Study Population
We assembled a cohort of females with at least one pregnancy episode and
had continuous enrollment from 30 days before pregnancy start date to 6
weeks after pregnancy end date, allowing for multiple pregnancy episodes
per female. We used an algorithm that was developed and validated in
multiple U.S. and U.K. administrative databases.20Claims for pregnancy markers, procedures and outcomes were identified
from diagnosis, procedure, DRG, HCPCS, and laboratory test codes (Table
S1).20-24 First, pregnancy outcome was assigned to
each pregnancy episode. Pregnancy end dates were assigned using the
first occurrence of the most reliable outcome code. Pregnancy start
dates were then estimated using gestational age claims and pregnancy
markers. Live birth episodes were retained for this analysis. We linked
mothers and newborns by matching on the offspring’s earliest date of
insurance coverage within 10 days of the mother’s live birth claims
code.25 Mother-to-infant linkage was successful for
90% of live births and did not differ by maternal cancer status.
Mothers without a matched infant were retained in the dataset. Plurality
was determined by number of infants matched to each pregnancy episode.
(Table S1).
We assembled a cohort of female AYA cancer survivors querying for cancer
codes (Table S2),26 excluding non-melanoma skin. We
required > 2 diagnostic codes pertaining to the same
cancer site within 12 months.27 To identify new cancer
diagnoses, we required an observed 6-month period of continuous
enrollment without another cancer diagnosis prior to the index cancer
diagnosis claim date. We retained those of AYA age (15-39) at first
cancer diagnosis claim.
Merging the pregnancy and AYA cancer survivor cohorts, we generated a
cohort of AYA cancer patients and females without AYA cancer who
delivered live births (eFigure 1). For live births in AYA cancer
survivors, pregnancy start dates that occurred after the index cancer
diagnosis claims date were considered exposed. We retained the first
live birth after cancer diagnosis for AYA cancer survivors and the first
live birth episode in the dataset for each female without cancer that
occurred between the ages 15-50 years of age.