Cancer associated with obstetric delivery: Results of linkage with the California cancer registry

Lloyd H Smith, Beate Danielsen, Mark E. Allen, Rosemary D Cress

Research output: Contribution to journalArticlepeer-review

268 Scopus citations


OBJECTIVE: This study provides revised population-based measurements for the occurrence rates of cancer associated with obstetric delivery and examines perinatal and cancer-related outcomes within the group of women with 4,846,505 obstetric deliveries in California, inclusive of the years 1991 through 1999. STUDY DESIGN: This observational study used a population-based retrospective review of cases identified as a result of computer linkage of maternal/neonatal hospital discharge and birth/death records with case files in the California Cancer Registry (CCR). The effect of timing of cancer diagnosis on clinical outcomes was studied by dividing the cases into three groups as follows: "prenatal" for cancer diagnosis within 9 months before delivery, "at delivery" for cancer diagnosis during delivery hospitalization, and "post partum" for cancer diagnosis within 12 months after delivery. Computerized records for 4,846,505 obstetric patients and 4,906,920 newborn infants comprising the linked vital statistics birth/patient discharge database (VS/PDD) were used to identity-match cases within the CCR case files. Cases of malignant disease were categorized into 22 anatomic or histologic subgroups. Perinatal clinical outcomes including preterm delivery, prolonged neonatal hospital stay, stillbirth, neonatal death, frequency of first trimester prenatal care, and cesarean delivery were analyzed by use of International Classification of Diseases, 9th Revision, Clinical Modification codes from the VS/PDD. Clinical cancer outcomes including cancer stage and vital status on follow-up were drawn from CCR records. Statistical comparisons for trends were performed with the Cochran-Armitage test, outcomes comparisons with the Fisher exact test, and survival comparisons were performed with the Cox proportional hazard model. RESULTS: Among 4,846,505 obstetric deliveries, 4,539 cases of invasive malignancy were identified for an observed occurrence rate of 0.94 per 1000 births. Sixty-four percent of the cases occurred post partum; cancers of the breast, thyroid, cervix, along with malignant melanoma, and Hodgkin's disease accounted for 64% of the cases. The timing of cancer diagnosis affected clinical outcomes: for all cancer cases as a group, the most favorable perinatal and cancer outcomes occurred in women whose cancer diagnosis was made 6 to 9 months before delivery (6% of cases). The most unfavorable perinatal and cancer outcomes were associated with cancer diagnosis made 0 to 3 months before delivery (14% of cases). For women whose cancer was diagnosed post partum, perinatal outcomes were minimally affected by the presumed existence of occult cancer at the time of obstetric delivery. CONCLUSION: The use of computer-linkage to the CCR files enhanced identification of cases of maternal malignancy associated with obstetric delivery. Cancer diagnosis was associated with approximately 1 in 1000 deliveries. Most cases were diagnosed after delivery and were comprised predominantly of cancers of the breast, thyroid, cervix, malignant melanoma, and Hodgkin's disease. A small group of women (approximately 1 per 5000 deliveries) are seen within a few months before delivery or at delivery with malignant disease, many of whom have rapidly progressing disease and may require high-risk perinatal and oncology services.

Original languageEnglish (US)
Pages (from-to)1128-1135
Number of pages8
JournalAmerican Journal of Obstetrics and Gynecology
Issue number4
StatePublished - Oct 2003


  • Cancer
  • Clinical outcomes
  • Computer linkage
  • Obstetric delivery
  • Pregnancy

ASJC Scopus subject areas

  • Medicine(all)
  • Obstetrics and Gynecology


Dive into the research topics of 'Cancer associated with obstetric delivery: Results of linkage with the California cancer registry'. Together they form a unique fingerprint.

Cite this