☰ Menu

Postgraduate Training Course in Reproductive Health 2004

Assessing maternal mortality due to induced abortion
A systematic review of the literature

Laura Gil, MD
Instituto Materno Infantil
Bogota, Colombia

See also presentation


In order to reduce the mortality rate due to induced abortion, health care providers and policy makers must be aware of the real incidence and burden of induced abortion world wide.  Estimates are difficult to make, due to inaccuracy of the reporting systems, the policies and the mechanisms for gathering information in the different countries. An effort to gather the available information based on reliable data is urgently needed. Gathering information concerning the incidence, morbidity and mortality on reproductive health issues is a sensitive topic because they reflect on the level of development of a country thus having not only public health implications, but also economic and political significance. Compiling epidemiological data may prove to be extremely difficult for some countries where no adequate flow of information exists. Data tend to be inaccurate due to underestimation of the causes of maternal death, even in countries with official records. Countries without registry systems often show high mortality rates, requiring reproductive health research and interventions (1).
By the year 2000, only 78 countries, representing 35% of the world’s population, adequately reported the cause of death in their civil registry (2). Although, ideally the information on maternal mortality should be gathered from reliable registration systems, other models have been developed and are in use:

  • The Reproductive Age Mortality Studies (RAMOS) is thought to be the most reliable but a time consuming and expensive method. It investigates all deaths of women of reproductive age and assesses whether it was related to pregnancy or not. A variety of sources can be used, like interviews with household members and health care providers as well as the review of clinical records.
  • Direct household surveys indirectly estimate the maternal mortality rate but require a large sample size and is therefore often imprecise. The indirect and direct sisterhood methods in which women are interviewed about the survival of all their adult sisters alleviates somewhat the problem with large sample size requirements but still pose the pitfall of wide confidence intervals. They also give retrospective information from around 12 years before the survey so they are not useful for inferences on short term changes. They also require stable populations in terms of migration, wars or any other factor of displacement.
  • Verbal autopsy may be helpful in places where no medical certificat of the cause of death is available but its reliability and validity have not been established and it may fail to identify deaths occurring early in pregnancy, late after delivery and those due to indirect causes like malaria or HIV infection.

The World Health Organization (WHO) has developed estimates on maternal mortality for 1990, 1995 and 2000 based on data including national statistics, direct reporting and population based studies and thus from highly heterogeneous data, having to adjust the figures for statistical analysis. Such data might not be comparable as they have been conducted at different times and the different statistic methods used. (2,3)
Furthermore, research on the epidemiology of induced abortion involves greater difficulties and poses a special challenge to the investigator. The gathering of data may be even more difficult than that of maternal mortality and morbidity globally due to the legal and moral implications prevailing in many settings. Legal abortion means the procedure is authorized, and as such, it can be recorded at the health care facility but this is not the case for illegal abortions. Twenty million out of approximately 46 million abortions annually are illegal, with the greater majority taking place in developing countries (1).
Many methods to estimate the incidence of induced abortion have been used, such as: the "illegal abortion provider survey," the "complications statistics" approach, the "mortality statistics" approach, self-reporting techniques, prospective studies, the "residual" method, anonymous third party reports, and experts' estimates. They often involve surveys for providers or users and individuals may fear legal or moral judgment leading to inaccuracies of the data and underestimation.(10). Other reasons for bias should be taken into account, for example, in the case of the "complications statistics" approach, that uses the information from the number of women treated for complications resulting from abortion. Such information may be misleading, since complications may be the result of various reasons and its rate may vary widely depending on the type of setting in which the abortion is performed, ranging from unskilled, non-sanitary interventions to trained doctors’ practices with adequate technology, and a very low complication rate (4).
All these efforts have resulted in the publication of very useful estimates on induced abortion incidence, based mainly on large surveys and some times relying on experts’ opinions.

How should abortion incidence be approached?

