MXQ 012

Epidemiologic Reviews Vol.
32, 2010
ª The Author 2010. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of DOI: 10.1093/epirev/mxq012
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org. Advance Access publication:
June 22, 2010
Family Planning and the Burden of Unintended Pregnancies
Amy O. Tsui*, Raegan McDonald-Mosley, and Anne E. Burke

* Correspondence to Dr. Amy O. Tsui, Population, Family and Reproductive Health Department, Johns Hopkins Bloomberg School
of Public Health, 615 North Wolfe Street, W4041, Baltimore, MD 21205 (e-mail: atsui@jhsph.edu).
Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Accepted for publication April 29, 2010.
Family planning is hailed as one of the great public health achievements of the last century, and worldwide
acceptance has risen to three-fifths of exposed couples. In many countries, however, uptake of modern contra-
ception is constrained by limited access and weak service delivery, and the burden of unintended pregnancy is still
large. This review focuses on family planning’s efficacy in preventing unintended pregnancies and their health
burden. The authors first describe an epidemiologic framework for reproductive behavior and pregnancy intended-
ness and use it to guide the review of 21 recent, individual-level studies of pregnancy intentions, health outcomes,
and contraception. They then review population-level studies of family planning’s relation to reproductive, maternal,
and newborn health benefits. Family planning is documented to prevent mother-child transmission of human
immunodeficiency virus, contribute to birth spacing, lower infant mortality risk, and reduce the number of abortions,
especially unsafe ones. It is also shown to significantly lower maternal mortality and maternal morbidity associated
with unintended pregnancy. Still, a new generation of research is needed to investigate the modest correlation
between unintended pregnancy and contraceptive use rates to derive the full health benefits of a proven and cost-
effective reproductive technology.
contraception; contraceptive behavior; family planning services; pregnancy outcome; pregnancy, unplanned;
reproduction
Abbreviations: CI, confidence interval; DHS, Demographic and Health Surveys; HIV, human immunodeficiency virus; IUD,
intrauterine device; OR, odds ratio.
INTRODUCTION ing primarily on the pill, female and male sterilization, and
condoms.
In 1999, the Centers for Disease Control and Prevention The prevalence of contraceptive use is similarly high in
identified family planning as one of 10 great public health European, many Latin American, and east and southeast
achievements in the United States during the 20th century Asian countries. Contraceptive use among partnered women
(1). Alongside other achievements, such as vaccination and aged 15–49 years in the developing world rose from 14% in
control of infectious diseases, access to family planning the mid-1960s (2) to 62% in 2008 (3) and from protecting
and contraceptive services was cited for social, economic, approximately 70 million to more than 600 million couples
and health benefits conferred through ‘‘smaller family from unintended pregnancies. Rapid adoption of contracep-
size and longer interval between the birth of children; in- tion has been documented in countries as diverse as Thai-
creased opportunities for preconceptional counseling and land, Iran, Egypt, and Colombia between the mid-1980s and
screening; fewer infant, child, and maternal deaths; and the mid-2000s (4). In low-income countries in sub-Saharan
use of barrier contraceptives to prevent pregnancy and Africa, south Asia, and Central America, use of modern
transmission of human immunodeficiency virus and contraception is more modest and is constrained by limited
other STDs [sexually transmitted diseases]’’ (1, p. 241). access to services and weak government programs. While
In the United States, contraceptive use among all types of contraceptive methods used vary across regions, the
women 15–44 years of age in 2002 was 61.9% in 2002 health and social benefits of family planning are widely
and considerably higher (72.9%) among married women. accepted across much of the world. Public sponsorship has
More than 45 million US women use contraception, rely- launched most national family planning programs targeting
152 Epidemiol Rev 2010;32:152–174

Family Planning and Unintended Pregnancies 153
low-income users, but modern contraceptive use has risen transmitted infection, unintended pregnancy, fetal wastage,
through sustained individual demand often met by an ex- stillbirth, and maternal and neonatal mortality. Successful
pansion of care from private providers. progression through these events can be measured by post-
Globally, the strength of government commitment tends partum health and survival of mothers and infants. Many
to be greater than actual funding levels or program imple- health technologies, including contraception, increase the
mentation efforts. Family planning has been cited as essen- likelihood that each transition occurs successfully. Table 1
tial to the achievement of Millennium Development Goals and Figure 1 convey the epidemiologic links among the
(5) by former United Nations Secretary General Kofi Annan events and the key interventions, particularly family plan-
(6), and, as such, part of the fifth Millennium Development ning, associated with the pathways.
Goal targets universal access to family planning as a key Starting with coital activity, the probability of conception
strategy for improving maternal health. The proportion of during any random act of intercourse is thought to range
governments in less-developed countries that provide direct between 3% and 5% (10). A recent study of conception risk
or indirect support for contraceptive access grew from 64% by menstrual cycle day among a small sample of North

in 1976 to 87% in 2009 (7). Global domestic spending on Carolina women found a likelihood of 3.1% per coition
population activities—which includes family planning, re- overall (11). In addition to pregnancy, coital activity also
productive health, sexually transmitted diseases/human im- carries the risk of sexually transmitted infection. The prob-
munodeficiency virus (HIV)/acquired immunodeficiency ability of acquiring an infection will depend on multiple
syndrome, and basic research—by governments, nongov- factors, including exposure to a partner infected with an
ernmental organizations, and consumers reached $18.5 offending organism as well as number of partners and con-
billion in 2006, but nearly half (45%) was allocated to sex- dom use. For example, Wawer et al. (12) found the proba-
ually transmitted diseases/HIV/acquired immunodeficiency bility of HIV-1 transmission to be 0.0082 (95% confidence
syndrome (8). Estimates of donor assistance in 2008 for interval (CI): 0.0039, 0.0150) per coital act in Rakai,
this sector total $10.6 billion, but only $0.25 billion Uganda. Sex protected by contraception and consistent con-
(2.4%) is directed toward supporting family planning (9), dom use addresses both pathways a and f in Figure 1 and
or approximately US $0.17 per woman of childbearing age reduces pregnancies that are unplanned and/or exposed to
in developing countries. sexually transmitted infections. Sexually transmitted infec-
The term ‘‘family planning’’ has been used synony- tion, such as chlamydia, gonorrhea, and syphilis, increases
mously with contraceptive practice, although the ability the risk of preterm birth, low birth weight, and stillbirth
to decide the number and timing of births can be achieved (i, d). Recurrent infections cause subfecundity (g) and
by a range of means, including contraception and assisted- infertility (h).
reproductive technologies. Voluntary abstinence—either The probability that conceptions become viable fetuses (b)
permanent or intermittent—elective abortion, and artificial and progress to term (d) is enhanced by maternal nutritional
insemination are other means commonly used by individ- well-being before pregnancy and nutritional status during
uals to achieve reproductive intentions. In this review, we gestation (13). The prevalence of spontaneous abortion can
focus on contraception and address unsafe abortion as a pre- range from 5% to 70% of pregnancies, depending on stage of
ventable outcome of failed contraceptive behavior or development. Worldwide, 22% of pregnancies, or about 42
methods. million, are electively terminated (c), of which 20 million
Our review incorporates both population-level and terminations happen under unsafe conditions, mostly in the
individual-level perspectives in assessing the research evi- developing world (14). Contraception plays a key role in re-
dence of contraceptive practice’s relation to the burden of un- ducing reliance on elective abortions and can avert as many
intended pregnancies. The review has 4 parts. After framing as 13%–15% of the maternal deaths that result from unsafe
the behavioral epidemiology that links sexual, reproductive, abortions (a-b-c-m). Figure 1 highlights the significance of
maternal, and newborn health outcomes, we briefly detail the protected coitus not only in preventing unintended pregnancy
measurement of unintended pregnancy and contraceptive and sexually transmitted infection but also in lowering expo-
practice. Next, we review findings from recent individual-level sure to subsequent morbidity and mortality risks.
studies of 1) fertility intentions and pregnancy and maternal
outcomes, 2) fertility intentions and contraceptive behaviors, Terminology and definitions in unintended pregnancy
and 3) contraceptive behaviors and unintended pregnancy research
outcomes. In the fourth part, we review research on the
population-level health implications of family planning need. Fertility-intention measures implicitly require individual
cognition of the ability to control the timing and number of
A reproductive behavioral epidemiology framework pregnancies. This recognition is near, but not completely,
universal in the world. Santelli et al. rightly describe preg-
With broad acceptance of contraception as a modern nancy intendedness as a ‘‘complex concept . . . encompassing
health technology, why do unintended pregnancies still con- affective, cognitive, cultural and contextual dimensions’’ (15,
stitute a health burden for women and their partners? The p. 94). The persistence and stability of individual fertility
answers lie in the health risks associated with sexual activity intentions, and thus their predictive value, have been ques-
and reproduction. Coition, conception, viable pregnancy, tioned in a number of studies (16–18).
fetal growth, parturition, and the puerperium separately Because population-level measures are primarily as-
carry health risks or undesirable outcomes, such as sexually sessed through cross-sectional, national household surveys,
Epidemiol Rev 2010;32:152–174

