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 projects, designed in partnership with federal funding organizations, to improve the quality of data used in clinical research. This project is one of two projects designed to link data from PCORnet and the U.S. Food and Drug Administration’s Sentinel medical product monitoring system.
What was the project about?
Zika is a virus that is spread by mosquitoes. Pregnant women who get Zika are at risk of having babies with serious birth defects like microcephaly, or smaller than normal head size. Tracking Zika is an important part of stopping its spread. However, current tracking methods are limited to specific parts of the United States and may miss certain cases.
PCORnet created a shared database system that includes data from patients’ electronic health records, or EHRs. This shared data system includes data from a clinical data research network in Florida that is part of PCORnet. In this project, the team wanted to learn whether the database had enough data to track symptoms of Zika in babies. They also wanted to know if the data were complete and accurate. Last, they wanted to create a new method for searching databases to track Zika.
What were the results?
What data were in the database? The project team found that many necessary data for tracking Zika were in the database. These data included diagnoses of Zika, demographics, and lab test results. However, some data were not in the database. The team had to search for this information in patients’ EHRs. The team grouped other data that wasn’t in the database in terms of how easy the information would be to find. This includes data that were
- Easy to find, like head measurements and whether the baby was born early
- Moderately difficult to find, like hearing screening results
- Very difficult to find, like vision screening results
Were the data in the database accurate? The project team found that data were accurate. For example, only 0.005 percent of babies had a medical record date that was before their date of birth. About 1.3 percent had head measurements that were too big or small to be real.
A doctor reviewed 36 records for babies whose data suggested possible microcephaly. The doctor found that 69 percent of those infants did have microcephaly.
Were the data in the database complete? The project team found that data elements were complete based on a review of 2,800 birth records from one hospital in 2016. For example, 99.8 percent of records had head measurement, and 99.8 percent had information on whether the baby was born early.
Could the team develop a new method to track Zika? The team created a new method to detect Zika by searching medical records from both mothers and babies.
Who was in the project?
The project included EHR data for 182,000 births from one health system in Florida. Birth records were from 2012 to 2017. The project team also looked at medical claims data from the Sentinel system. Sentinel is a tracking system for medical products that is run by the U.S. Food and Drug Administration.
What did the project team do?
The project team used different methods to answer each question. They looked at
- What data the database had for tracking Zika, and what data the project team needed to find in medical records
- How accurate the data were, by looking for errors and having a doctor review diagnoses of microcephaly
- How complete the data were, by looking for records with missing data
- Whether the team could create a new method for tracking Zika using the database
What were the limits of the project?
The data in the database do not contain some key information that may help track Zika. The project team had to review patient medical records to get these data.
Future projects can use the tracking method for other health conditions.
How can people use the results?
Research teams can use these methods when planning studies that use the PCORnet database to identify people with health problems.
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 projects, designed in partnership with federal funding organizations, to improve the quality of data used in clinical research. This project is one of two projects designed to link data from PCORnet and the U.S. Food and Drug Administration’s Sentinel medical product monitoring system.
Objective
To examine the capacity to track congenital Zika syndrome using data contained in clinical data research networks within PCORnet and to create a novel query strategy to track congenital Zika syndrome in the future
Project Design
Design Element | Description |
---|---|
Design | Retrospective descriptive analysis project |
Data Sources and Data Sets | Records from 182,000 births between 2012 and 2017 from OneFlorida Clinical Research Consortium |
Analytic Approach | EHR data review |
Outcomes | Aim 1. Data elements predicting Zika syndrome present in PCORnet CDM and EHRs Aim 2. Data entry error frequency Aim 3. Validation of microcephaly diagnoses Aim 4. Completeness of data for key birth characteristics needed to track birth defects from Zika syndrome Aim 5. EHR query strategy to identify evidence of Zika syndrome in birth records |
The project included electronic health record (EHR) data from 182,000 births linked using the PCORnet common data model (CDM). The CDM organizes data into a standard structure for use by researchers. Birth records were from the years 2012 through 2017. All records were from the OneFlorida Clinical Research Consortium, a PCORnet clinical data research network containing data for 15 million patients across Florida. The team also examined claims data from Sentinel.
The project had five aims:
Aim 1. Determine what data elements predictive of Zika syndrome exist in the CDM and the difficulty of extracting additional data elements from EHRs themselves.
Aim 2. Evaluate the frequency of data error entries in the CDM.
Aim 3. In a subset of birth records, evaluate the validity of data extracted from EHRs for Zika tracking by having a doctor review records and determine if births were accurately diagnosed with microcephaly.
Aim 4. In a subset of birth records, determine the completeness of data elements needed for tracking Zika syndrome.
Aim 5. Develop a novel query to search EHR and Sentinel claims data for evidence of congenital Zika syndrome.
Results
Aim 1. The project team found that many data elements necessary for detecting Zika syndrome existed in the CDM, such as diagnoses, demographics, and laboratory test results. The team categorized other data elements not currently in the CDM in terms of how difficult they would be to extract from EHRs:
- Relatively easy to extract, such as head circumference and estimated gestational age
- Moderately difficult to extract, such as hearing screening
- Very difficult to extract, such as results of vision screening
Aim 2. The project team found that 0.005% of observations within the CDM had a date that preceded patient date of birth, and 1.3% had out-of-plausible-range head circumference measurements.
Aim 3. Based on a doctor’s review of 36 records from one hospital, 69% of infants flagged as possible microcephaly cases had the condition.
Aim 4. The project team found that data elements needed for Zika tracking were highly complete based on a review of 2,800 birth records from one hospital in 2016. For example, 99.8% of records had head circumference data, and 99.8% had estimated gestational age at birth.
Aim 5. The team developed a query strategy to detect congenital Zika syndrome using SAS software. The strategy queries EHR and hospital claims data from both mothers and infants.
Limitations
The CDM data do not contain key data elements that may assist in detecting Zika syndrome, such as head circumference. The project team had to review underlying patient records to retrieve these data.
Conclusions and Relevance
The project team examined CDM data to identify important data elements that can be used to track Zika syndrome. The team examined the completeness and validity of data and created a search strategy to track cases of congenital Zika syndrome using EHR and claims data. Future projects can consider using CDM data to track other diseases and health conditions.
Future Needs
Future projects could investigate the accuracy and completeness of hospital claims data for use in tracking congenital Zika syndrome. Future projects could also extend the tracking method to other healthcare topics.