Background: Use of administrative data to examine the relationship of exposures to pregnancy requires pregnancy start and end dates. Information regarding the accuracy of the use of claims to estimate of these dates is needed.
Objectives: To assess the performance of an algorithm for identification of the pregnancy start and end dates within a commercial insurer's administrative database as compared to dates found in medical records.
Methods: In a retrospective study of pregnant women with psoriasis or chronic inflammatory arthritis, and a general population comparator group, an 8.5% random sample, stratified by pregnancy outcome, was identified from a large claims database using STORK (Systematic Tracking of Real Kids), a process to identify pregnancies and link mothers and babies in administrative claims. Medical charts were sought to validate outcomes and pregnancy dates. We identified a sample of 457 pregnancies, representing claims for 322 live births, 117 non-live births, and 20 unknown outcomes.
Results: Overall, 219 chart-based pregnancy end dates were available from the charts of 300 pregnancies for date-validation. Of these, 216 (98.6%) had an end date within 34 days of the claims-based end date, and 96.8% within 21 days. Pregnancy start dates were available in 266 charts where 240 (90.2%) were within 34 days of the claims-based conception dates and 229 (86.1%) were within 21 days. Pregnancy start dates within 21 days were validated for 100% of pre-term deliveries and 91.6% of single live births, but were less accurate for abortions, ranging from 72.7 for elective to 79.3% of spontaneous abortions within 21 days and from 75.0% of ectopic pregnancies to 86.2% of spontaneous abortions within 42 days. Pregnancy conception dates for multiple live births were validated for 48.3% within 21 days to 86.5% within 42 days.
Conclusions: Overall, our dating algorithms performed well in identifying pregnancy end dates for most outcomes and for calculating conception dates for single live births. Further refinement may improve dating of early terminations and further work is needed for pregnancies with multiple births.
