Articles

Unintended pregnancy and abortion by income, region, and

the legal status of abortion: estimates from a comprehensive
model for 1990–2019
Jonathan Bearak, Anna Popinchalk, Bela Ganatra, Ann-Beth Moller, Özge Tunçalp, Cynthia Beavin, Lorraine Kwok, Leontine Alkema

Summary
Background Unintended pregnancy and abortion estimates document trends in sexual and reproductive health and Lancet Glob Health 2020;
autonomy. These estimates inform and motivate investment in global health programmes and policies. Variability in 8: e1152–61

the availability and reliability of data poses challenges for measuring and monitoring trends in unintended pregnancy Published Online
July 22, 2020
and abortion. We developed a new statistical model that jointly estimated unintended pregnancy and abortion that
https://doi.org/10.1016/
aimed to better inform efforts towards global equity in sexual and reproductive health and rights. S2214-109X(20)30315-6
See Comment page e1106
Methods We developed a model that simultaneously estimated incidence of unintended pregnancy and abortion Guttmacher Institute,
within a Bayesian framework. Data on pregnancy intentions and abortion were compiled from country-based surveys, New York, NY, USA
official statistics, and published studies found through a literature search, and we obtained data on livebirths from the (J Bearak PhD, A Popinchalk MPH,
World Population Prospects. We analysed results by World Bank income groups, Sustainable Development Goal C Beavin BA, L Kwok BA); UNDP/
UNFPA/UNICEF/WHO/World
regional groupings, and the legal status of abortion. Bank Special Programme of
Research, Development and
Findings In 2015–19, there were 121·0 million unintended pregnancies annually (80% uncertainty interval [UI] Research Training in Human
112·8–131·5), corresponding to a global rate of 64 unintended pregnancies (UI 60–70) per 1000 women aged Reproduction, Department of
Sexual and Reproductive
15–49 years. 61% (58–63) of unintended pregnancies ended in abortion (totalling 73·3 million abortions Health and Research, WHO,
annually [66·7–82·0]), corresponding to a global abortion rate of 39 abortions (36–44) per 1000 women aged Geneva, Switzerland
15–49 years. Using World Bank income groups, we found an inverse relationship between unintended pregnancy (B Ganatra MD, A-B Moller MPH,
and income, whereas abortion rates varied non-monotonically across groups. In countries where abortion was Ö Tunçalp MD); and Department
of Biostatistics and
restricted, the proportion of unintended pregnancies ending in abortion had increased compared with the Epidemiology, University of
proportion for 1990–94, and the unintended pregnancy rates were higher than in countries where abortion was Massachusetts, Amherst, MA,
broadly legal. USA (L Alkema PhD)
Correspondence to:
Interpretation Between 1990–94 and 2015–19, the global unintended pregnancy rate has declined, whereas the Dr Jonathan Bearak, Guttmacher
Institute, New York, NY 10038,
proportion of unintended pregnancies ending in abortion has increased. As a result, the global average abortion rate USA
in 2015–19 was roughly equal to the estimates for 1990–94. Our findings suggest that people in high-income countries [email protected]
have better access to sexual and reproductive health care than those in low-income countries. Our findings indicate
that individuals seek abortion even in settings where it is restricted. These findings emphasise the importance of
ensuring access to the full spectrum of sexual and reproductive health services, including contraception and abortion
care, and for additional investment towards equity in health-care services.

Funding UK Aid from the UK Government, Dutch Ministry of Foreign Affairs, UNDP/UNFPA/UNICEF/WHO/
World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP),
and The Bill & Melinda Gates Foundation.

Copyright © 2020 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND
4.0 license

Introduction also because individuals with an intended pregnancy

With long-term global decreases in desired family size might be unable to continue a pregnancy, access to safe
and changes in the age at which people want to start and legal abortion as well as contraception increases
families,1–3 individuals increasingly spend more of their reproductive autonomy. Limits on access to services, in
lives trying to avoid unintended pregnancies. Evidence contrast, contribute to unintended pregnancy and, when
suggests that rates of unintended pregnancy have access is limited, unsafe abortion.6
declined world­wide,4,5 with increasing access to and use Despite global health commitments, such as the
of contraception likely contributing to these trends. Sustainable Development Goals (SDGs)7 and the Global
However, existing contra­ceptive methods are imperfect Strategy for Women’s, Children’s and Adolescent’s
and some people who do not want to get pregnant might Health (2016–30),8 the Guttmacher-Lancet Commission9
not want or be able to use these methods. As such, and has documented continued gaps in sexual and

www.thelancet.com/lancetgh Vol 8 September 2020 e1152

 Articles

Research in context
Evidence before this study used information on married women’s contraceptive needs
The scarcity and unreliability of data necessitated studies that and use of contraceptives to estimate abortion incidence only.
use a model-based approach to estimate rates of abortion and Our analysis also incorporates information on these measures
unintended pregnancy. Sedgh and colleagues used a model- in unmarried women. Additionally, we extended their
based approach to produce regional abortion estimates from framework to simultaneously estimate unintended pregnancy,
1990 to 2014. Bearak and colleagues then used these estimates unintended births, and abortion. We also published a study
and other data to estimate unintended pregnancy. protocol, which included new data classification processes.
Sedgh and colleagues found that abortion rates between Finally, our study includes data up until 2018 and estimates up
1990–94 and 2010–14 declined in Europe and northern until 2019.
America, and remained similar elsewhere. As of 2010–14, Implications of all the available evidence
they also found that abortion rates did not substantially differ In contrast to declines in the global unintended pregnancy
between countries where abortion was legally available rate since 1990–94, the global abortion rate declined through
compared with those where it was restricted. Subsequently, 2000–04 before returning to levels last seen in the early
Bearak and colleagues found that unintended pregnancy rates 1990s, as the proportion of unintended pregnancies ending in
declined in all world regions, but that the evidence for a decline abortion has increased. We also found substantial differences
was comparatively weak in Latin America and northern between low-income and high-income countries in the
America. They also found that women with unintended unintended pregnancy rate. This finding highlights global
pregnancies were less likely to obtain an abortion in countries economic inequality in women’s and couples’ access to sexual
where access to abortion was restricted. and reproductive health services. For countries in our study
Added value of this study that had restricted access to abortion, rates of unintended
We developed a new model that simultaneously pregnancies were higher than in countries where abortion was
estimated abortion and unintended pregnancy. Additionally, broadly legal. The proportion of unintended pregnancies
we built a new global database that incorporated data from ending in abortion had increased since 1990–94 in countries
166 countries, compared with 92 countries and 105 countries were abortion was legally restricted. As such, individuals with
from previous studies. Countries were invited to provide unintended pregnancies relied on abortion services even in
feedback on the data sources, methods and preliminary settings where abortion was restricted, potentially facing legal
estimates as part of a WHO country consultation process. and physical risks for doing so. Our findings emphasise the
Our methods and comprehensive data input resulted in need for continued investment in access to quality and
different estimates, and these estimates supersede those from comprehensive sexual and reproductive health care and,
previously published studies. Our analysis improves on equally importantly, a commitment to upholding sexual and
previous research in several ways. Sedgh and colleagues reproductive rights.

