The study is based on 2,870,932 singleton births, registered in the Swedish Medical Birth Registry during 1973–2002. Linking this register with the Swedish Cancer Registry, 414 births were identified of women with previously treated invasive breast cancer. Invasive breast cancer was defined according to World Health Organization's histological typing of breast tumors. In total, 331 of the 414 births were the first births subsequent to breast cancer diagnosis, with a mean time between breast cancer surgery and pregnancy of 37 mo (range 7–163). Second and third births subsequent to breast cancer diagnosis were excluded from the analysis in order to analyze one birth linked to each woman exposed to breast cancer who subsequently gave birth. In women without breast cancer, all singleton births were included.
The Swedish Medical Birth Registry of the National Board of Health Welfare has data from over 99% of all births in Sweden since 1973 [6
]. Starting with the first antenatal visit, antenatal, obstetric, and pediatric information are recorded in a standardized manner.
Maternal age was defined as age in completed years at the time of delivery. Parity was defined as number of previous births, including stillbirths. Information about maternal smoking was collected at the first antenatal visit. Women were classified in this study as nonsmokers or daily smokers (one or more cigarettes per day). The woman's country of birth was also recorded, and whether the woman was living with the child's father.
During the period of 1973–2002, three different editions of the International Classification of Diseases were used (ICD-8: 1955–1986; ICD-9: 1987–1990; and ICD 10: 1991–2002).
The variable hypertension was defined as essential hypertension (diagnosed before pregnancy and registered as a separate variable at the first antenatal visit), gestational hypertension, and/or preeclampsia/eclampsia using the following one or more diagnostic codes: ICD-8: 401 and 637; ICD-9: 642; and ICD-10: O10–O16.
Diabetes was defined as insulin-dependent diabetes present before pregnancy, non-insulin-dependent diabetes present before pregnancy, or gestational diabetes (ICD-8: 250; ICD-9: 250 and 648; and ICD-10: O24).
Complications during pregnancy and delivery were classified by a physician when the woman was discharged. For an outcome of pregnancy bleeding we used ICD-8: 632; ICD-9: 640–641; and ICD-10: O20 and O46; and for deliveries complicated by placenta previa, ante- or postpartum hemorrhage, retained placenta, abnormal bone pelvis, fetopelvic disproportion, malpresentation of fetus, prolonged labor, rupture of uterus, laceration of perineum, or other complications we used ICD-8: 651–662; ICD-9: 652–669; and ICD10: O45–O75. To register birth injury we used ICD-8: 772; ICD-9: 767; and ICD-10: P10–P15; and to register both major and minor malformations we used ICD-8: 740–759; ICD-9: 740–759; and ICD-10: Q0–Q99. All diagnostic codes were registered as none or one or more.
Estimated gestational age was based on the first day of the last menstruation and/or on ultrasound examination performed routinely at no later than 18 completed weeks of gestation. During the two last decades, anomaly screening ultrasound has been uniformly performed in Sweden.
The definition of a live birth was a newborn with a gestational age of 22 wk or more that showed any evidence of life, such as beating of the heart, pulsation of the umbilical cord, or definite movement of voluntary muscles. Stillbirth was defined as late fetal death occurring at 28 or more completed weeks of gestation. Preterm delivery was classified as very preterm (≤32 wk) or preterm (33 to 36 wk).
Infants defined as small for gestational age were those with birth weights more than two standard deviations below the mean birth weight for gestational age according to a Swedish reference curve.
Apgar score at 5 min was categorized as (1) below seven or (2) seven or greater because a previous population-based study in Sweden showed that a 5-min Apgar score under seven in term infants was associated with an increased risk of neonatal morbidity and neurologic impairment [7
The study was approved by the Ethical Committee for Human Studies of Uppsala University.
The present study uses a cohort design and logistic regression analysis to estimate odds ratios (ORs) for birth outcome. The exposure of interest was breast cancer prior to birth, and the events adverse birth outcomes as defined above. We considered age of mother, time period of delivery, country of mother's birth, history of infertility, family situation, smoking habits, hypertension, and diabetes to be possible confounders. After studying the distribution of these variables among women exposed to breast cancer those not, we kept age of mother, parity, and year of delivery as variables of interest—both as possible confounders and theoretically also as modifiers of the effect. We started the analysis by stratifying on age of mother, parity, and year of delivery, in periods of 5-y intervals. All crude estimates regarding birth outcome and infant characteristics were adjusted for age of mother (continuous), parity, and year of delivery. When there was a binary response variable, the logistic procedure with canonical (i.e., logit) link in SAS version 8.2 (SAS Institute, http://www.sas.com
) was used to calculate the OR and its 95% confidence interval (CI). When there was a multinomial response variable, the generalized (glogit) link function was used [8
]. No adjustments were made for multiple tests.