information from the shared database about patients who received prescriptions for opioids. The team looked for patterns in the patient and prescription data.
What were the limits of the project?
Some hospitals and health centers used both old and new database codes for the same types of care, so some data were duplicates. The project team is identifying ways to fix this issue in the future.
How can people use the results?
Researchers can use data gathered by PCORnet to identify prescriptions, diagnoses, and types of health care that can lead to opioid dependence.
Professional Abstract
PCORI funded the development of PCORnet®, the National Patient-Centered Clinical Research Network, to make research faster, easier, and less costly to conduct. PCORnet is made up of Partner Networks of healthcare systems, patients and communities, and health plans that harness the power of large amounts of health data.
PCORI supports brief, descriptive projects to assess the feasibility of conducting research using data gathered and shared securely through PCORnet. This project is one of several designed to test the network while addressing priorities identified by PCORI and its stakeholders.
Objective
To assess the feasibility of using PCORnet for surveillance of patient and provider risk factors related to the opioid epidemic
Project Design
| Design Element | Description |
|---|---|
| Design | Retrospective descriptive analysis project |
| Data Sources and Data Sets | Data from 2010 to 2017 for adult patients from 9 clinical sites in 3 PCORnet networks |
| Analytic Approach | Review of EHR data, Medicare claims, medical examiner reports, toxicology tests, prescription drug monitoring programs, and emergency medical services data |
| Outcomes |
Aim 1: Identify data elements in PCORnet CDM available for monitoring opioid dependence risk factors. Aim 2: Identify patient- and provider-related risk factors associated with opioid use. |
The project team had two aims:
- Aim 1. Identify the data elements in the PCORnet Common Data Model (CDM) that can be used to create a profile of opioid dependence risk factors, clinical processes, and outcomes.
- Aim 2. Use CDM clinical data to assess patient- and provider-related risk factors related to opioid dependence and assess the extent to which clinical policies or processes lead to reductions in opioid use.
The team compiled electronic health record (EHR) data along with insurance claims, results from prescription drug monitoring programs, toxicology tests, and medical examiner reports and added them to the CDM. They identified data elements in the CDM that researchers could use to measure opioid-related clinical activity and risks. These data elements included high-risk diagnoses or acute or chronic pain, prescribing patterns associated with increased risk for opioid dependence, outcomes such as overdose or hospitalization, and processes such as medication-assisted treatment or admission to substance abuse treatment facilities. The team also established criteria for three cohorts with different opioid use trajectories: patients with a cancer diagnosis, patients receiving inpatient cancer treatment, and patients who received opioids for non-cancer reasons. The team tested a data query with CDM data from nine clinical sites representing more than 15 million patients across three Clinical Data Research Networks (CDRNs).
Results
Aim 1. The team used patient demographic and diagnostic data; data on opioid prescribing patterns among clinicians and health systems; and data related to opioid prescriptions, testing, and treatment for dependence. Using data in the CDM, the team determined that 19% of all patients in the three cohorts were prescribed opioids in 2011; this decreased to 16% in 2017. Among patients without cancer, opioid exposure was 18% in 2011, and it decreased to 15% in 2017.
Aim 2. CDM data were adequate to assess risk of opioid exposure by patients’ age, sex, race, ethnicity, and other factors such as diagnosis. The team compared fatal and non-fatal overdoses with risk factors such as chronic opioid use. In addition, the team identified ICD-9 and ICD-10 codes that researchers could use in future analyses to study the types of care associated with opioid dependence risk.
Limitations
Some sites returned duplicate data using ICD-9-CM and ICD-10-CM/Procedure Coding System codes. The team is addressing this issue before making the data query available for broader use.
Conclusions and Relevance
The project team was able to use the PCORnet CDM to identify prescriptions, diagnoses, and types of clinical care associated with opioid use and perform separate analyses for patients with and without a cancer diagnosis.
Future Needs
Future projects could help clinical sites to expand the details captured about opioid prescriptions, such as formulations, strength, and units of measurement. These details may improve surveillance.