(CSEMP062) VALIDITY OF ADMINISTRATIVE HEALTH DATA CASE DEFINITIONS FOR IDENTIFYING POLYCYSTIC OVARY SYNDROME: A SYSTEMATIC REVIEW

Abstract:

Background:
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in females of reproductive age. Administrative health data offer the opportunity to evaluate health outcomes and disease epidemiology at a population-level, but variable administrative health data case definitions for PCOS have been used to date.

Objective: To estimate the validity of published administrative health data case definitions of PCOS compared with reference standards.

Methods:
Studies published up until July 4, 2023 were searched from Medline and Embase. Abstracts and full texts were reviewed in duplicate using Covidence. Studies were included if they compared the accuracy of codes for PCOS in an administrative database with a reference standard and had validity indices (sensitivity, specificity, positive predictive value (PPV), negative predictive value or measure of agreement). Original studies in English with human subjects and at least five PCOS patients (and ≥30 patients overall) were included. Quality was determined using the Quality Assessment for Diagnostic Accuracy Studies 2 tool. A meta-analysis was planned if there were sufficient data available.

Results:
Of 6,266 screened abstracts and 70 screened full texts, four studies met inclusion criteria. All four studies compared administrative health data definitions of PCOS to chart review as a reference standard. Two studies defined PCOS using Rotterdam Criteria, one study used self-report, and one used a clinical gold standard. All case definitions included the ICD-9 code 256.4 in the PCOS case definitions presented and three case definitions also included ICD-10 code E28.2. Two case definitions included ICD codes for related disorders, such as hirsutism, female infertility, or irregular menses. The range of ICD codes used in the case definitions was 2-17. Of the three studies that reported PPV for case definitions, PPV ranged from 68% to 96%. One study did not report PPV but found high agreement (Percent agreement= 90.3, κ=0.21 and percent agreement bias adjusted κ=0.81). Overall, the risk of bias of included studies was low. There were insufficient data for meta-analysis.

Conclusions:
We identified four studies that reported the validity of one or more administrative health data case definitions for PCOS in administrative databases compared with reference standards. The majority of case definitions for PCOS had a high PPV (>70%), suggesting that these case definitions are valid. Further validation of these case definitions is required in other administrative health databases to assess and compare their external validity for use in population health studies.