154 Tsui et al.
Table 1. Common Measures for Reproductive Epidemiology Outcomes and Range of Values
Observed Average
Outcome and Geographic Illustrative Key
Numerator Denominator Constant or Range (Study Pathway(s)b
Measure Reference Intervention
Source)a
Coital activity
Coition rate Coitions Day 100 0.12–0.27 per day Across 9 sub- a Contraception,
(Brown, 2000 (86)) Saharan African sexuality education
countries f Consistent
condom use
Conception
Conception rate Conception Coitions 100 2%–4% (Tietze, Wilcox et al.: a Contraception
1960 (10)); 3.1% North Carolina b Fetal nutrition
(Wilcox et al., 2001
supplementation
(11))

Pregnancy rate Pregnancies Exposed women 1,000 137 per 1,000 Worldwide a Contraception
women aged 15–44
b Fetal nutrition
years (Singh et al.,
supplementation
2009 (14))
Unintended Unintended Exposed women 1,000 57 per 1,000 Developing a Contraception
pregnancy rate pregnancy women aged 15–44 countries
c Elective
years (Singh et al., pregnancy
2009 (14))
termination
Sexually
transmitted
infection
Type-specific Type of Coitions 1,000 0.0082 (95% CI: HIV-1 transmission g Antibiotics,
infection rate acquired 0.0039, 0.0150) rate per coital act in sexual abstinence
infection (Wawer et al., 2005 Rakai, Uganda
h Antibiotics
(12))
f and i Consistent
condom use with
pregnant partner,
ART
Fetal growth
Preterm birth rate Livebirths Births 100 9.6% (95% CI: 6.2– Worldwide d Maternal and
occurring at 11.9) births (Beck fetal nutrition during
<37 weeks of et al., current and prior
gestation 2010 (87)) pregnancies
(contraception for
birth spacing)
Fetal mortality Fetal deaths Pregnancies 100 14.7% (Ventura United States in d Smoking
ratio et al., 2009 (88)) 2005 cessation,
elimination of
alcohol and other
drug use in
pregnancy
Pregnancy
termination
Miscarriage rate Spontaneous Exposed women 1,000 19.4 per 1,000 United States in d Smoking
fetal losses women aged 15–44 2005 cessation,
years (Ventura elimination of
et al., 2009 (88)) alcohol and other
drug use in
pregnancy
Induced Induced Exposed 1,000 29 per 1,000 Worldwide a Contraception
abortion rate abortions women women aged 15–44 to prevent
years (Singh et al., unwanted
2009 (14)) pregnancies, before
and after abortion
j, m Safe abortion
procedures,
postabortion
complications
management
Table continues

Table 1. Continued
Observed Average
Outcome and Geographic Illustrative Key
Numerator Denominator Constant or Range (Study Pathway(s)b
Measure Reference Intervention
Source)a
Induced Induced Pregnancies 100 41 in developed Worldwide a Contraception

abortion ratio abortions and 23 in to prevent
developing regions unwanted
in 1995 pregnancies
(Guttmacher
j, m Safe abortion
Institute, 1999 (89)) procedures,
postabortion
complications
management
Parturition
Birth rate Livebirths Exposed 1,000 39 (Ukraine) to 245 Worldwide a Contraception

(general fertility women (Niger) (Measure to avoid unintended
rate) DHS StatCompiler pregnancies
(47))
d Prenatal and
delivery care,
maternal and fetal
nutrition,
postpartum family
planning
Maternal Maternal Livebirths 100,000 3 (Denmark) to Worldwide e/k Management of
mortality ratio deaths 2,100 (Sierra complications
Leone) per 100,000 (hemorrhage,
livebirths (WHO, toxemia, sepsis);
2007 (90)) provision of
emergency
obstetric services,
skilled birth
attendance
n Cesarean
section
Stillbirth rate Stillbirths (fetal Deliveries 1,000 5 in developed Worldwide a Consistent
deaths in the countries to 32 in condom use
last 12 weeks of Africa and South
e Treatment of
pregnancy) Asia (Stanton et al., syphilis infection
2006 (91))
Infant mortality
Infant mortality Deaths of Livebirths 1,000 4 in western Worldwide l Infection
rate infants at <12 European to 95 in prevention,
months of age middle African immunization,
countries breastfeeding,
(Population rehydration;
Reference Bureau, contraception for
2009 (92)) birth spacing
Perinatal Deaths of Livebirths 1,000 6 in western Worldwide l Infection
mortality rate infants within Europe to 76 in prevention,
the first 7 days western Africa immunization,
(WHO, 2006 (93)) breastfeeding,
rehydration;
contraception for
birth spacing
Neonatal Deaths of Livebirths 1,000 3 in western Europe Worldwide l Infection
mortality infants within to 49 in western prevention,
the first 28 days Africa (WHO, 2006 immunization,
(93)) breastfeeding,
rehydration;
contraception for
birth spacing
Postneonatal Deaths of Livebirths 1,000 2.2 in the United Worldwide l Infection
mortality infants at States in 2006 prevention,
between 28 (NCHS), 4.3 in immunization,
days and 1 year Georgia in 2005, breastfeeding,
of life 63.5 in Swaziland in rehydration
2006 (MEASURE
DHS StatCompiler
(47))
Abbreviations: ART, antiretroviral therapy; CI, confidence interval; DHS, Demographic and Health Surveys; HIV, human immunodeficiency virus; NCHS, National
Center for Health Statistics; WHO, World Health Organization.
a
Observed average or range as given in the cited source (author(s), year, (reference no.)) of cross-national variation.
b
Italic letters refer to pathways identified in Figure 1.

156 Tsui et al.
unmarried women, and especially adolescents, may not be

included.
Terminology and definitions in family planning

research
Modern contraceptive methods can be categorized in sev-

eral ways. Hormonal methods include such products as oral
contraceptives, patches, vaginal rings, injectables, implants,
and levonorgestrel intrauterine contraception. Nonhormonal
methods include male and female condoms and other barrier
methods, as well as copper intrauterine devices (IUDs).
Implants and intrauterine contraception, and sometimes in-
jectables, are also categorized as long-acting, reversible

