Legal Abortion Worldwide: Incidence and Recent Trends By Gilda Sedgh, CONTEXT: Information on abortion levels and trends can inform research and policies affecting maternal and Stanley K. reproductive health, but the incidence of legal abortion has not been assessed in nearly a decade. Henshaw, Susheela Singh, METHODS: Statistics on legal abortions in 2003 were compiled for 60 countries in which the procedure is broadly legal, Akinrinola Bankole and trends were assessed where possible. Data sources included published and unpublished reports from official and Joanna national reporting systems, questionnaires sent to government agencies and nationally representative population Drescher surveys. The completeness of country estimates was assessed by officials involved in data collection and by in-country and regional experts. Gilda Sedgh is senior research associate, RESULTS: Inrecentyears,morecountriesexperiencedadeclineinlegalabortionratesthananincrease,amongthosefor Stanley K. Henshaw is which statistics are complete and trend data are available. The most dramatic declines were in Eastern Europe and senior fellow, Susheela Central Asia, where rates remained among the highest in the world. The highest estimated levels were in Armenia, Singh is vice president Azerbaijan and Georgia, where surveysindicate that women will have close to three abortions each, on average, in their of research, Akinrinola lifetimes. The U.S. abortion rate dropped by 8% between 1996 and 2003, but remained higher than rates in many Bankole is director of Northern and Western European countries. Rates increased in the Netherlands and New Zealand. The official abortion international research and at the time of rate declined by 21% over sevenyears in China, which accounted fora third of the world’s legal abortions in 1996. Trends writing Joanna in the abortion rate differed across age-groups in some countries. Drescher was research associate—all at the CONCLUSIONS: The abortion rate varies widely across the countries in which legal abortion is generally available and Guttmacher Institute, has declined in many countries since the mid-1990s. New York. Perspectives on Sexual and Reproductive Health, 2007, 39(4):216–225, doi: 10.1363/3921607 In recent decades, abortion has received considerable The last assessment of abortion levels in countries where attention, and its legality and availability have often legal abortion is generally available* was conducted nearly generated controversy. Even in countries where abortion adecadeago.1 Accordingtothatstudy,uptothemid-1990s, is not a contentious issue, contraception is usually a less legal abortion rates had been falling in many parts of the expensive and less taxing means of avoiding unintended world, either as contraceptive prevalence was increasing or births. Accurate information on abortion levels and trends as contraceptives were being used more effectively. The can help donors, policymakers and program planners investigators speculated that although the legalization of assess the extent to which women experience unintended abortion may initially result in an increase in the number of pregnancies, and can facilitate the development of policies reported abortions in countries where desired fertility is and programs to respond to unmet need for effective low,abortionrateswilleventuallydeclineasaccesstofamily contraceptive services. Accurate measures of abortion planning education and contraceptive services increases.2 can also inform the public discourse by providing impar- Updated information on abortion incidence is needed tial, empirical evidence of abortion prevalence. to assess recent trends and current patterns in induced abortion across countries and regions and in different age- groups, and to document abortion incidence in countries *Countries are included in this category if abortion is allowed for social or in which legal abortion has recently become available. economic reasons or without specification as to reason. A few countries In this article, we present statistics on the level of with more restrictive formal laws are included because sources indicate induced abortion in 60 countries and territories in which that legal abortion is broadly available: Australia, Hong Kong, Israel, New † Zealand and Spain, where abortion is permitted to protect the woman’s legal abortion was generally available in 2003. We assess mental health, and South Korea, where it is permitted to protect her the completeness of the available abortion data, and physical health. wherever statistics of comparable quality are available †Dependent territories with separate abortion statistics or legislation are for 1996 (the most recent year for which similar statistics 1 treated as countries. have been published ), we report trends in abortion Ed. note: This article was first published in International Family Planning incidence. In addition, for countries with sufficiently Perspectives, 2007, 33(3):106–116. complete data on legal abortion, we provide recent 216 Perspectives on Sexual and Reproductive Health age-specific abortion rates, which can help identify ment reporting systems, were always assumed to be groups of women who are experiencing the greatest dif- incomplete because some women do not report their ficulties in preventing unintended pregnancy. abortions, particularly in face-to-face interviews,5,6 and surveys are subject to sampling error and random variation. METHODS For all but four of the 33 countries considered to have DataCollection complete counts, the data were government statistics. For We employed methods similar to those used in the last Puerto Rico7 and the United States,8 data collected through review of the incidence of abortion worldwide.1 For each surveys of all providers were used; the U.S. estimate for 2000 country or territory with a population of at least one was projected to 2003.9 For Australia, we used a published million in which legal abortion was generally available estimate based on insurance claims and hospital statistics.10 in 2003, we sought data on the total and age-specific For Switzerland, we used data compiled by a private or- numbers of abortions. Several sources of data were used: ganization using reports from local government offices.11 published abortion data from national statistical offices or Government statistics were available for 24 of the 27 relevant government agencies; for countries that lacked countries with estimates that are incomplete or of such data, responses to a standardized questionnaire or unknown completeness. For South Korea, Turkey and formal inquiry that we sent to appropriate government Turkmenistan, only survey-based estimates were available. agencies, sometimes with the help of local contacts; for The survey in South Korea was administered to married a few countries, abortion data from the Council of Europe women only, and the survey in Turkey was administered or the European Region of the World Health Organiza- to ever-married women only. In 11 countries where both tion;3,4 and for countries whose official statistics were official statistics and nationally representative survey data deemed incomplete or unavailable, abortion estimates were available, both sets of findings are shown. In all but from nationally representative population surveys. one of these countries, survey estimates were based on Most of the statistics presented here are based on women’s reports of their induced abortions obtained in the official reports of the numbers of induced abortions one-, three- or five-year period prior to the survey. The performed according to the laws of the countries. How- abortion estimate for India is derived from a facility-based ever, the available abortion statistics are not always survey. In eight countries with incomplete reporting or complete. For example, providers do not always report reporting of uncertain completeness (China, Georgia, all abortions they perform, even if legally required to do Hong Kong, India, Japan, Romania, Turkey and Vietnam), so. In some countries, only abortions performed at public abortion estimates from the same data source are available facilities are reported, whereas large proportions of for 1996, and these are presented as well. abortions are performed at private facilities. In addition, We also obtained estimates of the population of women medication abortions or early surgical procedures may be aged 15–44 in five-year age-groups and numbers of live underreported in some countries. births to calculate abortion rates and ratios for 2003. We asked local experts to assess the extent to which the When available, we used data from country statistical reported statistics represented all legal abortions that had offices. For countries where population estimates were been performed. These experts included researchers, not available, we used estimates published by the Council officials from government agencies involved in abortion of Europe or the United Nations. We interpolated where data collection, and program administrators and pro- necessary to obtain midyear population estimates. viders who were familiar with abortion reporting prac- tices. We also ensured that published statistics did not Analysis include spontaneous abortions, and confirmed that they We calculated age-specific abortion rates (the number of included menstrual regulations (or ‘‘mini-abortions’’) abortions per 1,000 women in each five-year age-group), where these are recorded separately. Furthermore, we the general abortion rate (the number of abortions