The problem of illegally induced abortion: results from a hospital-based study conducted at district level in Dar es Salaam


correspondence Vibeke Rasch, Department of Obstetrics and Gynaecology, Odense University Hospital, DK 5000 Odense C, Denmark. E-mail:


Summary Illegal abortion is known to be a major contributor to maternal mortality. The objective of the study was firstly to identify women with illegally induced abortion, (IA) and to compare them with women admitted with a spontaneous abortion (SA) or receiving antenatal care (AC), and secondly to describe the circumstances which characterized the abortion. The population of this cross-sectional questionnaire study comprised patients from Temeke District Hospital, Dar es Salaam, Tanzania. After an in-depth confidential interview, 603 women with incomplete abortion were divided into two groups: 362 women with IA and 241 with SA. They were compared with 307 AC women. IA women were significantly younger, more often better educated, unmarried, nulliparous and students than AC women. Regarding civil-status, educational level, proportion of nullipara and proportion of students, SA patients were similar to AC women. These results lend support to the assumption that the in-depth confidential interview made it possible to distinguish IA women from SA women.


According to the World Health Organization about 585000 women die each year during pregnancy or immediately after giving birth or after an induced abortion. Over 99% of these maternal deaths occur in low-income countries ( WHO 1997). Unsafe abortion – a procedure performed either by a person lacking the necessary skills or in an environment lacking minimal medical standards or both – is a major contributor to maternal mortality. Unsafe abortions are responsible for 50000–100000 preventable deaths each year world-wide ( WHO 1994). In addition there are millions of women who are suffering from chronic morbidities and disabilities as a consequence of unsafe abortion.

Throughout sub-Saharan Africa, induced abortion is highly restricted, with only a few countries permitting abortion unless the woman's life is threatened by the pregnancy. Even where laws are liberal, the availability of services is so poor and the requirements for an elective procedure are so great that many abortions continue to be clandestine and unsafe ( Rogo 1993; Sims 1996; Machungo et al. 1997a ).

Data on the extent of induced abortion in sub-Saharan Africa are inconsistent and substantial under-reporting is a major problem. In Zaire, 15% of ever-pregnant women aged 13–49 reported having had an induced abortion ( Shapiro & Tambashe 1994). In Kenya, South Africa, Nigeria and Uganda the incidence among adolescents varied from 1.5% to 29% ( Ajayi et al. 1991 ; Agyei & Epema 1992; Brabin et al. 1995 ; Buga et al. 1996 ).

Data collected from women with clinically suspected induced abortion are notoriously unreliable. In countries where the intervention is illegal, women who have an induced abortion can hardly be expected to admit to it. Thus most studies dealing with illegal abortion have been based upon a classification method using clinical findings, whereas we found classification based upon an empathic interview more fruitful ( Rasch et al. 2000 ).

In Tanzania, abortion is illegal with few exceptions, and the exact magnitude of the problem of induced abortion is unknown. The prevalence of complications among women hospitalized for an incomplete abortion indicates that such terminations are common ( Justesen et al. 1992 ). In 1993, Mpangile et al. found that of 965 women with incomplete abortion, 455 (47%) had an induced abortion ( Mpangile et al. 1993 ). We have found that as many as 60% of allegedly spontaneous abortions in Dar es Salaam are in fact induced pregnancy terminations ( Rasch et al. 2000 ). In Ilala District, Dar es Salaam 1991–93, Urassa et al. (1996) showed that unsafe induced abortion caused 15% of maternal deaths.

The contraceptive prevalence rate in Tanzania is low. According to the Tanzania Knowledge, Attitudes & Practices Survey 1994 (KAP), 80% of Tanzanian women knew of a contraceptive method, but only 11% were currently using modern contraception and 21% ever had done so ( Weinstein et al. 1995 ). Adolescents are in a special situation. According to public opinion, young girls should abstain from sexual relations or get married soon after menarche. However, as societies in Africa change and adopt more and more of Western culture, teenage conduct becomes unpredictable and teenagers are freer than ever to engage in sexual affairs ( Liljestrom et al. 1998 ). Until recently population policy stated that contraception should be used for child spacing only. Hence, family planning information and advice was given primarily to married clients with at least one child. In the early 1990s authorities began to acknowledge the problem of pregnancies among adolescents, and population policy was changed in 1994. The guidelines now state that ‘all males and females of reproductive age, including adolescents irrespective of their parity and marital status, shall have the right of access to family planning information, education and services’ ( Ministry of Health 1994).