contraceptive methods. Surgical sterilization is a permanent
Figure 1. Pathways between sexual, reproductive, maternal, and method of family planning.
newborn health outcomes. Motivational pathways are represented Contraceptive effectiveness is a measure of the success of
by dashed lines, behavioral pathways by solid lines.
typical use of a method. It incorporates efficacy, that is, how
well a method works when used consistently and correctly,
and a host of other factors, such as ease of compliance. Gen-
such as the National Survey of Family Growth, the Demo- erally, long-acting, reversible methods and sterilization are the
graphic and Health Surveys (DHS), and the Reproductive most effective (>99% protection against pregnancy over
Health Surveys, pregnancy intendedness is based on female a year of use), with very low pregnancy rates among typical
respondents’ retrospective, potentially biased recall of the users that approach perfect-use rates. Once initiated, these
planned status of the last or a recent pregnancy: Right before methods are relatively user independent. Shorter-
you became pregnant with your (nth, last) pregnancy . . . , acting hormonal contraceptives are generally in the next tier
Thinking back to just before you got pregnant with [name of of effectiveness. Included are such methods as pills, patches,
child] . . . , or At the time you became pregnant with [name and vaginal rings. These methods have high efficacy, but po-
of child] . . . . Prospective studies assess women’s future tential problems with compliance (missed doses, unreliable
pregnancy intentions and their strength: Are you trying to supply) result in higher real-world pregnancy rates. For exam-
get (or keep from getting) pregnant now? How important is ple, the typical pregnancy rate for the combined oral contra-
avoiding a pregnancy to you? Furthermore, in most studies, ceptive pill is 8% in the first year of use (22). Barrier methods
pregnancy intendedness is dichotomized (intended/unin- are somewhat less effective (pregnancy rates of 15%–32%),
tended, wanted/unwanted), despite recognition that it is followed by contraception that relies on timed intercourse,
a complex and nuanced concept. This dichotomization such as withdrawal or fertility-awareness methods.
may be due to the limitations of data collection or measure- Perhaps one of the best-measured reproductive behaviors,
ment instruments, but it does raise questions about what is contraceptive practice has been assessed extensively at the
being missed in current analyses. population level for more than 4 decades by using the ‘‘con-
The desired timing of the next pregnancy is used to assess traceptive prevalence rate.’’ Technically, this is not a rate but
unintendedness. Generally, a pregnancy that follows a wom- a proportion—the percentage of exposed women reporting
an’s report of not wanting any additional births is classified current use of any contraception, including male methods.
as ‘‘unwanted,’’ whereas one that happens before a desired Exposure involves being of reproductive age (15 years to 44
point in time for her is ‘‘mistimed.’’ A pregnancy desired at or 49 years) and sexually active or in a marital or stable
the time is considered ‘‘wanted.’’ Some pregnancies to union. Contraceptive ‘‘method mix’’ is also a measure of
women who are unsure of their intendedness are classified much interest because it is a proxy for method availability
as being of ‘‘unsure’’ or ‘‘ambivalent’’ status. Most studies and client choice (23). It may reflect preferences of women
reviewed here adopt a 3-level classification, with ‘‘unsure’’ or couples or it may reflect limits regarding supply or pro-
combined with ‘‘wanted’’ intendedness. vider bias (24).
Pregnancy-intention measures provide the denominator
for unmet contraceptive-need measures in the developing MATERIALS AND METHODS
world (19). The most widely used concept of unmet contra-
ceptive need is a woman exposed to the risk of pregnancy Guided by the framework in Figure 1, we reviewed the
and not currently using contraception who wants to space or recent research literature on the magnitude and strength of
limit future childbearing. The standard DHS definition for the relation between pregnancy intentions and reproductive,
unmet need means that a woman 1) is married or in a con- maternal, and newborn health outcomes first (Table 2) and
sensual union, 2) is between the ages of 15 of 49 years, 3) is then contraception. Table 3 summarizes studies of pregnancy
capable of becoming pregnant, 4) wants to have no more intentions and contraception behaviors, and Table 4 includes
children or no children for at least 2 years, and 5) is using studies of contraceptive behaviors and pregnancy out-
neither a traditional nor a modern method of contraception comes, specifically the incidence of unintended pregnancy
(20, 21). One obvious limitation of this definition is that and elective pregnancy termination. Nearly all studies are

observational and most cross-sectional, limiting the rigor of We identified 3 studies, all in the United States, that in-
the evidence and the reliability of further synthesis. cluded multivariable analyses of population-based data, and
To investigate the association of pregnancy intentions one prospective survey. ‘‘Birth outcomes’’ were primarily
with birth and maternal outcomes, we conducted an initial preterm birth and low birth weight, although the definitions
search by using the PubMed and Embase databases. The of these outcomes varied somewhat across studies. The re-
initial search included the terms ((pregnancy intention) sults are inconclusive. Two studies found that associations
OR (unplanned pregnancy) OR (unintended childbearing) between unintended pregnancy and birth outcomes varied
OR (unintended fertility) OR (unwanted pregnancy)). by race or ethnicity, with black women and Latinas having
Searches for studies on birth outcomes were identified by increased odds of negative outcomes if the pregnancy was
using the search terms (birth outcome OR neonatal outcome unintended versus intended (28, 29). A third study (30)
OR prematurity OR preterm birth OR low birth weight) OR found higher odds of low birth weight if the pregnancy
(maternal outcome OR maternal health OR maternal mor- was unwanted (vs. wanted; adjusted OR ¼ 1.16, 95% CI:
bidity OR maternal mortality). These searches were then 1.01, 1.33). These 3 studies assessed pregnancy intentions

combined. Abstracts of retrieved results were then reviewed retrospectively, and all used a similar question to do so
to identify relevant articles. We also reviewed bibliogra- (Table 2). Results on fertility intentions and birth outcomes
phies from selected articles to aid with complete review. for developing countries are more difficult to find. A World
Given previously published reviews of a related nature Health Organization report analyzed DHS data from 5 de-
(25, 26), we limited searches to articles published in English veloping countries (Bolivia, Egypt, Kenya, Peru, and the
in 2004 or later. We also limited our review to studies that Philippines). The authors concluded that the effects of un-
were prospective or longitudinal, were population based, intendedness on the child’s subsequent immunization status
and included multivariate analyses. and growth were inconsistent across countries (31).
To evaluate recent studies of the relation between pregnancy
intentions and contraceptive use, we conducted a second Pregnancy intentions and maternal behaviors and
search of the PubMed and Embase databases with the search health outcomes
terms (‘‘pregnancy, unplanned’’[MeSH Terms] OR (‘‘preg-
nancy’’[All Fields] AND ‘‘unplanned’’[All Fields]) OR ‘‘un- Research findings on fertility intentions’ effects on ma-
planned pregnancy’’[All Fields] OR (‘‘unintended’’[All ternal behaviors and health outcomes are even sparser than
Fields] AND ‘‘pregnancy’’[All Fields]) OR ‘‘unintended preg- for pregnancy outcomes. We identified 4 studies that met our
nancy’’[All Fields]) AND (‘‘contraception’’[MeSH Terms] search criteria (Table 2). The strength of our review is lim-
OR ‘‘contraception’’[All Fields]). Between 2004 and 2009, ited by the different maternal outcomes selected by each
256 English-language studies were published. Abstracts of group of authors, ranging from antepartum behaviors such
retrieved results were reviewed to identify eligible studies. as smoking, to pregnancy complications such as hyperten-
We again limited our review to studies that were prospective sion, to postpartum depression and breastfeeding. One study
or longitudinal, were population based, and included multivar- (30) found no association between pregnancy intention and
iate analyses. maternal outcomes, while another (32) found unintended-
ness associated with decreased odds of early prenatal care
(vs. intended: adjusted OR ¼ 0.54 for mistimed and OR ¼
RESULTS OF INDIVIDUAL-LEVEL STUDIES 0.34 for unwanted) and significantly increased odds of
postpartum depression (adjusted OR ¼ 1.34 and OR ¼
Pregnancy intentions and birth outcomes
1.98, respectively). Shapiro-Mendoza et al. (33) found that,
Rigorous research on the relation between pregnancy in- compared with intended pregnancies, unwanted and mis-
tentions and pregnancy outcomes is limited. What has been timed pregnancies were marginally, but not significantly
published generally focuses on short-term neonatal out- associated with longer durations of breastfeeding of infants
comes, such as prematurity. Many of the studies are older born in the 3 years prior to the 1990 Paraguay and 1994
and are methodologically limited. A recent systematic re- Bolivia DHSs (adjusted hazard ratio ¼ 0.90, 95% CI: 0.7,
view (27) concluded that unintended pregnancies, compared 1.2 and adjusted hazard ratio ¼ 0.87, 95% CI: 0.7, 1.0,
with intended pregnancies, were associated with higher respectively). What most of these studies have in common
odds of such neonatal outcomes as low birth weight (odds is their focus on immediate outcomes. Data on long-term
ratio (OR) ¼ 1.36, 95% CI: 1.25, 1.48) and preterm birth maternal or child outcomes are lacking.
(OR ¼ 1.31, 95% CI: 1.09, 1.58). However, because of
heterogeneity among studies—some studies adjusted for Pregnancy intentions and contraceptive use
potential confounding variables such as race, maternal
age, and prior low birth weight, while other studies did Five studies meeting our inclusion criteria were identified
not present adjusted results—the authors chose to incorpo- (Table 3), all of which were US based. Only one used a
rate only unadjusted odds ratios so as to include as many of prospective cohort design following up adolescent clinic
these studies as they could. Findings from previously pub- patients, whereas another analyzed survey data on recruits
lished studies are inconsistent, with some showing a negative at 2 US Army bases. The remaining 3 relied on data from
influence of intendedness on neonatal outcomes and others a nationally representative telephone survey or the 2002
showing no difference (25). Much of the available literature National Survey of Family Growth. Assessed contraceptive
is from developed countries. behaviors ranged from use at last sex, to consistent use, to

Table 2. Results From Key Studies of Fertility Intentions and Pregnancy and Maternal Outcomes
158 Tsui et al.