reproductive health and rights, including those related groups of countries where abortion is broadly legal to
to access to contraception and safe quality abortion care. countries where it is not. This evidence can inform efforts
In this context, timely evidence for the incidence of towards global equity in sexual and reproductive health
unintended pregnancy and abortion can motivate and rights.
investment and greater commitment to increase access
to services and inform policies and programmes. Methods
To strengthen this much needed evidence base, our This section presents a broad overview of methods
study makes several contributions. We developed a reflecting space constraints, with a complete description
new statistical model that jointly estimated unintended of the methods elsewhere.16
pregnancy and abortion. These outcomes contextualised We followed the Guidelines for Accurate and
one another, providing a more comprehensive under­ Transparent Health Estimates Reporting (GATHER)17
standing of unintended pregnancy and abortion. A statement in developing the database, analysis, and
study protocol was published to enhance transparency,10 presentation of the study.
and substantially more data were obtained (thereby
covering more countries) than previous efforts.4,10–14 With Data for unintended pregnancy and abortion
this new approach, we developed estimates from Data for pregnancy intentions were compiled from
1990–2019. country-based surveys and from studies found through
This Article summarises how estimates of unintended a literature search. These surveys and studies are
pregnancy, abortion, and the percent of unintended described in detail in the technical paper.16 In surveys
pregnancies that end in abortion compare across World done peri­ odically, a pregnancy was considered unin­
Bank income groups15 and SDG regions. We also compare tended if it occurred sooner than desired or if it was not

e1153 www.thelancet.com/lancetgh Vol 8 September 2020

 Articles

wanted at all. We obtained data on abortion from unintended pregnancy was defined in our model as a
published studies and official statistics. Official statistics function of the number of women (disag­gregated by
might be incomplete because of various issues, marital status) with an unmet need for contraception,
including the legality of abortion, differing reporting the number of women who used a contraceptive
requirements across countries, and under-reporting.18 method who experienced a method or user failure,
Official statistics were determined to be complete and the risk of pregnancy in each of these population
through a data classifi­ cation process outlined in the groups (figure 1). Similarly, the incidence of intended
study protocol and, if incomplete, were treated as pregnancies was defined as a function of the number of
minima (minima inform the model that the true women (disaggregated by marital status) with no need
abortion rate is likely to be no less than the observed for contraception and their risk of pregnancy. Abortion
rate).10,16 incidence was a function of group sizes, pregnancy
In total, 2415 datapoints were obtained for pregnancy rates, and propensities to have an abortion. Data for the
intentions and abortions from 166 countries for the time sizes of these groups, and contraceptive method mix,
period of 1990–2019. 516 observations were compiled on were developed by the United Nations Population
unintended fertility for 139 countries. For abortion data, Division.19–23
we obtained 1899 observations from 105 countries, with Pregnancy outcomes were abortions, livebirths,
1019 observations treated as minima. and fetal losses (ie, miscarriages and stillbirths;
As some countries had data for one outcome but appendix p 1). We estimated fetal losses using an See Online for appendix
not the other, we had data for abortions or pregnancy approach derived from life tables of pregnancy loss by
intentions for 166 countries and territories. Of these gestational age in which there was, on average, one
166 countries, we had point data for 150 countries; that is, fetal loss for every ten abortions and one fetal loss
for the other 16 countries, data were all minima, maxima for every five livebirths.18,19,24
(maxima inform the model that the true abortion rate is We used Bayesian hierarchical random walk models to
likely to be no more than the observed rate), or ranges.16 produce estimates for 195 countries and territories
We had data for abortions and pregnancy intentions for (appendix p 2–3). For each country, resulting pregnancy
77 countries and, of these 77 countries, we had point data and abortion estimates were based on: (1) available
for both outcomes for 50 countries. abortion and pregnancy data, (2) infor­mation on contra­
ceptive needs and use by marital status, (3) model-based
Modelling strategy estimates of a series of parameters for variability, across
Our modelling strategy was informed by the proximate countries and periods, in subgroup-specific pregnancy
determinants of fertility.18,19 In brief, the incidence of rates and probabilities that a pregnancy would end in

Women Pregnancies Abortions Livebirths Fetal losses

Unintended Abortions from Livebirths from Fetal losses from

unintended pregnancies unintended pregnancies unintended pregnancies

Married Married Married Married Married

Unmet Unmet Unmet Unmet Unmet

Unmarried Unmarried Unmarried Unmarried Unmarried

Unmet Unmet Estimate Unmet Unmet Unmet
Estimate percent of
Married pregnancy Married pregnancies Married Married Married
Failure rates for each Failure aborted Failure Failure Failure
population for each
Unmarried group* Unmarried population Unmarried Unmarried Unmarried
Failure Failure group† Failure Failure Failure