The fact that so many women risk death, injury and social or criminal consequences to terminate a pregnancy clearly demonstrates how desperately they wish to delay or avoid having children. Hence women who have undergone abortion constitute an important group with unmet family planning needs. The purpose of our study was firstly to identify women with illegally induced abortion (IA) and evaluate whether an empathic interview method makes it possible to separate IAs from SAs (women with spontaneous abortion). Secondly, we wanted to compare IA women with SA women and antenatal care (AC) patients regarding their socio-economic and reproductive characteristics. Our third objective was to describe the circumstances of the abortion.

Subjects and methods

Dar es Salaam, the capital of Tanzania, has about 2 million inhabitants ( Bureau of Statistics 1992) and consists of three districts (Kinondoni, Ilala and Temeke). Each district has a district hospital and the city has one referral hospital. Three categories of women were recruited for the study. The first group (IA) consisted of women who admitted to having undergone an illegal abortion, the second group (SA) of women who maintained that the abortion was spontaneous and the third group (AC) of pregnant women attending an antenatal clinic. Patients were enrolled consecutively at Temeke District Hospital (TDH) in Dar es Salaam.

Temeke District covers an area of about 700 km2. According to the 1988 census, its population was 401 776. From projection the 1996 population was estimated to be 590408, including 118082 women of childbearing age. The annual increase in population is approximately 4.8%, partly due to a high fertility rate and partly to immigration from rural areas ( Bureau of Statistics 1992).

Study groups

Data were collected for 5 calendar months (1.2.1997 – 30.6.1997) at the emergency gynaecological ward at TDH. 603 patients with the diagnosis of incomplete abortion were interviewed privately and confidentially by a trained interviewer to determine whether or not their abortion had been secretly induced. Only women who admitted to this were included in this group. As reported elsewhere (Rasch et al. 2000) we found that 60% of women presenting with an incomplete abortion admitted to having induced it. The 241 women (40%) who denied an induced abortion during the confidential interview were classified as SA. During the same period 307 patients at the antenatal clinic of TDH were consecutively recruited as a referent population and interviewed.


Interviews were performed at the gynaecological ward either on admission or just after the patient had recovered. The interviewer was of the same culture as the patients and developed her empathic skills through guided interviews in which one of the authors (VR) participated. Confidentiality was ensured by taking the patient to a separate room and commencing the dialogue by approaching the patient's general life situation. It was stressed that the interviewer should not take any kind of notes or ask whether the abortion had been induced before she had gained the patient's confidence. At the antenatal care clinic, a specially trained nurse midwife who worked at the clinic performed the interviews. Interviews were conducted in Kiswahili and the answers entered in the questionnaire by the interviewer in English.

Socio-economic and reproductive characteristics

The three groups were compared by using a questionnaire to obtain data on age, civil status, education, occupation, dwelling characteristics and religion and on number of previous pregnancies and their outcome, parity and number of currently living children.

Circumstances characterizing the induction

IA women were asked where the abortion had taken place, who performed it, what method was used, how much it cost and who the first confidant was.


Data were recorded using Epi Info vs. 6.03 for Epidemiology and Disease Surveillance from the CDC, Atlanta, USA. Data were entered twice, the two data sets were compared by the Validate program and questionable entries reconciled. We used the Statistical Package for the Social Sciences (SPSS vs. 8.0) for analysis and calculated the odds ratio (OR) by a case-control approach using IA women and SA women (cases) as dependent variables against AC women (controls). The associations between IA or AC and SA or AC and socio-economic and reproductive characteristics are presented as OR with 95% confidence intervals (CI) according to Cornfield ( Brown 1981). To adjust for the possible confounding of the women's educational, occupational and family situation by age and marital situation, logistic regression was performed. Effect modification was evaluated by stratified analyses.


All patients were informed that participation in the study was voluntary and that it would have no impact on their further treatment whether they participated or not. Informed oral consent was obtained, since most women feared lack of anonymity if they consented in writing.


Socio-economic characteristics

The findings are summarized in Table 1. A significantly higher proportion of IA women and a significantly lower proportion of SA women were aged < 25 years than AC women. More than half of the IA patients were < 20 years. There was no association between religious belief and outcome of pregnancy. The vast majority of IA women were single, a result that differed significantly from AC women. No difference was found between SA and AC women. IA women lived with their parents, relatives or a friend more often and with a husband less frequently than AC women. When adjusted for age and marital situation, the proportion of IA women living with parents, relatives, a friend or alone remained significantly higher ( Table 2). 74% of SA and 79% of AC women were cohabiting with their husband. Compared to AC women, a significantly higher proportion of SA women were living alone or with a friend.