Author(s),
Fertility
Year Study Location/ Study Analytic Confounders Comparison Outcome(s) Multivariate Summary of
Intention Definition
(Reference Design/Year(s) Population Sample Size Controlled Group(s) Assessed Analysis Resultsa Findings
Measure
No.)
Pregnancy Outcomes
Afable- California Women with 15,331 ‘‘At the time ‘‘I wanted to get Family income, Intended Preterm birth White: unsure, OR ¼ Relation

Munsuz and Maternal and livebirths in women you got pregnant then’’ respondent’s vs. unsure, (<37 weeks of 1.44 (95% CI: 1.08, between
Braveman, Infant Health February– pregnant, how (intended), ‘‘I education, paternal mistimed, gestation) 1.92); unwanted, pregnancy
2008 (28) Assessment, May; California did you feel wanted to get education, age, unwanted (4 OR ¼ 1.31 (95% CI: intention and
pooled 5 resident, about getting pregnant later’’ parity, marital groups) 0.89, 1.91); mistimed, preterm birth
rounds of English or pregnant?’’ (mistimed), ‘‘I status, OR ¼ 1.08 (95% CI: varies by women’s
cross-sectional Spanish didn’t want to socioeconomic 0.83, 1.41). Immigrant racial or ethnic
survey data, speaking, get pregnant factors that Latina: unsure, group. After
1999–2003 aged 15 then or in the affected preterm OR ¼ 1.49 (95% CI: adjustment for the
years; future’’ birth 1.08, 2.06); unwanted, socioeconomic
addresses on (unwanted), and OR ¼ 1.24 (95% CI: and demographic
birth certificate ‘‘I wasn’t sure 0.87, 1.76), mistimed, variables, being
what I wanted’’ OR ¼ 1.03 (95% CI: unsure about
(unsure). 0.80, 1.34). No pregnancy
association among intention
black or US-born significantly
Latina women. elevated odds of
preterm birth
among immigrant
Latinas but not
among white,
black, or US-born
Latina women.
Keeton and CDC (Atlanta, US federal/ 47,956 ‘‘Thinking Unintended Maternal age, Intended vs. 1) Composite Relative risk for Intended
Hayward, Georgia) state women back to just pregnancies: ‘‘I marital status, unintended measure of intended pregnancy: pregnancy at
2007 (29) PRAMS data, cooperative with before you got didn’t want to be tobacco use, VLBW (<1,500 VLBW/VPT, black: a young age was
1993–2001 questionnaire singleton pregnant, how pregnant then or alcohol use, receipt g) and VPT birth OR ¼ 1.19 (95% CI: associated with
of women with birth from did you feel at any time in of prenatal care in (<32 weeks’ 1.02, 1.38); white: higher risk of poor
a recent 10 states about the future, or I the first trimester, completed OR ¼ 1.08 (95% CI: birth outcomes
livebirth drawn becoming wanted to be total number of gestational age) 0.92, 1.30). NICU (VLBW/VPT and
from each pregnant?’’ pregnant later.’’ prenatal care visits, and 2) admission admission, black: NICU admission)
state’s birth Intended medical to the NICU OR ¼ 0.93 (95% CI: for both whites and
certificate file pregnancies: ‘‘I complications of 0.81, 1.07); white: blacks. Intended
wanted to be pregnancy OR ¼ 1.08 (95% pregnancy was
pregnant (hypertension or CI: 0.96, 1.21). protective against
sooner, or I diabetes), and birth NICU admission
wanted to be history (a with advancing
pregnant then.’’ combination of maternal age.

parity and history of
prior preterm or low
birth weight births)
Mohllajee CDC (Atlanta, US federal/state 87,087 ‘‘Thinking Unintended: Maternal age, Intended vs. Birth outcomes: Ambivalence: low Women with
et al., 2007 Georgia) cooperative women back to just mistimed (‘‘I maternal race, ambivalent, low birth weight birth weight, OR ¼ unwanted
(30) PRAMS data, questionnaire who gave before you got wanted to be maternal ethnicity, mistimed, (<2,500 g), 1.15 (95% CI: 1.02, pregnancies had
1996–1999 of women with birth pregnant, how pregnant later’’), maternal unwanted preterm delivery 1.29). Mistimed: low an increased
a recent between did you feel unwanted (‘‘I education, marital (<37 weeks), birth weight, OR ¼ likelihood of
livebirth drawn 1996 and about didn’t want to be status, parity, and small for 0.92 (95% CI: 0.86, preterm delivery
from each 1999 in 18 becoming pregnant then or prenatal care, gestational age 0.97). Unwanted: and compared
state’s birth states pregnant?’’ at any time in previous low birth (birth weight preterm delivery, with women with
certificate file the future’’), or weight infant or <10th percentile OR ¼ 1.16 (95% CI: intended

ambivalent (‘‘I premature delivery, for the infant’s 1.01, 1.33). No other pregnancies.
don’t know’’). smoking during the gestational age associations found. Women who were
Intended: ‘‘I pregnancy, based on race ambivalent toward
wanted to be drinking during the and parity) their pregnancies
pregnant pregnancy, and had increased
sooner’’ or ‘‘I previous low birth odds of delivering
wanted to be weight or preterm a low birth weight
pregnant then.’’ birth infant infant; in contrast,
women with
mistimed
pregnancies had
a lower likelihood.
Maternal Outcomes
Hardee et al., Indonesia, Random and 796 Ever Subjective Background Ever vs. Low, medium, For being in low (vs. Women with a prior
2004 (94) cross-sectional quota sample experienced (respondent characteristics never had an and high high) psychological unintended
survey data, of women unintended report) associated in unintended measures of well-being: OR ¼ 2.8 pregnancy were
1996 aged 15–49 pregnancy bivariate analysis pregnancy psychological (95% CI: 1.5, 5.1; in more likely to
years with at with psychological well-being medium (vs. high) report lower
least 1 child well-being psychological well- measures of
being: OR ¼ 2.1 (95% psychological
CI: 1.2, 3.8). well-being.
Cheng et al., Maryland Random sample 9,048 ‘‘Thinking back Unintended: Sociodemographic Intended vs. Inadequate folic Folic acid use: Unintended
2009 (32) PRAMS of postpartum to just before mistimed (‘‘I factors mistimed and acid use, mistimed, OR ¼ 2.17 pregnancy had
database mothers you got wanted to be unwanted cigarette use, (95% CI: 1.78, 2.64); a statistically
(births in pregnant, how pregnant later’’), alcohol use, unwanted, OR ¼ 2.33 significant
2001–2006) did you feel unwanted (‘‘I postpartum (95% 1.71, 3.19). association with
about didn’t want to be depression, Early prenatal care: some negative
Family Planning and Unintended Pregnancies

becoming pregnant then or postpartum mistimed, OR ¼ 0.54 prenatal care
pregnant?’’ at any time in contraceptive (95% CI: 0.44, 0.67); behaviors and
the future’’). use, early unwanted, OR ¼ 0.34 with postpartum
Intended: ‘‘I prenatal care (95% CI: 0.26, 0.45). depression.
wanted to be Cigarette use
pregnant (prenatal): unwanted,
sooner’’ or ‘‘I OR ¼ 2.07 (95% CI:
wanted to be 1.47, 2.92);
pregnant then.’’ (postpartum):
unwanted, OR ¼ 1.86
(95% CI: 1.35, 2.55).
Breastfeeding for >8
weeks: unwanted,
OR ¼ 0.74 (95% CI:
0.57, 0.97). Postpartum
depression: mistimed,
OR ¼ 1.34 (95% CI:
1.08, 1.68); unwanted,
OR ¼ 1.98 (95% CI:
1.48, 2.64).
Table continues
159
160 Tsui et al.
Table 2. Continued
Author(s),
Fertility
Year Study Location/ Study Analytic Confounders Comparison Outcome(s) Multivariate Summary of
Intention Definition
(Reference Design/Year(s) Population Sample Size Controlled Group(s) Assessed Analysis Resultsa Findings
Measure
No.)
Mohllajee CDC (Atlanta, US federal/state 87,087 ‘‘Thinking back Unintended: Maternal age, Intended vs. Maternal No associations with All adjusted odds
et al., 2007 Georgia) cooperative women to just before mistimed (‘‘I maternal race, ambivalent, outcomes: pregnancy intention ratios for

