Effect of parity, gravidity, previous miscarriage, and age on risk of Down's syndrome: population based study

BMJ 1998; 317 doi: (Published 03 October 1998) Cite this as: BMJ 1998;317:923
  1. Annabelle Chan, senior medical consultant, pregnancy outcome unit (chan.annabelle{at},
  2. Kieran A McCaul, head, health statistics unita,
  3. Rosemary J Keane, midwife, pregnancy outcome unita,
  4. Eric A Haan, directorb
  1. aEpidemiology Branch, South Australian Health Commission, PO Box 6, Rundle Mall, Adelaide, South Australia 5000, Australia
  2. bSouth Australian Birth Defects Register, Department of Medical Genetics and Epidemiology, Women's and Children's Hospital, North Adelaide, South Australia 5006
  1. Correspondence to: Dr Chan
  • Accepted 27 February 1998

The increased risk of Down's syndrome with maternal age underlies the recommendation for older pregnant women to be offered screening by amniocentesis or chorionic villus sampling. Recently Schimmel et al suggested that increased parity was an independent risk factor for Down's syndrome, but their study was not population based and did not include terminations of pregnancy.1

Using statewide statistics on births and terminations of pregnancy we investigated whether the risk of Down's syndrome is increased independently of maternal age by maternal parity, gravidity, or previous miscarriage.

Subjects, methods, and results

South Australia has around 20 000 births annually. The state collects data on birth defects that include maternal characteristics in both its perinatal and abortion statistics. These statutory collections are complemented by notifications from health professionals to the South Australian Birth Defects Register of defects in children detected within the first 5 years of life and by cytogenetic and necropsy reports. Each case of Down's syndrome included in these collections has been cytogenetically confirmed.

The effects of parity, gravidity, number of previous miscarriages, and mother's age (by single year of age) on risk of having a fetus with Down's syndrome were modelled separately using Poisson regression; then the effects of parity, gravidity, and previous miscarriage were modelled separately after adjustment for the effect of mother's age. Overdispersion was detected in all the Poisson models constructed, and an overdispersion factor was estimated using the square root of Pearson's χ2 divided by the number of degrees of freedom.2 The analysis was performed using PROC GENMOD in SAS.3 Analyses were undertaken for 1986-95 and 1986-90, which was similar to the period of study of Schimmel et al (1981-9) and preceded the gradually increasing use of maternal serum screening for Down's syndrome.

Analysis using births and terminations of pregnancy showed no significant increase in risk for increase in parity or gravidity (table). When only births were analysed for 1986-95, the increased risks with increase in parity (P<0.001) and gravidity (P<0.01) were not significant after adjustment for age (P=0.46 and P=0.75 respectively); similar results were obtained for 1986-90 for increase in parity. The risk was not increased with the number of previous miscarriages, but the increase in risk with age was constant (P<0.001).

Relative risk (per unit increase in variable) of Down's syndrome for age, parity, gravidity, and previous miscarriage, 1986-95 and 1986-90, South Australia

View this table:


Schimmel et al studied live births alone among women of high parity attending one hospital. Their analysis of maternal age by five year age groups might have contributed to spurious results through truncation.1

The low risk found in women of low parity may have resulted from the inadvertent exclusion of women of low parity who had had Down's syndrome diagnosed in their fetus prenatally elsewhere and spontaneously miscarried or terminated the pregnancy.1 Another Australian population based study found that older pregnant women who had had three or more previous births were less likely than those of lower parity to undergo amniocentesis or chorionic villus sampling, although all were eligible for a medical service rebate.4 We found a similar differential use of prenatal diagnosis in univariate analysis among South Australian women who were pregnant in 1991-6 (odds ratio 0.55 (95% confidence interval 0.51 to 0.61)) and also found a similar higher use of these tests among women who had had a previous termination of pregnancy. These and other differences identified may reflect different degrees of knowledge about the availability of tests, concern about the possibility of having a disabled child, or attitudes towards termination of pregnancy.4

With the operation of selection factors and increasing use of prenatal diagnosis, risk estimates of Down's syndrome need to be based on population data that include births and terminations of pregnancy.


We acknowledge the role of South Australian midwives and neonatal nurses in providing perinatal data and of doctors in providing data on congenital abnormalities; the contribution of staff of the South Australian Births Defects Register and the Pregnancy Outcome Unit in processing and collating data on babies with Down's syndrome, births, and terminations of pregnancy; and the staff of the department of cytogenetics and molecular genetics, Women's and Children's Hospital, and of the department of cytogenetics, Queen Elizabeth Hospital, for providing cytogenetic data.

Contributors: AC initiated and planned the study, reviewed the literature, performed the univariate analysis, participated in the interpretation of data, and wrote the paper. KAM performed the Poisson regression analysis and participated in the planning of the study, the interpretation of data, and writing the paper. RJK participated in the planning of the study, the validation of some case details, the retrieval and preparation of data for analysis, and editing the manuscript. EAH participated in planning the study, interpreting the data, and editing the manuscript. AC and KAM are guarantors for the paper.


  • Funding None

  • Conflict of interest None


View Abstract