Table 1.  Socio-economic characteristics of women with illegal abortion (IA), with spontaneous abortion (SA) and attending antenatal (AC) care Thumbnail image of
Table 2.  Adjusted OR (aOR) for selected variables (IA = women with illegal abortion, SA = women with spontaneous abortion and AC = women attending antenatal care) Thumbnail image of

The civil status was also reflected in the women's occupational situation. The majority of SA and AC women worked as housewives, whereas significantly fewer IA women did so. More than half of the IA women stated that they were students. This result differed significantly from AC women. After adjustment for age and marital situation the association between having an IA and being a student became less pronounced although it remained significant. Almost half of the IA women were educated above primary school level, whereas significantly fewer AC patients were. This association did not change after adjustment for age and marital situation. Regarding educational level and proportion of students, SA women did not differ from AC women.

Reproductive characteristics

Approximately two-thirds of the AC women stated that their pregnancy was planned. This finding was similar to SA women and in contrast to IA patients, only 1% of whom had a planned pregnancy. IA women had significantly less often given birth before and significantly fewer previous spontaneous abortions than SA and AC women. The proportion of women who had an IA before was almost identical among IA and SA women, whereas none of the AC patients admitted to a previous IA. ( Table 3). The women's parity is reflected in the number of their living children. A significantly higher proportion of IA women had no living children whereas a lower proportion had three or more compared to AC women. When adjusted for age and civil status, however, the association changed and women with more than three living children were then more likely to have an IA when compared with women having one or two living children.

Table 3.  Reproductive characteristics of women with illegal abortion (IA), with spontaneous abortion (SA) and attending antenatal (AC) care Thumbnail image of

Circumstances characterizing induced abortion

Most women (60%) were above the 12th gestational week when they had their pregnancy terminated; the mean gestational age was 13.2 weeks ( Table 4). More than half stated as the reason for abortion that they were still at school and feared expulsion. In 79% abortion had been induced at a health clinic or a hospital, the vast majority by either a health clinic worker or a doctor. The most frequently used method was stated to be dilatation and curettage. The average cost was 33 300 Tsh. equivalent to 50 US$ (the monthly salary for a nurse in Tanzania is around 36 000 Tsh.). 12% of the women did not know the price, since their partner had arranged and paid for it. One-third had not told anybody about their situation. 63% of those who had confided in somebody first told the responsible male partner about the pregnancy; others told girlfriends or relatives; only a few had informed their mothers.

Table 4.  Circumstances characterizing the induction Thumbnail image of


In a study on illegal abortion, the main problem is the selection of patients to be investigated. In a previous study on a similar study population we found that using complications (infections and trauma) and the statement ‘unplanned pregnancy’ as criteria of illegally induced abortion was less useful due to under reporting and severe misclassification (Rasch et al. 2000). Instead we chose to classify as IA women only those who, after an in-depth confidential interview, admitted that their abortion was induced. As expected, there was marked similarity of socio-economic and reproductive characteristics between SA and AC women, lending support to the assumption that misclassification of women with IA as SA is not important in this study.

60% of patients admitted with an incomplete abortion stated that it had been induced. An Ethiopian study has shown that 53% of women admitted with incomplete abortion had an IA ( Abdella 1996). In other parts of the world where induced abortion is illegal, unsafe abortions are known to be widespread and women are frequently admitted to hospitals with incomplete abortion after attempting to terminate their pregnancy ( Costa & Vessey 1993; Fonseca et al. 1996 ). The large number of women admitted with IA in our study reflects the fact that the Tanzanian family program does not reach those most in need of contraceptives.

The profiles of IA, SA and AC women differed in various aspects. IA women were significantly younger, more often single, students and nullipara than AC women. SA women did not differ in the same respects from AC women. 88% of IA women were < 24 years, and 55% were < 20 years. Apparently the new family planning guidelines are not being adhered to in practice, as it seems that many adolescent girls still have an unmet contraceptive need. Other studies from sub-Saharan Africa have also reported high percentages of teenagers among women with IA; in Kenya 30% of IA women were teenagers ( Lema et al. 1996 ); in Mozambique, 44% ( Hardy et al. 1997 ), and in Nigeria, 52% ( Anate et al. 1995 ). Adolescent girls are reluctant to attend family planning clinics as they fear the moralizing and judgemental attitudes of the staff and are afraid of being recognized by acquaintances who might inform their guardians. Without access to preventive contraceptive measures many adolescent girls count on access to abortion ( Barker & Rich 1992; Rasch & Silberschmidt 2000). The fact that adolescents do not make use of the existing facilities indicates that special services attuned to adolescents' values and attitudes are needed.

IA women were more often single and living with their parents or friends than SA and AC women. These findings agree with an earlier Tanzanian study ( Justesen et al. 1992 ) and correspond well with profiles of women with IA in Kenya ( Sjostrand et al. 1995 ; Lema et al. 1996 ), in Nigeria ( Ogunniyi & Faleyimu 1991), in Ethiopia ( Madebo & Tsadic 1993) and in Mozambique (Machungo et al. 1997). A significantly higher proportion of SA than AC women lived alone, which might reflect a reluctance among some SA women to admit having had an IA.