(30) PRAMS data, questionnaire who gave you got wanted to be maternal ethnicity, mistimed, nausea, kidney found by authors. unintended,
1996–1999 of women with birth pregnant, how pregnant later’’), maternal unwanted or bladder Premature rupture of mistimed, and
a recent between did you feel unwanted (‘‘I education, marital infections, membranes: adjusted ambivalent were
livebirth drawn 1996 and about didn’t want to be status, parity, premature OR ¼ 1.37 linearly related and
from each 1999 in 18 becoming pregnant then or prenatal care, rupture of (unintended), adjusted greater than 1
state’s birth states pregnant?’’ at any time in previous low birth membranes, OR ¼ 1.03 (mistimed), (intended) but
certificate file the future’’), or weight infant or vaginal bleeding, adjusted OR ¼ 1.06 were not
ambivalent (‘‘I premature delivery, diabetes, high (ambivalent) statistically
don’t know’’). smoking during the blood pressure, compared with significant.
Intended: ‘‘I pregnancy, and and premature intended.
wanted to be drinking during the labor
pregnant pregnancy
sooner’’ or ‘‘I
wanted to be
pregnant then.’’
Shapiro- 1990 Paraguay Most recent 2,845 ‘‘At the time you Intended: ‘‘I Child’s sex, maternal Unwanted and Discontinued Unwanted vs. intended: Compared with
Mendoza and 1994 births to children < became wanted to get age, maternal mistimed vs. breastfeeding Bolivia, OR ¼ 0.87 intended
et al., 2007 Bolivia DHSs, women of 3 years of pregnant with pregnant then.’’ education, intended status at the time (95% CI: 0.7, 1.0); pregnancies,
(33) 3-year reproductive age in [name of last- Mistimed: ‘‘I mother’s marital of the survey (or Paraguay, OR ¼ 0.90 unwanted and
retrospective age in the 3 Bolivia born child], did wanted to get status, residence, 36 months of (95% CI: 0.7, 1.2). mistimed
survey data years prior to and 1,837 you want to pregnant later.’’ currently working, age) Mistimed vs. intended: pregnancies were
surveys children < become Unwanted: ‘‘I presence/type of Bolivia, OR ¼ 0.91 marginally
3 years of pregnant then, did not want any toilet facility, parity, (95% CI: 0.8, 1.1); associated with
age in did you want to more children.’’ currently pregnant, Paraguay: OR ¼ 0.91 longer duration of
Paraguay wait until later, modern (95% CI: 0.8, 1.1). breastfeeding in
or did you want contraceptive user both Bolivia and
no more Paraguay, but
children at associations were
all?’’ not statistically
significant; parity
did not modify the
association.
Abbreviations: CDC, Centers for Disease Control and Prevention; CI, confidence interval; NICU, neonatal intensive care unit; OR, odds ratio; PRAMS, Pregnancy Risk Assessment Monitoring System; VLBW, very
low birth weight; VPT, very preterm.
a
Unless otherwise indicated, adjusted odds ratios and 95% confidence intervals are presented.
resumed use following discontinuation, to type of method never used them. An extensive comparative analysis of 8
used. Fertility intentions were assessed in terms of wanting countries (39) assessed contraceptive failure rates for differ-
another pregnancy in the future, number of unintended preg- ent reversible contraceptive methods. Adjusting for poten-
nancies, and importance of avoiding a future pregnancy. tial confounders, the authors found the probability of an
A number of studies reported a high percentage of women accidental pregnancy among contraceptive pill users to
who did not intend to become pregnant but did not use range from 0.19 in Zimbabwe to 1.24 in Indonesia, from
contraception (5, 34, 35). On the other hand, adolescent 0.05 for injectable users in Bangladesh to 0.42 in Colombia,
clients committed to not becoming pregnant had higher odds and for IUD users from 0.04 in Egypt to 0.26 in Indonesia.
of using contraception 3 months later (adjusted OR ¼ 9.12, These results, of which most are statistically significant at
95% CI: 7.75, 12.30), and US Army recruits, irrespective of P < 0.01, are akin to typical-use failure rates found for
gender, not intending to have a baby in the next 6 months modern reversible contraceptives.
had higher odds of using an efficacious contraceptive Three US-state-based studies (41, 42, 44) similarly dem-
method (adjusted OR ¼ 1.14, 95% CI: 1.09, 1.20). Women onstrated that some contraceptive use or knowledge is

who had experienced 1 or more unintended pregnancies had advantageous in comparison to nonuse in reducing the inci-
notably higher odds (2.1 times) of using long-acting dence of unplanned pregnancy, but the measured protection is
methods compared with those with no such pregnancies not as high as one might expect. For example, in the Colorado
and reduced odds (OR ¼ 0.7) of using the pill or condom study, women having unprotected sex, compared with those
(36). Somewhat surprisingly, with 2002 National Survey of using birth control, had an adjusted odds ratio of 1.67 (95%
Family Growth data, Wu et al. (5) found no association CI: 1.11, 2.52) of having an unintended pregnancy within a 3-
between wanting to avoid a pregnancy in the future and year follow-up period. The magnitude of the adjusted odds
consistent use of contraception during months at risk in ratio value, although statistically significant, is not substantial
the past year. However, Vaughn et al. (37), with the same considering what should be a higher efficacy level from con-
data, found that the probability of resuming contraceptive traceptive protection.
use among women who achieved their desired family size Two studies on contraception and repeat abortion inci-
was significantly higher than for those discontinuing use dence (45, 46) showed that reliable contraception is signif-
when more children were wanted (adjusted hazard ratio ¼ icantly associated with reduced odds of repeat abortion. In
1.10, 95% CI: 1.04, 1.67). These empirical analyses suggest Goodman et al.’s study (45), 6.1% of postabortion IUD in-
that intentions to avoid pregnancy are associated with the sertion cases had repeat abortions compared with 15.3% of
use of contraception, but far from perfectly. controls; the adjusted hazard ratio was 0.37 (95% CI: 0.26,
0.52). Although only 60% of participants at risk of unin-
Contraceptive behaviors and fertility outcomes tended pregnancy in the St. Petersburg study (46) used re-
liable contraception at last sex, the odds of having 2 or more
Another set of studies examined the association of contra- abortions was significantly higher for those using unreliable
ceptive behaviors with subsequent pregnancy outcomes. Nine methods or no protection.
studies published since 2004 met our inclusion criteria, 7 of
which focused on the incidence of unintended pregnancy
(38–44) and 2 (45, 46) on repeat abortion as outcomes (Table RESULTS OF POPULATION-LEVEL STUDIES
4). Of the studies, 4 were US-state based (Rhode Island, Contraception benefits for reproductive health
Colorado, Oregon, and California), while the others were
conducted in international settings. Heterogeneity in study Demographers have long studied the relation between
design across studies limited our synthesizing the findings family planning and fertility at the population level and have
in a concise way. Contraceptive behaviors of interest range drawn implications about satisfying contraceptive need on
from general types of methods used, to specific use of fertility, abortion, and mortality rates in the developing
symptom-thermal or hormonal methods, to prepregnancy world. In this section, we review study findings from
knowledge of emergency contraception. cross-national survey analyses that project the impact of
The studies located in developing countries utilized contra- lowering unintended pregnancy on reproductive and child
ceptive calendar data and related respondent use patterns to the health outcomes. As background, we use the StatCompiler
subsequent incidence of unintended pregnancy. A Guatemalan tool (47) to compile DHS-based national estimates of the
study (38) examined the influence of women’s contraceptive proportion of pregnancies that are unintended, either mis-
discontinuation on unintended pregnancy, finding that those timed or unwanted, and the proportion of women of repro-
who used contraception but discontinued for a reason other ductive age using modern contraception. Figure 2 shows
than a desired pregnancy had a high relative risk ratio of 14.58 that the values from 158 DHSs conducted in 68 African,
(95% CI: 10.07, 21.12) of having an unwanted (vs. intended) Asian, Latin American, and Caribbean countries between
pregnancy. Those who did not use contraception also had 1991 and 2007 do not closely fit either a linear or curvilinear
a higher relative risk ratio of 3.94 (95% CI: 3.03, 5.10) of trend because levels of unintended pregnancy tend to
having a mistimed pregnancy and a relative risk ratio of be higher than use of effective contraception. In fact, the
6.17 (95% CI: 4.39, 8.67) of having an unwanted pregnancy. relatively flat, curvilinear, and positive trend line suggests
A Nigerian study (43) found that women who have ever that unintended pregnancy levels rise, rather than fall, with
used traditional or modern contraceptives have higher odds modern birth control use. Only a few country data points
of experiencing an unwanted pregnancy than those who show high use and low levels of unintendedness.