This information may be approached as rates or ratios depending on the total population taken into account. Induced abortion rates correlate the number of abortions with the populations, most commonly, the number of abortions per 1000 women aged 15-44 whereas the ratio refers to the number of abortions per 100 known pregnancies. The last is more likely to depend on the fertility rate of the studied population and reflects therefore the desire of women regarding the wanted family size. The two approaches provide different but complementary types of information and assessing the differences between them is useful to understand the time trends of a given society and regional differences (5).
A third interesting approach is the measure of the total first abortion rate (TFAR) which gives an idea on the proportion of women that undergoes induced abortion during their reproductive life span. This measurement is not easy to perform in every setting because it requires a registry of all abortions performed, specifying if there was a previous history of abortion, for at least 30 years to cover at least one single reproductive period for a given generation. This technique has the advantage of showing generational changes in the use of abortion. (6)
Some interesting approaches include those compiled by the Alan Guttmacher institute periodically in the United States. This approach allows for an observation on the national trends given that no complete official registration is available from 1996 on induced abortion. A survey on 2442 potential providers was performed resulting in the report of a total of 1.31 million abortions performed in the 2000 for an abortion rate of 21.3 abortions per 1000 women in reproductive age (reference).
In Nigeria, a developing country with highly restrictive abortion laws, and a much more difficult scenario, a large survey was conducted on a nationally representative sample of 672 health facilities to estimate the total annual number of abortions and to describe the provision of abortion- related services in 1998. The author found a rate of 18 to 25 per 1000 women aged 15-44 and a ratio of 9 to 12 per 100 known pregnancies (7).
In South Australia, were abortion has been legal for the past decades and a reliable registration system is available, the TFAR has been calculated to be 29%. Meaning that about 29% of South Australian women born around 1955 and exposed to legal abortion throughout their reproductive lifetime experienced an induced abortion. This proportion was found to be higher for later cohorts of women born in 1960-1980 (e.g., at least 31% for those born in 1960) (6).
To our knowledge, it has never been tried to synthesize these data by performing a systematic review. When a critical appraisal of the literature is performed, reliable comparable data may emerge and appropriate statistical analysis may than lead to useful conclusions.
A great effort is currently being undertaken to achieve this goal through the systematic review and critical appraisal of the available literature published between 1996 and 2003. This work will render information on more reliable information ever produced on maternal mortality, with a separately evaluating every cause of mortality and to give an idea of the burden of each of them.


A systematic review of the literature providing data on maternal mortality and morbidity was conducted. This review had no language restrictions and included all the existing electronic data bases such as Medline, Popline, CAB, Sociofile, Cinhal, Econlit, Ebase, BIOSISS PAIS International, the grey literature database (SIGLE), the Cochrane Database of Systematic Reviews, the Database of Abstract of Reviews of Effectiveness and the Cochrane Controlled Trials Register. Also the regional WHO on-line databases and other electronic sources like the existing web pages from ministries of health of every country and any other retrieval by “maternal mortality”, hand a reference searching and personal contacts. All full texts of the relevant studies were critically appraised regarding the study design, sampling, characteristics of the population studied, setting and type of data reporting as follows:

  • Data quality assessment: whether the type of data reported was an estimate or an actual count, the type of maternal death definition if it was included (up to 42 day postpartum, up to 1 year postpartum or 42 day postpartum regardless of the cause of death) and whether the deaths were confirmed to be maternal and how (confidential enquiry, verbal autopsy, others).
  • General information regarding the country, the region and the study period.
  • Characteristics of the study: type of study, how the sampling was carried out, the data source, and the lowest unit (cluster vs. individual), type and size and demographic characteristics of the population studied; information on the extent and quality of follow up and type of setting of the study (national, province, city, medical facility, etc) and place of the abortion (home, hospital, etc) also description of the intervention if applicable and the forms of reporting the data (crude, adjusted for confounding variables, etc)
  • Data quality definitions: The definitions for induced abortion when available, the gestational age at which the abortions were induced and the characteristics of the abortion (whether legal or not, safe or unsafe) were also recorded in proper questionnaires to be read and classified by an electronic scanner.

