Perinatal risk for infants of unmarried mothers over a period of 20 years.

The perinatal events of the infants of 444 unmarried mothers, 3.7% of the total Northern Finland birth cohort from 1966, were compared with those of infants of 11,525 married mothers (95.5%), and a similar comparison was made between 395 (4.2%) unmarried mothers and 7516 (80.3%) married mothers in a second Northern Finland birth cohort in 1985-86. 1336 mothers, 14.3% of the mothers in this later cohort, were cohabiting. Divorced and widowed mothers were excluded from both cohorts. The infants of the unmarried mothers had a significantly lower mean birth weight, were more likely to be small for their gestational age (SGA), of low birth weight (LBW) (below 2500 g) and had a higher incidence of pre-term births than those of the married mothers in both cohorts. Perinatal mortality was significantly higher among the unmarried mothers only in the former cohort. These differences in perinatal events diminished markedly after adjustment for maternal age, parity, height, years of schooling and smoking habits, but did not totally disappear. The difference in the incidence of pre-term births diminished, but remained significant in both cohorts. The difference in mean birth weight, in the incidence of LBW infants and in perinatal mortality remained significantly less favourable to the unmarried mothers only in the 1966 cohort. It seems that the gap between the married and unmarried mothers had diminished. The incidence of SGA infants did not differ significantly between the married and unmarried mothers in either cohort after adjustment for the background variables. The cohabiting mothers formed an intermediate group between the married and single unmarried mothers in respect of perinatal events, but were close to the married mothers. In raw figures, the mean birth weight in this group was significantly lower and the incidence of SGA infants higher than among the married mothers, but these differences also disappeared after adjustment of the background variables.