Livebirths from Fetal losses from

Intended intended pregnancies intended pregnancies

Married Married Married Married

No need No need No need No need

Unmarried Unmarried Unmarried Unmarried

No need No need No need No need

Figure 1: Process model for estimating rates of unintended pregnancy and abortion
Unmet=unmet need for contraception. Failure=contraceptive failure. No need=no need for contraception. *A hierarchical time series model can be used to estimate pregnancy rates and propensities
to have an abortion. This model enables sharing of information across countries and time. †To fit the model, use data for women and births from the United Nations Population Division, and use all
data (by population group) for abortion and pregnancy intention. Fetal losses are calculated by the formula: 0·2 livebirths + 0·1 abortions.

www.thelancet.com/lancetgh Vol 8 September 2020 e1154

 Articles

abortion, and (4) birth rate estimates. Model assumptions, For results by SDG regional groupings, we presented
treatment of data (including accounting for random and major regions (except for Australia and New Zealand,
systematic error across, and within, data classes), and which we distinguished from Oceania). Results for
model validation exercises are detailed in the technical each subregion are in appendix p 5. For standardisation,
paper.16 we used each country’s classifications from 2019 when
reporting time trends by income group or the legal
Reported estimates status of abortion.17,26 For results by World Bank income
We computed point estimates with 80% uncertainty groups, we used country groups published in 2019.15
intervals (UIs) using the medians and 10th and We used the 2019 classification for legal status and
90th percentiles of posterior distributions of model World Bank income group for all time periods in the
parame­ters. A higher probability of change corresponded analysis to minimise estimated trends being an artifact of
to a greater certainty about the direction of change (ie, a countries shifting between categories during the analysis
greater or lesser rate of abortions or unintended period or due to changes in the classification methods.
pregnancies between two timepoints being compared). Additional groupings and indicators produced by our
We did not perform null-hypothesis significance testing, model that were beyond the scope of this Article are in
For the Guttmacher institute for several reasons.25 However, we present the posterior the appendix (pp 7–9) and on the Guttmacher Institute
website see probabilities wherever below 90%. and WHO websites.
gu.tt/GlobalAbortion
We computed point estimates of outcomes of interest
For the WHO website see
http://www.who.int/
using the posterior medians for each country and for each Role of the funding source
reproductivehealth/unintended- grouping of countries separately for each 5 year period. The funders of the study had no role in study design, data
pregnancy-and-abortion We report annual estimates for 5 year time periods (ie, all collection, data analysis, data interpretation, or writing of
rates, percentages, and numbers applied to each year in the report. The corresponding author had full access to all
the 5 year time period or, in other words, they were the the data in the study and had final responsibility for the
average across each 5 year span of time). For analyses in decision to submit for publication.
which countries were grouped by income and legality, we
also considered the possibility that an estimate might not Results
have been representative for most of the countries in that Incidence of unintended pregnancy
category. If large countries had estimates that differed Global levels and trends
substantially from other countries within the same group, In 2015–19, there were on average 121·0 million
this could obfuscate the relationship between income or unintended pregnancies each year (UI 112·8–131·5
legal status of abortion and rate of abortion or unintended million; appendix p 6), which corresponded to a global
pregnancy. Addressing this issue, we computed unweigh­ annual rate of 64 unintended pregnancies per
ted medians within groups of countries (appendix p 4). 1000 women aged 15–49 years (60–70) (figure 2;
When comparing two groups of countries, if the weighting appendix p 10). Among all pregnancies, 48% (46–51)
substantively affected the interpretation of the com­ were unintended (appendix p 7).
parison, we explained how. In 1990–94, the global annual unintended pregnancy
rate was 79 pregnancies (UI 74–84) per 1000 women aged
between 15–49 years (table 1). It declined by
100 Unintended pregnancy rate 12 points (7–15) between 1990–94 and 2000–04, to
Abortion rate
67 unintended pregnancies (63–72; appendix p 5)
Rate per 1000 women aged between 15 and 49 years

79
72 per 1000 women. The estimated decline in the global
75 67 66 65 64
unintended pregnancy rate was steepest before 2000 and
the rate was 64 unintended pregnancies (60–70) per 1000
women in 2015–19. Because of global population growth,
the annual number of unintended pregnancies increased
50 40 39 39
37 35 37 by 13% (4–22) between 1990–94 and 2015–19, from 107·6
million (101·0–115·3) to 121·0 million (112·8–131·5),
even as the rate declined, over the analysis period.
25
World Bank income groups
For World Bank income groups in 2015–19, the
0
unintended pregnancy rates were negatively associated
1990–94 1995–99 2000–04 2005–09 2020–14 2015–19 with income (figure 3). The average annual unintended
Period (years) pregnancy rate in high-income countries was 34 preg­
Figure 2: Global annual rates of unintended pregnancy and abortion per 1000 women aged between
nancies per 1000 women aged 15–49 years (UI 32–37),
15–49 years compared with 66 pregnancies per 1000 women aged
Shaded areas indicate 80% uncertainty intervals. 15–49 years (61–73) in middle-income countries. In