Once all the above mentioned data are extracted by the reviewers, consolidate data will be produced.


After retrieval and critical appraisal of all articles concerning maternal mortality and morbidity, a total of 137 articles related with induced abortion were found of which, 58% were found to be of medium quality according to the study design and the methodology used. 37% of the articles were from developing countries. The final analysis and interpretation of the data will be the next step.


Induced abortion defined as the elective termination of pregnancy during the first or second trimester by surgical or medical means is one of the most frequent gynaecological procedures, and with few complications when performed in safe conditions (8). Even though induced abortion carries a lower mortality risk than that of delivery itself it is not so in every setting. The higher incidence of unsafe induced abortions observed in many developing countries reflects an evident social and economical inequity in such regions. The availability of reliable data on the incidence of induced abortion and its burden in terms of mortality and morbidity are of great value when trying to develop policies on such topic and in the design of services to be offered in each setting identifying the most vulnerable situations and subjects based on individual findings for specific regions. An estimated of 38 to 50% of all pregnancies that happen every year in the world are unintended despite of a general increase of the use of modern contraceptive methods. The burden of all this unwanted pregnancies is reflected on a high number of induced abortions performed every year, calculated to be 45.5 million in 1995; 44% of them, performed under illegal conditions. Twenty-six percent of all pregnancies and around 50% of this unwanted pregnancies end up in induced abortion (8).
Many questions have been raised when analysing epidemiological data on induced abortion. The majority of them remain to be answered by future research.

What Is The Real Incidence Of Abortion Worldwide And How Accurate Are The Data?

The incidence of abortion world wide was calculated to be 35 per 1000 women aged 15-44 and 26 per 100 known pregnancies for 1995. The recollection of this data was based on countries’ registration when available and in many cases, on estimates. The error margin of these estimates can be as high as 20%. For high quality surveys the estimated variation can be as low as 4% (8) whereas estimates based on population surveys may underestimate the rate of abortion as much as 50% (9). This estimates range from the low rate of 11 per 1,000 in Western Europe through the world average of 33–37 per 1,000 in Africa, Asia, and Latin America, to the high rates of 78–83 in Cuba and Vietnam and 90 per 1,000 in Eastern Europe. The highest incidence ever reported was 240 per 1000 women, found in Romania in 1984. The rate in developed countries is 35 per 1000 women of reproductive age very similar to that in developing countries, of 34. In contrast the ratio on the former is 42 per 100 known pregnancies compared with 23 in the latter. This difference between the rate and the ratio is due to a lower fertility rate in the developed countries (10)

Which Factors Determine The Differences Between Countries And Its Time Trends?

Induced abortion incidence varies widely between different regions depending on many factors. The investigation of these factors may lead to a better understanding of this phenomenon. The assessment of the determinants of variations of induced abortion must take into account several variables like couples' reproductive preferences (which clearly determines the fertility rate of a region), the prevalence and effectiveness of contraceptive practice to implement these preferences, and the probability of undergoing an abortion to avoid an unintended birth when contraception fails or is not used (11).

Availability of contraception

The relationship between levels of contraceptive use and the incidence of induced abortion is very controversial. Common sense would indicate that abortion rate should decrease as contraceptive use increases but this is not always the case and this does not imply that they are not directly associated. Some observations suggest an increase in induced abortion despite increased availability of contraception. Different patterns have been recognized when trying to correlate this two variables in different countries and other co-factors have been identified such as the quality of the contraceptive methods and the total fertility rate (TFR).

The Effectiveness of the Contraceptive Methods Used

Data from countries where the information on abortion rates and contraceptive use is well recorded, show that as the proportion of users of highly effective methods increases, the rate of induced abortion decreases. When the prevalence of modern contraceptives is around 70%, the abortion rate is typically in the range of 10-30 abortions per 1000 women in the and when this prevalence falls to 40 to 60% abortion rates increase to 30-50 % (12).