Table 3. Results From Key Studies of Fertility Intentions and Contraceptive Behaviors
162 Tsui et al.

Author(s),
Study Analytic Contraceptive Multivariate Summary
Year Study Fertility Intention Confounders Comparison
Location/Design/ Sample Definition Outcome(s) Analysis of
(Reference Population Measure Controlled Group(s)
Year(s) Size Assessed Resultsa Findings
No.)
Bartz et al., Moderately Adolescent 289 sexually Trying to get Yes or no Age, previous ‘‘No’’ answer Use of Trying to get Half of coital
2007 (34) large US city, females experienced pregnant, trying answers to, Are pregnancy; to intention contraception pregnant: OR ¼ events for

longitudinal, recruited females in 2004 not to get you trying to get random measure; at time of 0.17 (95% CI: adolescent females
2004 from 3 follow-up of the pregnant, pregnant now? intercept for unimportance coitus, measured 0.12, 0.23); trying committed to not
primary care original cohort of intensity of Are you trying to person- of commitment 3 months later not to get getting pregnant
adolescent 287 enrolled commitment, keep from getting specific and partner’s pregnant: OR ¼ were unprotected.
clinics subjects importance of pregnant now? correlations desire 7.84 (95% CI: Contraceptive
partner’s desire Importance of not 6.36, 9.68); protection during
that she get getting pregnant committed to not sex was
pregnant at this time in my getting pregnant, significantly
life; importance OR ¼ 9.12 (95% associated with
of partner CI: 6.75, 12.30); intention to avoid
wanting her to partner wants me getting pregnant.
get pregnant to get pregnant,
OR ¼ 0.46 (95%
CI: 0.35, 0.61)
O’Rourke United States, US Army 1,095 male and Pregnancy PRAMS-based Gender, age, Other levels of Contraceptive Pregnancy On the basis of
et al., 2008 cross-sectional, recruits female first-term intention multidimensional education, pregnancy efficacy nonintention score multivariate
(35) 2003–2004 at Fort soldiers, sexually assessment ethnicity, binge intention score (highest level (1–4); adjusted ordinal regression
Bliss, Texas, active but not codes: 1 ¼ plan drinking in the of birth control OR ¼ 1.14 (95% analysis of 845
and Fort pregnant or an a baby in the next past month used at last CI: 1.09, 1.20) with men and women,
Gordon, expectant father 6 months, 2 ¼ sexual each unit increase pregnancy/
Georgia ambivalent (don’t intercourse) in pregnancy paternity intention
know), 3 ¼ no intention score was found to be
intent in the next significantly
6 months, 4 ¼ no associated with
intent ever choice of
efficacious
contraceptive
method. One-third
of the population
attempting to
avoid pregnancy
used no birth
control at the last
sexual intercourse.
Frost and United States, Nationally 1,641 women 1) Number of 1) Number of Parity, fatalistic 1) 0 unintended Specific use of 1 unintended Women
Darroch, cross-sectional representative using reversible unintended unintended attitude about pregnancies, a reversible pregnancies(vs. experiencing 1
2008 (36) telephone sample of 1,978 contraception at pregnancies, 2) pregnancies ever contraception, 2) very method: pill, 0): pill ¼ 0.73, unintended
survey, 2004 women aged risk of importance of experienced, 2) reason for important long acting P < 0.05 (standard pregnancies had
18–44 years unintended avoiding how important it method use, (injectable, error not reported); higher odds (2.1) of
pregnancy pregnancy is to avoid would change patch, IUD, long acting ¼ 2.10, using long-acting
(heterosexually pregnancy (very, method if cost ring, and P < 0.001; methods and
active in the past somewhat, was not an implant), condom ¼ 0.66, reduced odds for
year, not a little/not issue, type of condom, P < 0.01. pill and condom use

pregnant, up to 2 important) provider and other Somewhat (vs. (0.7). Women
months (withdrawal, very) important: reporting little or no
postpartum, periodic pill ¼ 0.96, not importance about
trying to become abstinence, significant/long avoiding pregnancy
pregnant, and spermicides, acting ¼ 0.91, not had reduced odds
not sterile for and other significant/ of using the pill (0.4)
either barrier) condom ¼ 0.83, and elevated odds
contraceptive or not significant/ of using other
noncontraceptive other ¼ 2.01, P < methods such as
reasons) 0.001; a little/not rhythm or
(vs. very) withdrawal (2.6).
important: pill ¼
0.38, P < 0.001/
long acting ¼ 0.77,
not significant/
condom ¼ 1.14,
not significant/
other ¼ 4.42, P <
0.001
Wu et al., United States, Nationally 3,687 women Wants a baby Yes answer to, Age, race/ No (another Consistent Multinomial 10% of women
2008 (5) NSFG 2002, representative at risk of in the future Do you want ethnicity, baby not use (use of logistic at risk of an
retrospective sample of 7,643 unintended a (another) education, wanted) contraception adjusted relative unplanned
survey data on women aged pregnancy baby some parity, in the past risk ratio with pregnancy
contraceptive 15–44 years (heterosexually time in the insurance year during all consistent use as persistently did not
use active in the future? coverage, months at risk the reference use birth control
past year, not number of of pregnancy), outcome: nonuse ¼ over a 1-year
pregnant, sexual inconsistent use 1.1 (95% CI: 0.8, period. No
trying to become partners, (use of 1.7), inconsistent significant

pregnant, and months of contraception in use ¼ 1.2 association was
they or their sexual activity the past (95% CI: 0.88, 1.7) found between
partner not sterile year during future pregnancy
or do not suspect some months intentions and
being sterile when at risk contraceptive
of pregnancy), behavior.
nonuse (no
use of
contraception
during all
months of
pregnancy
risk)
Vaughn et al., United States, Nationally 7,106 episodes Wants no more Yes answer Nulliparous, Wants more Resumption of Cox proportional Women wanting
2008 (37) NSFG 2002, representative of use children married or children contraceptive hazards model no more children
retrospective sample of 7,643 cohabiting, use within 12 adjusted hazard have a higher
survey data on women aged previous months after ratio and 95% CI; probability of
contraceptive 15–44 years method used discontinuation wants no more resuming
use is hormonal, children ¼ 1.10 contraceptive use
previous (95% CI: 1.04, within 12 months of
method failed, 1.67) discontinuation
covered by (1.10) than those
private wanting more
insurance children.
163
Abbreviations: CI, confidence interval; IUD, intrauterine device; NSFG, National Survey of Family Growth; OR, odds ratio; PRAMS, Pregnancy Risk Assessment Monitoring System.
a
164 Tsui et al.
Reasons for this apparent contradiction may vary from
% Using Modern Contraception