e1155 www.thelancet.com/lancetgh Vol 8 September 2020

 Articles

Unintended pregnancy rate per 1000 women aged Abortion rate per 1000 women aged between Unintended pregnancies ending in abortion (%)
between 15–49 years 15–49 years
1990–94 2015–19 Change from Probability 1990–94 2015–19 Change from Probability 1990–94 2015–19 Change from Probability
(80% UI) (80% UI) 1990–94 to of change (80% UI) (80% UI) 1990–94 to of change (80% UI) (80% UI) 1990–94 to of change
2015–19 (%) 2015–19 (%) 2015–19 (%)
(80% UI) (80% UI) (80% UI)
World 79 64 –18% 100% 40 39 –3% 62% 51 61 18% 100%
(74 to 84) (60 to 70) (–24 to –11) (37 to 44) (35 to 44) (–14 to 10) (48 to 54) (58 to 63) (12 to 26)
Sub-Saharan Africa 103 91 –12% 97% 27 33 24% 87% 26 37 41% 99%
(96 to 112) (86 to 96) (–19 to –4) (20 to 34) (29 to 38) (–3 to 65) (21 to 31) (34 to 40) (18 to 76)
West Asia and north 126 86 –31% 99% 61 53 –14% 71% 48 61 25% 96%
Africa (103 to 158) (67 to 114) (–45 to –16) (40 to 91) (34 to 78) (–40 to 21) (39 to 58) (51 to 69) (6 to 50)
Central and south 89 64 –28% 99% 40 46 15% 71% 45 72 59% 100%
Asia (76 to 107) (59 to 70) (–40 to –15) (29 to 56) (42 to 51) (–17 to 60) (37 to 54) (69 to 75) (35 to 93)
East and southeast 60 58 –4% 61% 38 43 13% 74% 63 74 16% 99%
Asia (53 to 70) (48 to 73) (–19 to 15) (35 to 42) (34 to 54) (–9 to 42) (54 to 70) (65 to 80) (7 to 29)
Latin America 94 69 –27% 100% 35 32 –8% 66% 37 47 26% 98%
(88 to 100) (61 to 79) (–35 to –16) (31 to 38) (25 to 41) (–28 to 21) (34 to 40) (41 to 53) (9 to 45)
Europe and northern 67 35 –47% 100% 46 17 –63% 100% 69 49 –29% 100%
America (62 to 72) (33 to 39) (–53 to –40) (42 to 50) (15 to 20) (–67 to –56) (66 to 72) (45 to 53) (–34 to –24)
Australia and New 42 38 –9% 80% 19 15 –19% 87% 45 41 –11% 79%
Zealand (36 to 48) (32 to 45) (–21 to 5) (18 to 20) (12 to 19) (–37 to 3) (39 to 53) (33 to 50) (–27 to 7)
Oceania (excluding 82 78 –4% 61% 22 34 51% 85% 28 44 54% 95%
Australia and New (61 to 116) (58 to 113) (–21 to 16) (8 to 53) (16 to 66) (–8 to 161) (13 to 48) (27 to 60) (8 to 147)
Zealand)
UI=uncertainty interval.

Table 1: Rates of unintended pregnancy and abortion, and proportion of unintended pregnancies ending in abortion, globally and for SDG regions between 1990–94 and 2015–19

low-income countries, the average annual unintended 120 Low-income countries

pregnancy rate was 93 pregnancies per 1000 women aged Middle-income countries
15–49 years (88–100). High-income countries

Since 1990–94, the unintended pregnancy rate declined 100

among all income groups (appendix p 5). High-income
Rate per 1000 women aged between 15 and 49 years

and middle-income countries experienced a decline of

21% (UI 15–27 for high-income countries and 13–28 for 80
middle-income countries), followed by a decline of 18%
(12–24) in low-income countries.
60
SDG regional groupings
Annual average unintended pregnancy rates varied across
SDG regions by nearly a factor of three (table 1;
40
appendix p 10). Rates were 35 unintended pregnancies per
1000 women aged 15–49 years (UI 33–39) in Europe and
northern America and 38 pregnancies per 1000 women
20
aged 15–49 years (32–45) in Australia and New Zealand.
Rates were nearest the global rate in east and southeast
Asia (58 unintended pregnancies per 1000 women
0
[48–73]), central and south Asia (64 pregnancies per 1000 Unintended pregnancy rate Abortion rate Unintended pregnancies
women [59–70]), and Latin America (69 pregnancies per ending in abortion (%)
1000 women [61–79]). The unintended pregnancy rate was Unintended pregnancy and abortion rates, and percentage of unintended pregnancies ending in abortion
by World Bank income groups
86 pregnancies per 1000 women (67–114) in west Asia and
north Africa, and 91 pregnancies per 1000 women (86–96) Figure 3: Unintended pregnancy and abortion rates, and percentage of unintended pregnancies ending in
in sub-Saharan Africa. abortion by World Bank income group, 2015–19
Error bars indicate 80% UIs.
We found evidence of a decline in the unintended
pregnancy rate throughout the SDG regions during the
30-year time period. The unintended pregnancy rate Africa. It declined by 28% (15–40) in central and south
declined by 47% (UI 40–53) in Europe and northern Asia and 27% (16–35) in Latin America. In sub-Saharan
America, and by 31% (16–45) in west Asia and north Africa, the rate declined by 12% (4–19).

www.thelancet.com/lancetgh Vol 8 September 2020 e1156

 Articles

Elsewhere, the unintended pregnancy rate declined between 15 and 49 years (13–16) in high-income countries.
from 1990–94 to 2015–19 by 4% (UI –15 to 19) in east and The corresponding percentages of unintended preg­
southeast Asia and 9% (–5 to 21) in Australia and New nancies ending in abortion were 66% (62–69) in middle-
Zealand. However, in those two regions, the probabilities income compared with 40% (37–44) in low-income
of change were 61% and 80% respectively. countries and 43% (40–46) in high-income countries.
From 1990–94 to 2015–19, high-income countries had a
Incidence of abortion and percent of unintended 31% (UI 24–37) decline in the abortion rate and a
pregnancies ending in abortion 12% (6–18) decline in the percent of unintended
Global levels and trends pregnancies ending in abortion (appendix p 8). Abortion
In 2015–19, there were 73·3 million (UI 66·7–82·0) rates in low-income and middle-income countries were
abortions each year on average (appendix p 6), which similar in 2015–19 compared with 1990–94. However, we
corresponded to a global annual rate of 39 abortions found evidence that the proportion of unintended
(35–44) per 1000 women aged 15–49 years. These pregnancies ending in abortion increased between these
estimates reflect that 61% (58–63) of unintended two periods by 22% (15–31) in middle-income countries
pregnancies ended in abortion. and by 26% (11–46) in low-income countries.
The global abortion rate was similar in 2015–19,
compared with 1990–94. However, there was evidence of SDG regional groupings
a decline through 2000–04 (with a probability of change Abortion rates varied across SDG regions by approxi­
of 97%) and then an increase (probability of change of mately a factor of three (table 1; appendix p 10), similar to
92%; figure 2). The proportion of unintended pregnancies the spread across regions in unintended pregnancy rates,
that ended in abortion increased 18% (UI 12–26) over in 2015–19. By contrast, the proportion of unintended
the 30-year period from 51% (48–54) to 61% (58–63) in pregnancies ending in abortion differed by up to a factor
2015–19 (table 1). of approximately two (table 1).
Comparing 1990–94 with 2015–19, there was the most
World Bank income groups evidence of a trend in Europe and northern America,
Whereas unintended pregnancy rates decreased mono­ where the annual average abortion rate declined by 63%
tonically across income groups in 2015–19, abortion rates (UI 56 to 67; table 1). We also found evidence that the
varied across income groups in a more nuanced average annual abortion rate declined in west Asia and
way (figure 3), being highest in middle-income countries, north Africa by 14% (–21 to 40), and in Australia and New
and lowest in high-income countries. In middle- Zealand by 19% (–3 to 37). However, the probability of
income countries, the annual average abortion rate was change for west Asia and north Africa was only 71% and
44 abortions per 1000 women aged between 15 and for Australia and New Zealand was only 87%. In sub-
49 years (UI 39–49), by contrast with 38 abortions per Saharan Africa, the abortion rate has been roughly level
1000 women aged between 15 and 49 years (33–43) in low- since 1995 (although there was some evidence of an
income countries and 15 abortions per 1000 women aged increase relative to 1990–94 [probability 87%]). In central