  • China: in the context of the 'One-child policy', as soon as women start using highly effective methods, the abortion rate dropped to almost 0 after 5 years
  • In Kazakhstan, Uzbekistan and the Kyrgyz Republic where abortion was widely available and was the preferred method for family planning due to of limited access to contraception, abortion rates were as high as 180 per 1000 women. In the last decade, with contraceptive use became more frequent, increasing up to 50%, the abortion rate has decreased by the same amount.

Total Fertility Rate

An increase in the use of contraception is related to a decrease in abortion rate in populations with stable or slow decrease of fertility rates as has been demonstrated by the analysis of the national registry of Bulgaria, Switzerland (1980-1995) and Tunisia (1975-1995) (11). This pattern has not been observed in all countries. In Cuba, for example, an increase in both abortion rate and contraceptive use has been observed. This is believed to be due to a rapid fall in the overall fertility rate which could not be met by the less rapidly increasing contraceptive use. This trend was also seen in Denmark, Netherlands, the United States, Singapore and South Korea, during the 1970’s, followed by a period of stabilization of the fertility rate with a further continuous decline on abortion rates.

Legal status

Although legal restrictions to abortion exist in many countries, they do not result in a lower abortion rate. Currently, six in 10 women--55% of those in the developing world and 86% in the developed world--live in countries where abortion is permitted on broad grounds. By contrast, 25% of women live in countries, overwhelmingly in the developing world, where abortion is prohibited altogether or allowed only to save a woman's life. A country's abortion rate is not closely correlated with whether abortion is legal or not. For example, abortion levels are high in Latin America, where abortion is highly restricted. At the same time, abortion rates are quite low throughout Western Europe, where the procedure is legal and widely available. Also, Eastern and Western Europe have the world's highest and lowest abortion rates, respectively, yet abortion is generally legal throughout the continent. Although the average incidence tends to be alike between countries with opposite legislations, complications and mortality rate differ due to the high incidence of unsafe abortion in illegal settings (1, 10). Changes in legislation can have dramatic effects on legal abortion rates. If these changes are not accompanied by corresponding changes in levels of contraceptive use or fertility, it is more likely that legal abortions will replace illegal abortions or vice versa rather than that the overall abortion level will change.
The well known case of Romania illustrates it very well: restrictions on abortion and contraception imposed during de 1960 has little effect on birth rates and demonstrating that abortions were still available despite legislation but under unsafe conditions. with. 86% of maternal deaths due to induced abortion. The number of abortion-related deaths dropped precipitously following the lifting of the restrictions on abortion in December 1989. (1)

Who is at a Greater Risk for Induced Abortion?

Many authors have tried to identify the high risk groups of women that are more prone to have abortions and although the characteristics vary from country to country, depending on the culture, the legislation, religion and TFR, some groups have been identified. In many developed countries, a trend for increasing abortion rates in teenagers has been observed. Such is the case in England in which the only age group to show a continuous increase since 1969 has been girls aged 11 to 14 years (13). In Japan the abortion ratio for women younger than 20 increased during the period of 1975 to 1995 from 18 % between 1976 and 1980 to 30 % between 1991 and 1995 (14). Another interesting observation is the fact that younger users of abortion services tend to come later in pregnancy, which carries a higher risk of complications (15). The risk of exposure to an unintended pregnancy and therefore an induced abortion deals also with decreasing age at first intercourse. A peak before menopause and between women with high parity has also been observed (16).

How to Reduce the Mortality Rate Due to Induced Abortion?