Regarding occupation, 56% of IA women stated that they were students, which differed significantly from SA and AC women among whom the same figures were 0% and 1%, respectively. This striking difference reflects that in the event a Tanzanian school girl becomes pregnant she is dismissed from school. This situation leaves pregnant schoolgirls with very few choices. They can either continue their pregnancy and consequently be expelled from school, a situation which may have huge implications for their future lives, or they can decide to have an IA and thereby run the severe health risk associated with such a procedure. Other studies have shown an equally high proportion of students among women with IA. In a similar study population in Nigeria around half of the women with IA were reported to be students ( Konje & Obisesan 1991). In an Ethiopian study, 31% of IA women were students ( Abdella 1996).

Almost half of the IA women were educated above primary school level, whereas this was the case for only 17% of SA and 16% of AC women. Other studies have reported similar findings. In Kenya 53% of patients with IA had attained secondary school education ( Lema et al. 1996 ); in two Ethiopian studies 67% and 72% were educated above primary school level ( Madebo & Tsadic 1993; Abdella 1996).

A significantly higher proportion of IA than AC women were nullipara. When adjusted for age and marital situation, however, the difference became insignificant, indicating that the high proportion of nullipara among IA women is a reflection of the women's age and their marital situation. In a Kenyan study 77% of women with IA were nullipara ( Ankomah et al. 1997 ); in Nigeria, 82% ( Megafu & Ozumba 1991). A significantly higher proportion of SA than AC women had three or more living children. This finding might reflect a reluctance among women of high parity to attend antenatal care ( Mbizvo et al. 1997 ). Hence our finding might serve as an illustration of the problem of achieving a representative study sample of pregnant women when using a hospital-based approach. On the other hand, the difference might also indicate that some IA women with three children or more have been misclassified as SA due to unwillingness among these women to admit having had an IA. Other studies have shown the same unwillingness among elderly women who are suspected of having had an IA ( Justesen et al. 1992 ; Sjostrand et al. 1995 ). The proportion of women who previously had an induced abortion was almost identical among IA and SA. None of the AC women admitted having undergone an IA. This striking difference probably suggests that the interview technique used at the antenatal clinics was insufficient in elucidating this specific issue.

Most (60%) induced abortions had been performed after the 12th week of gestation. This result is rather worrying as the health risk associated with induced abortion increases with gestational week ( Potts et al. 1977 ). The vast majority of induced abortions had been performed in a health unit by a professional, a trend also reported in other studies ( Konje & Obisesan 1991; Okonofua et al. 1992 ; Machungo et al. 1997b ). This finding supports the assumption that in settings with commercialized medicine, poorly equipped private hospitals appear to be the major providers of IA. In 34% of cases the inductions had been performed by doctors, in 44% by other health personnel, and 22% by an unskilled person. Whether or not the ‘doctors’ were actually doctors is difficult to determine, since most lay people refer to any hospital worker, especially those who wear white uniforms, as doctors. Our results agree with a Tanzanian study which reported that 22% of the abortionists were ‘doctors’, 65%‘other health clinic workers’ and 13%‘quacks’ ( Mpangile et al. 1993 ). The abortionist profession is reflected in the method most frequently reported, dilatation and curettage.

The average cost of an induced abortion was 33300 Tsh., equivalent to the monthly salary of a nurse at the time of the study. Other studies have reported equally high costs ( Mpangile et al. 1993 ; Machungo et al. 1997b ). It can be argued that the women in our study represent a privileged group who were able to pay for an IA. Not all Tanzanian women are able to raise this amount of money and consequently are bound to continue their pregnancy. Our findings underline that all women, regardless of their socio-economic situation, must have access to safe legal abortion services in the event of contraceptive failure.

In conclusion, the empathic approach used in classifying women with incomplete abortion as either IA or SA enabled us to distinguish between these two groups. The characteristics of IA women differed from SA and AC women; they tended to be single, childless, less than 20 years old, still at school and better educated than AC women. Our results indicate that there is a serious discrepancy between the existing, revised and progressive family planning guidelines that do address adolescents' needs – and the way in which they are dealt with in practice. There is a need for developing youth-friendly clinics attuned to adolescents' values and attitudes if young people's sexual and reproductive needs are to be addressed. Furthermore, as IA women rarely leave hospital with the means to avoid another unwanted pregnancy, they need family planning services to break the vicious cycle of unprotected intercourse, unwanted pregnancies and subsequent illegal abortion.


We are grateful to the women who shared their experience and time with us and to the research assistants who gave several months of their time during the period of data collection. Grants from Odense University, Denmark, are gratefully acknowledged.