Unintended pregnancy rate per 1000 women aged Abortion rate per 1000 women aged between Unintended pregnancies ending in abortion (%)
between 15–49 years 15–49 years
1990–94 2015–19 Change from Probability 1990–94 2015–19 Change from Probability 1990–94 2015–19 Change from Probability
(80% UI) (80% UI) 1990–94 to of change (80% UI) (80% UI) 1990–94 to of change (80% UI) (80% UI) 1990–94 to of change
2015–19 (%) 2015–19 (%) 2015–19 (%)
(80% UI) (80% UI) (80% UI)
Abortion broadly legal 72 58 –19% 99% 44 40 –8% 73% 61 70 15% 100%
(66 to 80) (53 to 66) (–28 to –9) (39 to 49) (36 to 47) (–20 to 9) (56 to 65) (65 to 73) (8 to 23)
Abortion broadly legal 76 50 –34% 100% 46 26 –43% 100% 61 53 –13% 100%
(excluding India and (72 to 80) (46 to 54) (–39 to –29) (43 to 50) (24 to 30) (–49 to –36) (59 to 63) (50 to 56) (–18 to –8)
China)
Abortion restricted 91 73 –20% 100% 33 36 12% 82% 36 50 39% 100%
(86 to 97) (68 to 79) (–25 to –14) (28 to 38) (32 to 42) (–4 to 30) (32 to 39) (46 to 53) (27 to 53)
Abortion prohibited 110 80 –27% 100% 35 40 11% 70% 32 50 52% 100%
altogether (100 to 123) (70 to 91) (–35 to –19) (27 to 48) (31 to 51) (–14 to 40) (27 to 39) (44 to 55) (30 to 78)
Abortion permitted 86 70 –19% 100% 31 36 15% 85% 36 52 41% 100%
to save the woman’s (80 to 93) (63 to 77) (–26 to –12) (27 to 38) (30 to 43) (–3 to 35) (33 to 41) (48 to 56) (28 to 57)
life
Abortion permitted to 92 75 –18% 100% 33 36 8% 73% 36 47 32% 100%
preserve health (86 to 99) (70 to 81) (–24 to –12) (28 to 38) (31 to 41) (–8 to 27) (32 to 39) (44 to 51) (20 to 47)
UI=uncertainty interval.

Table 2: Rates of unintended pregnancy and abortion, and proportion of unintended pregnancies ending in abortion, by legal status of abortion for years 2015–19

e1157 www.thelancet.com/lancetgh Vol 8 September 2020

 Articles

120 Abortion broadly available

Abortion broadly available,
excluding China and India
Abortion restricted
100
Rate per 1000 women aged between 15 and 49 years

80

60

40

20

0
High income Middle income Low income High income Middle income Low income
Rate of unintended pregnancy by income group Rate of abortion by income group

Figure 4: Rates of unintended pregnancy and abortion for World Bank income groups and legal status of abortion, 2015–19

and south Asia, we found evidence of a decrease in the within this category, but skewed the estimates because
average annual abortion rate followed by an increase: of their large populations. Therefore the estimated
from 40 abortions per 1000 women (29–56) in 1990–94, averages when including these countries did not reflect
to 35 abortions (28–44) in 2004–05, to 46 abortions (42–51) the majority of countries within these categories
in 2015–19. However, the probability of change for the (appendix p 4).
initial decrease was only 78%, in contrast to 94%, for We found that unintended pregnancy rates were
the subsequent increase. generally higher in settings where abortion is restricted
Most regions saw an increase in the percent of than in settings where it is broadly legal (table 2). Where
unintended pregnancies ending in abortion during the abortion is restricted, the annual average unintended
30-year time period (table 1). The percent of unintended pregnancy rate was 73 (UI 68–79) per 1000 women in
pregnancies ending in abortion increased by 59% 2015–19 (table 2). The abortion rate for countries where
(UI 35–93) in central and south Asia, and by 41% (18–76) abortion is restricted was 36 (32–42), and the abortion
in sub-Saharan Africa. Elsewhere, we found evidence that rate was similar regardless of the type of legal restriction
the percent of unintended pregnancies ending in abortion (table 2). For countries where abortion is broadly legal,
increased by 16% (7–29) in east and southeast Asia, and the rates were 58 (53–66) for unintended pregnancy and
by 26% (9–45) in Latin America. A clear exception to this 40 (36–47) for abortion. The unintended pregnancy rate
general pattern was found in Europe and northern was 50 (46–54) and the abortion rate was 26 (24–30) for
America. There, the percent of unintended pregnancies countries where abortion is broadly legal, excluding
ending in abortion declined by 29% (24–34) from 1990–94 India and China.
to 2015–19. We found that around half of unintended pregnancies
ended in abortion—whether abortion is restricted or
Legal status of abortion broadly legal—excluding India and China (table 2).
Results for 2015–19 Among all countries where abortion is broadly legal, 70%
We grouped countries where abortion is prohibited (UI 65–73) of unintended pregnancies ended in abortion.
altogether, permitted only to save a woman’s life or to
preserve physical or mental health (referred to hereafter Results by income and legal status of abortion, 2015–19
as where abortion is restricted), and where abortion Within any World Bank income group, rates of unin­
is available on request or on broad socioeconomic tended pregnancy were higher in countries with restrictive
grounds (referred to hereafter as broadly legal; table 2). abortion laws (figure 4). High-income countries where
In these comparisons, we report averages across abortion is broadly legal had the lowest annual average
countries where abortion is broadly legal, as well as in rate of unintended pregnancy at 30 (UI 28–32) pregnancies
countries in this category except for China and India. per 1000 women aged between 15 and 49 years. The
There is evidence that these countries were atypical highest rate was 101 unintended pregnancies (94–110) per