70
country to country. It is worth recalling that the definition 60
of ‘‘contraceptive prevalence’’ does not presume correct
50
and consistent use. For example, reported prevalence of con-
doms may not account for the often sporadic nature of their 40
use. Similarly, high rates of contraceptive discontinuation 30
(also not captured in the contraceptive prevalence rate) could 20
counteract the potential impact of contraceptive use on un-
10
intended pregnancies. There may also be supply-side deter-
minants of contraceptive use, such as availability of method 0
0 10 20 30 40 50 60 70
choice or restricted access, that contribute to high unintended
% of Recent Pregnancies Unintended (Mistimed or Unwanted)
pregnancy rates. Women may be better able to articulate an
unintended pregnancy than they are to avail themselves of
Figure 2. Relation between national rates of unintended pregnancy

and practice the means to prevent one with contraception.
and modern contraceptive prevalence among women aged 15–49
years across 158 Demographic and Health Surveys in developing
Coitus, conception, infection, and contraception countries, 1991–2007 (47).
The scale of sexual activity, reproduction, and their po-

tential risks at the population level are challenging to visu-
alize. With a majority of the 1.74 billion reproductive-age of unintended pregnancy. Although abortion rates are similar
females being sexually active and a probability of concep- across developed and developing regions, many abortions
tion during unprotected coition of 3 in 100 (11), each year as that take place in the latter areas are unsafe (53, 54). Two-
many as 720 million conceptions may occur. The majority thirds of pregnancies in the developing world occur to women
of conceptions (60%–70%) will be spontaneously miscar- not using contraception (55). If contraception can reduce the
ried, leaving approximately 239 million identified pregnan- incidence of unintended pregnancies, it will lower the risk of
cies, of which 136.2 million will progress to livebirths, 33 death and disability due to unsafe abortions. In Uganda, for
million being unwanted. Another 46 million pregnancies example, current use of contraception, compared with no
will be electively terminated. contraceptive use at all, has resulted in 150,000 fewer abor-
The Guttmacher Institute (14) estimates a pregnancy rate tions (56). Meeting the existing level of 41% unmet need
of 137 per 1,000 women aged 15–44 years in the developing there would further reduce this number. In Guatemala, where
world and an unintended pregnancy rate of 57 per 1,000, or 32% of pregnancies are unintended and abortion is illegal,
82.3 million mistimed or unwanted pregnancies. The unin- 12% of pregnancies end in induced abortion. Eight of 1,000
tended pregnancy rate has declined since the mid-1990s, women of reproductive age were hospitalized in 2003 be-
largely because of increases in contraceptive use, propor- cause of complications of unsafe abortion, a figure that
tionately faster in the developed than the developing world. may well underestimate the magnitude of the problem. The
Preventing sexually transmitted infection and HIV trans- relatively low prevalence of modern contraceptives (43%)
mission, as well as unplanned pregnancies, during coition is and high unmet need (28%) are acknowledged to contribute
a priority in many low-income countries. HIV research has to abortions and related morbidity (57).
only recently acknowledged the importance of integrating Often, however, ineffective contraceptive use, rather than
family planning into HIV prevention and care programs. For nonuse, contributes to unintended pregnancy: for many east-
HIV-positive women who seek to postpone or delay a preg- ern European and south Asian countries, as many as two-
nancy, family planning is a proven and cost-effective thirds of abortions are due to contraceptive failure, mostly
method for preventing mother-to-child transmission of from traditional method use, and one-third are due to unmet
HIV (48). A study of 14 developing countries (49) reports need for contraception (58). In developed countries, it has
that, for 1.342 million HIV-positive women in need of peri- been reported that most abortions occur as a result of con-
natal HIV prevention, a potential 71,945 infant HIV infec- traceptive failure and a small proportion are due to nonuse
tions and 423,211 births can be averted through increased of contraception. One study found that 84% of women seek-
family planning use. As the client load in need of HIV ing abortion reported recent contraceptive use about
diagnosis and antiretroviral therapy increases, the financial the time they conceived, compared with 16% reporting
and service burden for programs to meet that demand nonuse (59).
elevates the importance of responding to clients interested Increased contraceptive uptake is generally associated
in spacing and limiting childbearing (50, 51). It is estimated with reduced numbers of abortions. Since 1995, abortion
that family planning can avert as many or more vertical rates have decreased worldwide. The greatest declines are
pediatric infections and HIV deaths as scaling up antiretro- in eastern Europe, concurrent with an increase in access to
viral efforts to prevent mother-to-child transmission (52). modern contraceptive methods (60). Westoff’s (58) analysis
of contraceptive use and abortion rates for 12 eastern Euro-
Contraception, conception, and parturition pean and south Asian countries shows a strong negative
correlation between prevalence of contraceptive methods
While safe, legal induced abortion has few health conse- and abortion rates. That is, with some exceptions, countries
quences for the woman, the need for abortion is an indication with the highest uptake of modern contraceptive methods

generally also have the lowest abortion rates. Westoff (58) maternal deaths would be 19% higher. A recent Guttmacher
estimates that, if unmet need in those countries were re- Institute study (14) found that fulfilling unmet contraceptive
duced to zero and traditional contraceptive methods were need can prevent an additional 150,000 maternal deaths an-
replaced by modern ones, the number of induced abortions nually; a study in rural Bangladesh (69) found that, between
would be lowered by 55%. Satisfying unmet need results in 1979 and 2005, the fertility decline was responsible for a 30%
an average 23% reduction in abortions. reduction in maternal deaths. In Uganda, even with substan-
tial unmet need for contraception, current use of contracep-
Contraception in relation to gestation and birth
tion has resulted in 490,000 fewer maternal deaths compared
with no contraceptive use (56). Similarly, Egypt’s maternal
intervals
mortality ratio was reduced 50% between 1992 and 2000,
Elevated risks of neonatal, infant, and child mortality and a development concurrent with increased uptake of family
of child malnutrition were statistically linked to short birth planning and other maternal health improvements (70). Sto-
intervals (less than 30 months) in an analysis of DHS data ver and Ross’ (71) recent analysis suggested that declines in

from 17 developing countries (61). The adjusted odds of total fertility rates between 1990 and 2005 in developing
neonatal, infant, and under-age-5-years child deaths were countries, attributable to contraceptive use, likely averted
1.67, 1.85, and 1.91 times significantly higher if the birth 1.2 million maternal deaths.
interval was 18–23 months compared with 36–47 months. Contraceptive use in particular may also disproportion-
Conde-Agudelo et al. (62) found, with Latin American data, ately impact women’s risk of maternal mortality at either
greater risk of preterm, low birth weight, and small-for- end of the reproductive age span. Adolescence and older
gestational-age infants associated with short interpregnancy reproductive age elevate maternal mortality risk, as does
intervals of less than 6 months compared with 18–23 parity greater than 4 births (31). Providing contraceptive
months. In terms of maternal health, 2 studies (63, 64) re- services to these groups can reduce the maternal mortality
ported higher risks of premature rupture of membranes, pre- ratio by as much as 58% (55).
eclampsia, high blood pressure, and anemia with For every maternal death, as many as 30 more women
interpregnancy intervals of less than 6 months in Latin may suffer disability or injury due to complications from
America (in comparison to 18–23–month intervals) and pregnancy, childbirth, or abortion (72). Levine et al. (73)
Bangladesh (in comparison to 27–50–month intervals). estimated that unwanted fertility and unsafe abortion ac-
The interval between pregnancies is an important window count for 12%–30% of maternal disability-adjusted life
during which contraceptive benefits for maternal health can years across the developing regions of the world. As many
be experienced. Although the empirical evidence on birth as 1.27 million years of life are lost and another 0.76 million
spacing and maternal and newborn outcomes is strong, that years of life with disability are due to this maternal burden
for contraceptive use between pregnancies is weak (65). of disease in sub-Saharan Africa alone. The global disease
Contraception’s benefits need to be empirically differenti- burden associated with unmet family planning need among
ated from those of lactation and other protective behaviors reproductive-aged women is one of the greatest contributors
in the birth interval. Since breastfeeding can extend over to disability-adjusted life years in the developing world,
many months, particularly in sub-Saharan African coun- accounting for 7.4 million disability-adjusted life years
tries, overlap with contraceptive use confounds estimates among women aged 15–44 years, according to 2006 esti-
of the latter’s effects. One multicountry analysis of preg- mates (74). This issue trumps other risk factors such as
nancy outcomes found that 12 months of contraception-only anemia (4.7 million disability-adjusted life years) and smok-
coverage in the preceding birth interval can reduce the mor- ing (1.6 million). Anemia itself is often due to pregnancy,
tality risk for the next newborn by 31.2%, while 12 months which suggests that the all-cause burden of unmet need is
of contraceptive use overlapping with breastfeeding reduces even higher.
the risk by 68.4% (66). This same study of DHS data for 19 Maternal disability is also due to complications of unsafe
developing countries found an average of 3–4 months of abortion and childbirth, such as prolonged or obstructed
contraceptive use overlapping with breastfeeding. labor resulting in vesicovaginal fistula. A recent prospective
study evaluated morbidity in Mombasa, Kenya, among
Family-planning-averted births and maternal morbidity women in the first year postpartum. The authors observed
and mortality a 50% incidence of anemia, an 11% incidence of HIV, and
39% with an unmet need for family planning (75). The same
Liu et al. (67) estimated that nearly 230 million births are authors previously found that postpartum morbidity among
averted annually by global contraceptive use, or 1.7 times HIV-positive women was higher in uninfected women (76).
the current number of livebirths. Averted births and preg- In Kenya, where 44% of births are unplanned (77) and con-
nancies reduce the size of the denominator of maternal and traceptive prevalence is 39%, it is logical to assert that those
infant mortality rates. Although it is difficult to attribute postpartum women who delivered children from an unin-
change in the maternal mortality ratio to a particular cause, tended pregnancy, due to unmet need for family planning,
evidence exists that meeting the need for family planning can suffered unnecessary and completely preventable disability.
reduce maternal mortality. An analysis of DHS data indicated A Mexican study evaluating the impact of family planning
a strong negative correlation between maternal mortality ra- on maternal morbidity (78) used historical data and generated
tios and contraceptive prevalence rates (49). Another study comparisons between the current standard of care there and
(68) estimated that, without contraception, the number of a model in which World Health Organization benchmarks for

