Background: Interest is increasing to evaluate pregnancy outcomes using claims databases by health authorities. However, the extent of the feasibility is unclear.
Objectives: To evaluate the feasibility of identifying pregnancy outcomes in women (16 years and older) with sickle cell disease (SCD) between Jan 2016 to Dec 2018. Three different administrative claims databases were assessed: i) Truven MarketScan Commercial Claims and Encounters (CCAE) ii) Truven MarketScan Multi-state Medicaid (MDCD) iii) Optum ClinFormatics (Optum) database.
Methods: The databases were assessed for accurate identification of patients with SCD, family ID variable, identification of pregnancies, last menstrual period (LMP), continuous enrollment (CE) from 105 days prior to LMP (to account for drug exposure) date to 12 months post-delivery date for at least one pregnancy. SCD was defined using at least one inpatient or two outpatient claims between Jan 2016 to Dec 2018. Databases were searched for discharge diagnoses and medical procedures indicative of pregnancy outcomes [MacDonald et al (2019)] using the Instant Health Data (IHD) platform (Panalgo®). The pregnancy ICD-10 codes (live births, mixed births, spontaneous abortion, elective termination, and unclassified delivery) requiring an interval of at least 29 days between two recorded ones, were defined as pregnancy outcomes. As the LMP date was not recorded, it was estimated by subtracting the standard gestational age from the earliest date of the pregnancy claim.
Results: MDCD database lacks family ID; however, CCAE and Optum have family ID variable enabling linkage of maternal to infant data. Table 1 presents results for the assessment:
Table 1 Comparison of number of SCD patients in three databases Database No. of women No. of pregnant women No. of pregnant women with CE (%) MDCD 6,673 1,814 535 (29.5) CCAE 14,805 667 275 (41.2) Optum 6,786 224 92 (41.1)
Conclusions: The claims databases may be used for pregnancy surveillance. The presence of family ID variable in CCAE and Optum enables linkage of maternal to infant data, thus offering the possibility to evaluate pregnancy and infant outcomes. Limitations of claims databases include the low proportion of patients with CE and the potential for misclassification of exposures and outcomes due to the lack of validation of LMP, pregnancy outcomes, and other variables. An extended feasibility of assessment of claims databases as a potential data source for pregnancy surveillance is needed. Depending on the context and with additional validation to mitigate the known limitations, these databases should be considered as a potential resource, especially in rare diseases, if adequate data are not available from other sources.