www.thelancet.com/lancetgh Vol 8 September 2020 e1158

 Articles

1000 women aged between 15–49 years in low-income The substantial decline in the rate of unintended
countries with restrictive abortion laws. pregnancies is noteworthy because increased availability
In high-income countries where abortion is broadly of contraception and access to family planning services
legal, the annual rate of abortion was 11 abortions cannot guarantee a decline in unintended pregnancies.
(UI 11–12) per 1000 women aged between 15 and 49 years Although family planning programmes have led to an
(figure 4). The abortion rate in high-income countries increase in the prevalence of contraceptive use, changes
with restrictive laws, and middle-income and low- in fertility preferences, as well as the shift from
income countries regardless of legal status, was higher permanent methods, could both lead to increased rates
than that of high-income countries where abortion is of unintended pregnancy.
broadly legal, and ranged from 32 abortions (24–41) to For World Bank income groups, our findings indicated
48 abortions (42–56) per 1000 women aged between an inverse relation between income and unintended
15–49 years. pregnancy, resulting in substantial differences between
The proportion of unintended pregnancies ending in low-income and high-income countries. For 2015–19,
abortion was similar for high-income countries where low-income countries had the highest unintended
abortion is broadly legal (38% [UI 35–40]), and low- pregnancy rate and the lowest proportion of unintended
income countries where abortion is legally restricted pregnancies ending in abortion. These findings could
(39% [34–43]). Among other income and legality relate to global economic inequalities that influence
groupings, the proportion of unintended pregnancies people’s ability to access sexual and reproductive health
ending in abortion varied with no clear pattern, from services, including contraception services and abortion
44% (40–49) in low-income countries where abortion is care. However, this evidence is descriptive and we caution
broadly legal, to 76% (70–80) in middle-income countries against inter­preting these patterns as though they show
where abortion is broadly legal. a causal effect of income.
We generally found evidence of a decline in the average
Trends by the legal status of abortion annual unintended pregnancy rate throughout SDG
From 1990–95 to 2015–19, the abortion rate declined by regions. The trends in abortion rates were more nuanced,
43% (UI 36 to 49) in settings where abortion is broadly which can be explained by how the proportion of
legal, excluding China and India. The abortion rate unintended pregnancies ending in abortion generally
increased by 12% (–4 to 30) in countries that highly increased throughout the 30 year period. This trend
restrict access to abortion, although the probability of could reflect increases in access to abortion, such as
change was only 82% (appendix p 4). through the spread of medication abortion, or a stronger
The percent of unintended pregnancies ending in motivation to avoid unintended births. The one exception
abortion was not always similar in countries where is the substantial decline in the unintended pregnancy
abortion is legally restricted compared with countries and abortion rates in Europe and northern America.
where abortion is broadly legal. In countries that restric­ We found no evidence that abortion rates were lower in
ted abortion, the percent of unintended pregnancies settings where abortion was restricted. The gradation in
ending in abortion increased in every 5 year period in our type of restrictions also made little difference on the
analysis. The cumulative increase between 1990–94 and abortion rate due to unintended pregnancy rates being
2015–19 was 39% (UI 27–53; table 2). By contrast, in substantially higher in countries where abortion was
countries where abortion is broadly legal, excluding China restricted. This finding means that some women in these
and India, there was a 13% (8–18) decrease in the percent restrictive settings must take legal and physical risks to
of unintended pregnancies ending in abortion. seek abortion care, corroborating findings in earlier
studies28–31 that show that women with unintended
Discussion pregnancies rely on abortion even in settings where
Over the 30 year period analysed there had been a decline abortion is restricted.
in the unintended pregnancy rate, which shows that Moreover, we found that China and India, which
more people than ever, as a proportion of the population comprised 62% of women who were at reproductive age
who are at reproductive age, were able to limit or space in countries where abortion was broadly legal, skewed the
their childbearing.27 Despite this decline, approximately averages in countries where abortion was broadly legal.
half of all pregnancies were unintended. By contrast to Averaging among all other countries where abortion is
declines in the global unintended pregnancy rate, the broadly legal, abortion rates were higher among countries
global abortion rate was roughly equal in 2015–19 and where abortion was restricted. These countries could
1990–94. Declines in the global unintended pregnancy differ for several reasons, including differences in fertility
rate slowed after 2000–04, after which the proportion of preferences, and access to and use of contraception.
unintended pregnancies ending in abortion began to We also found that the proportion of unintended
increase. It is possible that increases in individuals’ pregnancies ending in abortion increased in countries
motivation to avoid an unintended birth, as well as where abortion was legally restricted. This finding points
greater access to services, contributed to both trends. to the need for research to understand how individuals

