Prevalence of morbidity associated with abortion before and after legalisation in South Africa

BMJ 2002; 324 doi: (Published 25 May 2002) Cite this as: BMJ 2002;324:1252
  1. Rachel JewkesGender and Health Group (rachel.jewkes{at}, directora,
  2. Heather Brown, specialist obstetricianb,
  3. Kim Dickson-Tetteh, clinical director, Reproductive Health Research Unitb,
  4. Jonathan Levin, senior statistician, Biostatistics Unita,
  5. Helen Rees, executive director, Reproductive Health Research Unitb
  1. a Medical Research Council, Private Bag X385, Pretoria 0001, South Africa
  2. b Chris Hani Baragwanath Hospital, PO Bertsham 2013, Johannesburg, South Africa
  1. Correspondence to: R Jewkes
  • Accepted 7 November 2001

Abortion on request has been legal in South Africa since 1997. The Choice in Termination of Pregnancy Act of South Africa 1996 allows abortion on request up to 20 weeks' gestation. Since it was introduced, 40 000 legal abortions have been performed annually.1 A national study in 1994 on morbidity associated with incomplete abortion (illegally induced and spontaneous miscarriage) assisted the act's passage through parliament.2

Characteristics of women, clinical findings on admission relating to abortion, and changes in hospital management, South Africa, 1994 (n=308) and 2000 (n=761). Values are percentages.

View this table:

We studied the impact of legislative change on morbidity and medical management by repeating the 1994 study of morbidity due to incomplete abortion among patients presenting to public hospitals in 2000.3

Participants, methods, and results

Over different three week periods between May and August, we collected data on all women presenting to selected public hospitals with incomplete abortions under 22 weeks' gestation. We excluded legally induced and threatened abortions. The sampling frame included all public hospitals in the nine provinces of South Africa responsible for treating women with gynaecological problems in 2000. The sample was stratified randomly by province and category of hospital (district, regional, and tertiary). In each stratum, two hospitals were selected with the sampling probability proportional to size (number of beds). Our sample consisted of 47 hospitals, as five of the provinces had one or no hospital in the tertiary stratum. A data capture sheet for each woman was completed from the hospital records by healthcare staff.

We used three clinical severity categories for data analysis and interpretation.4 Calculations were based on population estimates for 1999 of 13 478 000 women aged 12-49 years and 1 106 000 live births, where appropriate. The analysis took into account the complex sample design, which was a stratified multistage sample and was not self weighting. We used the Rao Scott F test (part of the Stata package) to compare the categorical variables with the 1994 study. 2 5

The methods of the two studies differed in the sampling of the hospitals. In 1994 the sample was stratified only by the number of beds: all hospitals with over 499 beds and a random sample of hospitals with under 500 beds were sampled.

All 47 sampled hospitals responded, returning a total of 761 data capture sheets. In 1994, 803 data capture sheets were returned. Three hospitals had no cases of abortion during the study period. The incidence of incomplete abortion per 100 000 women aged 12-49 years was 362 (range 282-441) compared with 375 (299-451) in 1994 (difference 13 (95% confidence interval Embedded Image123 to 97) per 100 000). The rate of incomplete abortion per 100 000 live births was 44 (34-54) compared with 42 (33-50) in 1994 (difference 2 (Embedded Image11 to 15) per 100). There was one death in the 2000 study period compared with three in the 1994 study.

The table shows the characteristics of the women, clinical findings on admission, and changes in hospital management. In 2000, only 7.8% of transfusions were given to women with haemoglobin concentrations of over 86 g/l. Antibiotics were not given to 55.5% and 52.0% of women in the medium and high severity categories, respectively.


Legalisation of abortion in South Africa immediately decreased morbidity but the magnitude was not substantial, possibly because morbidity was already lower than in many countries. The lack of change may reflect additional covert induced abortion activity, perhaps through the use of misoprostol in unregistered settings. There has been a trend towards lower technology. While more manual vacuum aspiration and less general anaesthetic and blood transfusion is commendable, antibiotic use and pain relief seem inadequate. The trend towards lower technology partially reflects success of training programmes for induced abortion; however, our findings suggest that further structured training in the use of manual vacuum aspiration with paracervical block and appropriate use of antibiotics and misoprostol would be beneficial.


Contributors: KD-T, HR, RJ, and JL wrote the proposal for the study. HB and KD-T managed the research and quality control. RJ, JL, HB, KD-T, and HR wrote the paper and project technical reports. RJ is guarantor.


  • Funding South African National Department of Health, Kaiser Family Foundation, and UK Department for International Development.

  • Competing interests None declared.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract