A data extraction algorithm for assessment of contraceptive counseling and provision

Area of Science:

• Reproductive Health
• Health Informatics
• Preventative Medicine

Background:

• Contraception counseling and provision are essential preventative services requiring real-time assessment for quality improvement.
• Direct observation is infeasible; indirect measures like chart review are acceptable.
• Computer-aided chart review using electronic medical records (EMR) offers efficiency and shareability.

Purpose of the Study:

• To create a comprehensive and accurate data extraction algorithm for assessing contraception counseling and provision rates in outpatient settings.
• To leverage EMR data for efficient and repeatable quality assessment.

Main Methods:

• Developed a data extraction algorithm using structured query language on EMR data from 8 outpatient clinics (August 2015-May 2016).
• Included nonpregnant women aged 14-45. Algorithm queried prescriptions, ICD codes, CPT codes, and search terms.
• Algorithm refined through 3 rounds of comparison with manual chart review to achieve high sensitivity and specificity.

Main Results:

• Analyzed 22,134 visits; 56.9% showed evidence of contraception counseling/provision.
• Algorithm captured 69.7% via search terms, 21.3% via prescriptions, and 8.9% via ICD codes.
• Sensitivity and specificity improved from 0.79/0.85 to 0.99/0.98 after refinement.

Conclusions:

• Data extraction algorithms are effective for computer-aided chart review of contraception services.
• Search-term capture from unstructured data is crucial for comprehensive assessment, especially for counseling without provision.
• The developed algorithm can be used by other institutions for real-time assessment, quality improvement, and comparative studies of contraceptive services.