e1159 www.thelancet.com/lancetgh Vol 8 September 2020

 Articles

obtain abortions, particularly in legally restrictive Our estimated numbers of abortions and unintended
settings, the safety of abortions in these settings,32 and pregnancies were higher than previous estimates.
the consequences of unsafe abortion on health and However, previous unintended pregnancy estimates were
wellbeing. Other aspects of people’s wellbeing, including within our UIs. Other studies previously estimated that
their experience of mistreatment and stigma,33,34 also 111·7 million unintended pregnancies4 and 59·3 million
warrant further study. abortions11 occurred annually between 2010–14, whereas
High-income countries where abortion is broadly legal our estimates for the same period were 118·0 million
had the lowest unintended pregnancy rate, abortion unintended pregnancies (UI 109·5–128·0) and 71·4 mil­
rate, and proportion of unintended pregnancies ending lion abortions (64·6–79·7).
in abortion. Among middle-income and low-income Similarly to Sedgh and colleagues,12 we estimated that
countries, there was not a clear relationship between the global abortion rate declined for approximately
legal restrictions and abortion rates, or the proportion 15 years after 1990–94. However, whereas they found
of unintended pregnancies ending in abortion. These little evidence of an increase thereafter, we found that,
findings could reflect differences in the quality and by 2015–19, the abortion rate returned to a similar rate to
capacity of national health systems in low-income and that estimated for 1990–94. This change was largely due
middle-income countries, and show the need for to differences in the model-based estimates for Asia,
additional investment in sexual and reproductive health although there are smaller differences (in either
care. Research is also needed for obstacles people face direction) in other regions.
(eg, social and economic obstacles) to exercising their Previous studies defined women of reproductive age as
reproductive autonomy, and the policies and programmes being between the ages of 15–44 years. Our study used
that can most effectively ensure all peoples’ sexual and the ages of 15–49 years to be consistent with the UN
reproductive health and rights. family planning indicators and the data used in our
This study has several limitations. We presented study.
estimates for broad regions and groupings of countries, One reason that our findings differed from previous
which masked variation across countries and population studies was that our statistical model closely corresponded
groups. We contextualised abortion within the context of to our theoretical framework, in contrast to earlier studies.
unintended pregnancy, but abortions can also occur Family planning indicators were used in our study to
following intended pregnancies. In addition, abortion predict unintended pregnancy, but Sedgh and colleagues
data are relatively sparse and have high UIs in countries used these indicators to estimate abortion. Their statistical
with restrictive laws compared with countries that do model had to assume no differences across countries
not, and data for unplanned birth rates are relatively within subregions in the proportion of unintended
sparse in countries with liberal laws compared with pregnancies ending in abortion because it could not use
countries with restrictive laws. data for the proportion of births that were unintended.
The increasing availability of medication abortion (ie, Bearak and colleagues4 previously produced estimates of
misoprostol with or without mifepristone) poses unintended pregnancies informed by this theoretical
challenges for measuring abortion incidence in many framework but, in their study, the unknowns to be
countries as use of these drugs occurs both inside and modelled were the proportions of births unintended; the
outside of formal health systems, meaning that it is likely abortion estimates from Sedgh and colleagues were
that abortions are occurring outside of the health sector treated as fixed. As a result, the pregnancy intention data
that we could not capture. Also, although marital status did not inform the previously published abortion
as well as contraceptive needs and use are key proximate estimates, and uncertainty in the abortion estimates were
determinants of pregnancy, these determinants alone do ignored when modelling unintended births.
not explain all differences between time periods or Finally, we built a substantially larger database than
between countries. These limitations, which were evident previous studies. Previous studies obtained abortion data
in the wide UIs for settings in which outcome data were for 92 countries and intention data from 105 countries.4,11
limited, reflect the continued need for—and investment In our study, we obtained abortion data for 104 countries
in—abortion research, especially in regions and and intention data from 139 countries. Our joint
countries with little available data. estimation approach and database constituted a
There were also many strengths to our estimation substantially more comprehensive effort.
approach. To the best of our knowledge, we incorporated
more data into our model than any of the previously Conclusion
published estimates of global trends in abortion and The Guttmacher-Lancet Commission recommended
unintended pregnancy, and we took advantage of newly a comprehensive package of essential sexual and
available estimates of unmet need among unmarried reproductive health and rights services, including
women. In addition, we used, for the first time, a contraception and safe abortion care, for inclusion in
comprehensive model that jointly estimated unintended national health systems.9 Our findings emphasise that
pregnancy and abortion. unintended pregnancy and abortion are experiences that

