Study population

The study cohort was identified in the Medical Birth Register and included all firstborn singleton infants of native Swedish parents, born after gestational week 34+0 during the years 1990-1993. In 1990, Sweden introduced pregnancy dating by ultrasound as a routine, and children born after 1993 did not have their school grades forwarded to the National School Register. Children delivered before gestational week 34 were excluded because this is considered inappropriate use of vacuum extraction in Sweden. For the child to be included in the study, his or her mother had to be alive and a resident of Sweden when the child reached 15 years of age (n = 148,881).

From this population we excluded infants with a malformation with possible clinical significance (n = 4558), infants delivered by forceps (n = 1369), and those born in breech presentations (n = 4399). Furthermore, we excluded children without a registered grade point average (n = 4691) or mathematics test score (n = 7832) at the age of 16 years. The definition of serious malformation was based on the diagnoses made by the attending pediatrician. It included all malformations with the exception of nondescended testicles, preauricular appendage, and minor malformations of fingers and toes.

The study population of 126,032 children was followed up at 15-16 years of age, at graduation from compulsory school (at the end of ninth grade) between 2006 and 2009.

Exposure variables

The exposure variables, modes of delivery, were collected from the Medical Birth Register and categorized into noninstrumental vaginal delivery, meaning a vaginal delivery without the use of any instrument, vacuum extraction, nonplanned cesarean section, and planned cesarean section. A cesarean section was defined as nonplanned if the operation was conducted after the onset of labor, either spontaneous or by induction, and defined as planned if it was performed before the onset of labor.

Pregnancy, birth, and perinatal variables

Information about the mother’s age at birth, mode of delivery, smoking habits in early pregnancy, maternal diseases, and complications during pregnancy and labor was collected from the Swedish Medical Birth Register, a national register with high-quality data on more than 99% of all births in Sweden. From the same register, perinatal data were collected about sex, gestational age, birthweight, and head circumference, whether the child was small for gestational age (SGA; <2 SD) or large for gestational age (LGA; >2 SD ); cephalic hematoma, intracranial bleeding, central nervous system complications, convulsions after birth, and Apgar scores at 5 minutes. Maternal diseases and pregnancy and labor complications, as well as neonatal diagnoses, were coded according to the International Classification of Diseases , ninth revision (ICD-9).

Socioeconomic variables

Data on parental education were collected from the Swedish Educational Register and defined as the highest formal education attained in the household 1 year before the child’s graduation. The educational level was categorized by years of education into less than 10 years (compulsory), 10-14 years (high school), and more than 14 (university). Data on social welfare recipiency and household disposable income were collected from the Total Enumeration Income Survey and was categorized into quintiles in which quintile 1 included the 20% of children in families with the lowest incomes. Residency was retrieved from the Register of the Total Population and categorized into big city, small city, and rural.

Maternal morbidity

Data from the Swedish Hospital Discharge Register 1973-2008 were used to create dichotomized variables associated with psychiatric and addictive disorders in the mother. Illicit drug and alcohol abuse (yes/no) indicates whether the mother had been discharged from a Swedish hospital with a diagnosis associated with use or abuse of alcohol or illicit drugs in ICD-9. Psychiatric disorder (yes/no) shows whether the mother had been discharged at least once from a Swedish hospital with a diagnosis in the psychiatric chapter in ICD-9.

Outcome variables

The study population was followed up at 15-16 years of age using 2 outcome variables: mean grade point average and the mean score on national mathematics tests. Data were collected from the National School Register kept by Statistics Sweden. The quality of the data in the National School Register is high, and summary statistics are published on a regular basis. The majority of all children (more than 90%) attend compulsory school, which lasts 9 years, usually between the ages of 7 and 16 years.

The register contains information on all school grades at the final examinations and graduation grades for each pupil leaving Swedish compulsory school. From 1998 onward, the Swedish school system has used alphabetic grades with 4 levels: fail, pass, pass with distinction, and pass with special distinction. Grades are awarded on a 3-level scale: a pass is counted as 10 merit points, a pass with distinction as 15 merit points, and a pass with special distinction as 20 merit points. The maximum merit rating is 320 points. Before selection for upper secondary schools, the pupil’s final school-leaving grade (ie, mean grade point average) is calculated as the sum of the points of the pupil’s 16 best grades on the leaving certificate.

The second outcome, the mathematics test score, was based on the sum of 4 subsample national tests in mathematics that all pupils attending the final year of compulsory school write. The maximum mathematics test scores in the years 2003-2006 ranged from 68 to 75.

Statistical methods

We conducted linear regression analyses using SPSS version 18.0 for Windows (SPSS Inc, Chicago, IL) to estimate crude and adjusted mean differences in mean grade points and mean mathematics grades with 95% confidence intervals in relation to mode of delivery. Noninstrumental vaginal delivery served as the reference group in the analyses.

To control for potential confounders, we used the following 3 models: model 1 was adjusted for year of graduation and sex only. For model 2 we added the following socioeconomic, demographic, and maternal morbidity variables to the confounders in model 1; maternal age, highest educational level in the household 1 year before graduation, urban/rural residency 1 year before graduation, single-parent household, maternal smoking, maternal drug abuse, maternal alcohol abuse, and maternal psychiatric diagnosis were added in model 2. In the final model, we added potential medical confounders to the variables in model 2: SGA, LGA, oligohydramnion, polyhydramnion, maternal diabetes, preeclampsia, head circumference, and gestational age.

To further control for residual perinatal confounding, subgroup analyses were conducted on a selected population of term children (gestational week >36), with Apgar score (>6) at 5 minutes, who were born without a diagnosis of fetal distress. The same linear regression models were used to test for the association mode of delivery and school grades (not shown in a table).

Our main analyses are based on children who graduated from compulsory school with registered mean grade point averages and mathematics test results. The children with no available grade records were analyzed separately to assess possible selection bias.