Ninety-five percent and 99% of all abortions performed in Africa and Latin America are illegal and thus often unsafe. In 1998, 30 unsafe abortions for each 1000 women in Latin America and the Caribbean were performed while the world average is 13 per each 1000 and 1 out of every 8 deaths related with pregnancy in this region are due to this practice (3). Short term complications of induced abortion like bleeding, infection and incomplete evacuation are rare and of no public health concern in countries where it is legal like in Denmark, where they are seen in less than 5% of the cases .In contrast, unsafe abortion, which represents as much as one third of all the abortions performed every year is a major cause of chronic and often irreversible health problems and even death for almost 50.000 to 100.000 women. Around 800.000 women every year are likely to obtain hospital treatment for the complications of induced abortion (17).
Reducing the rate of unsafe abortions is one of the most needed interventions helping to reduce the maternal mortality rate and also to alleviate to some extent the inequity among women’s situation around the world. Obviously, an increase on the availability of contraception as well as a shift from the use of traditional to modern methods will also have a great role in achieving this goal.


  1. Cynthia Dailard. Abortion in Context: United States and Worldwide. The Alan Guttmatcher Institute web page http://www.agi-usa.org/pubs/ib_0599.html
  2. http://www.who.int/reproductive-health/publications/maternal_mortality_2000/challenge.html
  3. Maternal Mortality in 2000: Estimates Developed by WHO, UNICEF and UNFPA http://www.who.int/reproductive-health/publications/maternal_mortality_2000/maternal_mortality_2000.pdf.
  4. Rossier C. Estimating induced abortion rates: a review. Stud Fam Plann. 2003 Jun;34(2):87-102. [PubMed]
  5. Van Look PFA, Von Hertzen H. Induced Abortion Special Programme of Research, Development and Research Training in Human Reproduction, World Health Organization, 1211Geneva27,Switzerland https://www.gfmer.ch/Books/Reproductive_health/Induced_abortion.html
  6. Chan A, Keane RJ. Prevalence of induced abortion in a reproductive lifetime. Am J Epidemiol. 2004 Mar 1;159(5):475-80. [PubMed]
  7. Henshaw SK, Singh S, Oye-Adeniran BA, Adewole IF, Iwere N, Cuca YP. The Incidence of Induced Abortion in Nigeria. Int Fam Plann Persp. 1998 Dec;24 No 4 : 156-164.
  8. Finer LB, Henshaw SK. Abortion incidence and services in the United States in 2000. Perspect Sex Reprod Health. 2003 Jan-Feb;35(1):6-15. [PubMed]
  9. Houzard S, Bajos N, Warszwawski J, de Guibert-Lantoine C, Kaminski M, Leridon H, Lelong N, Ducot B, Hassoun D, Ferrand M. Analysis of the underestimation of induced abortions in a survey of the general population in France. Eur J Contracept Reprod Health Care. 2000 Mar;5(1):52-60. [PubMed]
  10. Henshaw SK, Singh S, Haas T. The incidence of abortion worldwide.Int Fam Plann Persp. 1999 Jan;25(Suppl):S30-8. [PubMed]
  11. Bongaarts J, Westoff CF. The potential role of contraception in reducing abortion. Stud Fam Plann. 2000 Sep;31(3):193-202. [PubMed]
  12. Marston C, Cleland J. Relationships between contraception and abortion: a review of the evidence. Int Fam Plan Perspect. 2003 Mar;29(1):6-13. [PubMed]
  13. Ghebrehewet S, Ashton J. A review of induced abortion rates in England and Wales, 1969-1994.
  14. Goto A, Fujiyama-Koriyama C, Fukao A, Reich MR. Abortion trends in Japan, 1975-95. Stud Fam Plann. 2000 Dec;31(4):301-8. [PubMed]
  15. Induced abortion. Facts in brief . Alan Guttmatcher instiute web page http://www.guttmacher.org/pubs/fb_induced_abortion.html
  16. Tullberg BS, Lummaa V V. Induced abortion ratio in modern Sweden falls with age, but rises again before menopause. Evol Hum Behav. 2001 Jan;22(1):1-10. [PubMed]
  17. Prevention of unsafe abortion. World health organ tech rep ser. 1997; 871: i-vii. http://www.who.int/reproductiveHealth/publications/MSM_97_16/MSM_97_16_table_of_contents_en.html