www.thelancet.com/lancetgh Vol 8 September 2020 e1160

 Articles

are shared by many people globally, regardless of region, 14 Sedgh G, Henshaw S, Singh S, Ahman E, Shah IH. Induced
income group, and legal status. Our findings highlight abortion: estimated rates and trends worldwide. Lancet 2007;
370: 1338–45.
the need for continued commitment and investment 15 The World Bank. World Bank country and lending groups. 2019.
to ensure access to the full spectrum of quality https://datahelpdesk.worldbank.org/knowledgebase/
comprehensive sexual and reproductive health care. articles/906519-world-bank-country-and-lending-groups (accessed
Feb 12, 2020).
Fulfilling these commitments will not only result in 16 Bearak J, Popinchalk A, Ganatra B, et al. Global estimation of
better outcomes for all, but are also necessary to achieve unintended pregnancy and abortion using a Bayesian hierarchical
the targets for the Global Strategy for Women’s, random walk model. 2020. https://arxiv.org/abs/2007.09246
(accessed July 21, 2020).
Children’s and Adolescent’s Health (2016–30)8 as well as
17 Stevens GA, Alkema L, Black RE, et al. Guidelines for Accurate and
the SDGs, and universal health coverage. Transparent Health Estimates Reporting: the GATHER statement.
Contributors PLoS Med 2016; 13: e1002056.
JB and LA contributed to the conceptualisation of the estimation 18 Bongaarts J. A framework for analyzing the proximate determinants
approach. JB developed the statistical model and did the data analysis of fertility. Popul Dev Rev 1978; 4: 105.
with input from LA and AP. AP compiled and managed the data, 19 Bongaarts J. Modeling the fertility impact of the proximate
determinants: time for a tune-up. Demogr Res 2015; 33: 535–60.
with contributions from CB and LK. BG, OT, and A-BM led the country
consultation. JB prepared the first draft of the manuscript, with 20 UN. Estimates and projections of family planning indicators 2020.
New York: United Nations. https://www.un.org/en/development/
assistance from AP, and subsequent drafts were revised by AP with
desa/population/theme/family-planning/cp_model.asp (accessed
inputs from all coauthors. All coauthors convened periodically for
April 22, 2019).
technical exchanges about the estimation approach.
21 UN. World contraceptive use 2019. 2019. https://www.un.org/en/
Declaration of interests development/desa/population/publications/dataset/contraception/
We declare no competing interests. wcu2019.asp (accessed July 16, 2020).
22 Kantorova V. National, regional and global estimates and projections
Acknowledments of the number of women aged 15 to 49 who are married or in a union,
The authors alone are responsible for the views expressed in this Article, 1970–2030. New York: United Nations Population Division, 2013.
and they do not necessarily represent the views, decisions, or policies of 23 United Nations. World marriage data 2017. New York: Nations,
the institutions with which they are affiliated. Department of Economic and Social Affairs, Population Division,
References 2017. http://www.un.org/en/development/desa/population/
1 Bongaarts J, Mensch BS, Blanc AK. Trends in the age at reproductive publications/pdf/marriage/Metadata_World-Marriage-Data-2017.pdf
transitions in the developing world: the role of education. Popul Stud (accessed Nov 5, 2018).
2017; 71: 139–54. 24 Dellicour S, Aol G, Ouma P, et al. Weekly miscarriage rates in a
2 Darroch JE. Trends in contraceptive use. Contraception 2013; community-based prospective cohort study in rural western Kenya.
87: 259–63. BMJ Open 2016; 6: e011088.
3 Bongaarts J. Can family planning programs reduce high desired 25 WHO. The global abortion policies database. https://abortion-
family size in sub-Saharan Africa? Int Perspect Sex Reprod Health policies.srhr.org/(accessed Nov 22, 2019).
2011; 37: 209–16. 26 Center for Reproductive Rights. The world’s abortion laws. 2019.
4 Bearak J, Popinchalk A, Alkema L, Sedgh G. Global, regional, https://reproductiverights.org/worldabortionlaws (accessed
and subregional trends in unintended pregnancy and its outcomes July 16, 2020).
from 1990 to 2014: estimates from a Bayesian hierarchical model. 27 USAID. Desired number of children: 2000–2008. 2010.
Lancet Glob Health 2018; 6: e380–89. https://www.dhsprogram.com/pubs/pdf/CR25/CR25.pdf (accessed
5 Sedgh G, Singh S, Hussain R. Intended and unintended pregnancies July 16, 2020).
worldwide in 2012 and recent trends. Stud Fam Plann 2014; 28 Sedgh G, Sylla AH, Philbin J, Keogh S, Ndiaye S. Estimates of the
45: 301–14. incidence of induced abortion and consequences of unsafe abortion
6 Ganatra B, Gerdts C, Rossier C, et al. Global, regional, and in Senegal. Int Perspect Sex Reprod Health 2015; 41: 11.
subregional classification of abortions by safety, 2010–14: estimates 29 Chae S, Kayembe PK, Philbin J, Mabika C, Bankole A. The incidence
from a Bayesian hierarchical model. Lancet 2017; 390: 2372–81. of induced abortion in Kinshasa, Democratic Republic of Congo,
7 United Nations. Transforming our world: the 2030 agenda for 2016. PLoS One 2017; 12: e0184389.
sustainable development. New York, NY: United Nations, 2015. 30 Polis CB, Mhango C, Philbin J, Chimwaza W, Chipeta E, Msusa A.
8 Every Woman, Every Child. Global strategy for women’s, children’s Incidence of induced abortion in Malawi, 2015. PLoS One 2017;
and adolescent’s health. Every Woman Every Child. Every Woman 12: e0173639.
Every Child. June 24, 2016. https://www.everywomaneverychild.org/ 31 Sully EA, Madziyire MG, Riley T, et al. Abortion in Zimbabwe:
global-strategy/ (accessed Sept 18, 2017). a national study of the incidence of induced abortion, unintended
9 Starrs AM, Ezeh AC, Barker G, et al. Accelerate progress— pregnancy and post-abortion care in 2016. PLoS One 2018;
sexual and reproductive health and rights for all: report of the 13: e0205239.
Guttmacher–Lancet Commission. Lancet 2018; 391: 2642–92. 32 Singh S, Remez L, Sedgh G, Kwok L, Onda T. Abortion worldwide
10 Bearak JM, Popinchalk A, Sedgh G, et al. Pregnancies, abortions, 2017: uneven progress and unequal access. New York: Guttmacher
and pregnancy intentions: a protocol for modeling and reporting Institute, 2018.
global, regional and country estimates. Reprod Health 2019; 16: 36. 33 Gerdts C, Raifman S, Daskilewicz K, Momberg M, Roberts S,
11 Sedgh G, Bearak J, Singh S, et al. Abortion incidence between 1990 Harries J. Women’s experiences seeking informal sector abortion
and 2014: global, regional, and subregional levels and trends. Lancet services in Cape Town, South Africa: a descriptive study.
2016; 388: 258–67. BMC Womens Health 2017; 17: 95.
12 Sedgh G, Singh S, Shah IH, Åhman E, Henshaw SK, Bankole A. 34 Jayaweera RT, Ngui FM, Hall KS, Gerdts C. Women’s experiences
Induced abortion: incidence and trends worldwide from 1995 to with unplanned pregnancy and abortion in Kenya: a qualitative
2008. Lancet 2012; 379: 625–32. study. PLoS One 2018; 13: e0191412.
13 Henshaw SK, Singh S, Haas T. The incidence of abortion
worldwide. Int Fam Plan Perspect Dig 1999; 25: S30–38.

e1161 www.thelancet.com/lancetgh Vol 8 September 2020