Table 4. Results From Key Studies of Contraceptive Behaviors and Fertility Outcomes
166 Tsui et al.

Author(s),
Study Analytic Contraceptive Fertility Multivariate
Year Study Confounders Comparison Summary
Location/Design/ Sample Behavior Definition Outcome(s) Analysis
(Reference Population Controlled Group(s) of Findings
Year(s) Size Measure Assessed Resultsa
No.)
Barden- Guatemala, Pregnancies to 5,400 Contraceptive No method use Residence, age, Discontinued Whether Multinomial Discontinuation
O’Fallon retrospective reproductive- pregnancies to discontinuation any reported after at education, for wanted pregnancy logistic for reasons

et al., birth and aged women 4,118 women time in the 12 least 1 month religion, pregnancy; was wanted adjusted other than
2008 (38) contraceptive in the 3 years aged 15–49 months preceding (episode) of use indigenous discontinued at that time, relative risk pregnancy
use history prior to survey years livebirth ethnicity, for other wanted later ratios; carries a
covering 1999– household reason (mistimed), mistimed vs. higher and
2002 socioeconomic or not intended: statistically
status, parity, wanted at all nonuser ¼ significant
currently (unwanted) 3.94 (95% CI: relative risk
pregnant; 3.03, 5.10), of being
survey discontinued reported as
clustering for other mistimed or
reason ¼ 8.58 unwanted
(95% CI: than intended
6.55,11.25); pregnancy.
unwanted vs. Nonuse in the
intended: year prior to
nonuser ¼ pregnancy
6.17 (95% CI: carries a
4.39, 8.67), higher relative
discontinued risk of leading
for other to a mistimed
reason ¼ 9.78 or unwanted
(95% CI: than an
10.07, 21.12) intended
pregnancy.
Sedgh et al., Nigeria (8 3,020 women 2,978 women Ever used Used a modern Age, marital Never used Had an Traditional Women who
2006 (43) states), aged 15–49 contraception method; used a status, parity, unwanted only ¼ 2.62, ever used
cross-sectional years traditional method education, pregnancy P < 0.001 (no a modern or
household systematically only; never used socioeconomic standard error traditional
survey in 2002– sampled from status, religion, reported); method have
2003, weighted households in residence, modern ¼ higher odds
to population enumeration region 3.86, P < (3.9 and 2.6,
levels areas in 1 0.001 respectively)
urban and 1 of experiencing
rural local an unwanted
government pregnancy
area in each than those
state who never
used.
Frank- Germany, 1,599 women 900 women with Protected Abstinence in None Abstinent; Rates of Descriptive: The effectiveness
Hermann observational using the 17,638 cycles intercourse (with the fertile period, unprotected unintended rates and of using the
et al., 2007 prospective symptothermal motivated to barrier method) protected intercourse pregnancies 95% CIs; symptothermal
(40) cohort study method avoid intercourse in the per 100 protected ¼ method is
enrolled in pregnancy, fertile period, women 0.59% (95% comparable to
1985–2005 starting to use unprotected CI: 0.07, 2.13), abstinence
the intercourse in the abstinence ¼ practice and is
symptothermal fertile period 0.43% (95% superior to
method, and CI: 0.05, 1.55), unprotected

using only this unprotected ¼ intercourse.
or a barrier 7.46% (95%
method, aged CI: 4.15, 10.23)
19–46 years,
with an average
cycle length of
22–35 days, no
previous history
of infertility,
ovulating,
contributing at
least 12 cycles
of data
Kuroki et al., United States 542 clients 542 women, of Using a Reported use History of Not using Incident 0.69 (95% CI: Previous
2008 (42) (Rhode Island), attending 1,112 screened, hormonal of hormonal unplanned a hormonal unplanned 0.41, 1.14) unplanned
randomized primary care, consenting to contraceptive at contraception at pregnancy, age, contraceptive pregnancy pregnancy is
clinical trial gynecology, the trial (both baseline baseline race/ethnicity, significantly
of dual and family arms combined) education, associated
protection planning annual with a
intervention, clinics; aged household subsequent
1999–2003 14–35 years; income, parity, unplanned
English number of pregnancy.
speaking sexual partners Use of
in the past hormonal
month, coping contraception
score, at baseline
intervention arm lowers the
incidence of

unplanned
pregnancy
(adjusted
odds ¼ 0.69)
but is not
statistically
significant.
Table continues
167
168 Tsui et al.
Table 4. Continued
Author(s),
Study Analytic Contraceptive Fertility Multivariate
Year Study Confounders Comparison Summary
Location/Design/ Sample Behavior Definition Outcome(s) Analysis
(Reference Population Controlled Group(s) of Findings
Year(s) Size Measure Assessed Resultsa
No.)
Shlay et al., United States Medical records 642 women Did not use birth Whether any birth Age, race, Used birth Incident Did not use birth Nonuse of birth
2009 (44) (Denver, for 5,478 aged 12–44 control at the last control was used at poverty level, control at the unintended control ¼ 1.67 control at the

Colorado), women seen years provided sexual encounter; the last sexual number of last sexual pregnancy (95% CI: 1.11, last sexual
analysis of at the Denver with effective birth encounter; received previous encounter; 2.52); received encounter
medical records Metro Health contraception at control provided at a 3-month supply of pregnancies, not provided birth control at significantly
for STD clinic Clinic the initial STD the initial visit contraception free any history of with birth the initial visit increases the
clients, clinic visit, no of charge abortion, control at the ¼ 1.03 (95% odds of
2003–2006 intention to education, first first visit CI: 0.68, 1.58) unintended
become pregnancy pregnancy
pregnant, before age 17 (odds ¼ 1.67).
complete years, number Receiving birth
baseline and of sexual control at the
follow-up partners in the initial clinic visit
pregnancy past month, is not
history frequency of statistically
information, not sexual significantly
pregnant at the encounters over associated
initial visit, seen the past 4 with odds of
at least twice months incident
during the study pregnancy.
period
Goldsmith Oregon 2001 US federal/state 1,795 women (of Prepregnancy Before you got Maternal age, Yes, know of Unintended Don’t know; Oregon women
et al., PRAMS survey cooperative 2,490) who knowledge of pregnant . . . had maternal race/ emergency birth (either emergency unaware of
2008 (41) data, cross- questionnaire gave birth in emergency you ever read or ethnicity, contraception mistimed or contraception ¼ emergency
sectional of women with 2001 and contraception heard about maternal unwanted) 1.43 (95% CI: contraception
analysis a recent completed and emergency birth education, vs. intended 1.0, 2.1) before
livebirth drawn returned the control (the marital status, birth pregnancy
from each survey ‘‘morning-after’’ family income, have higher
state’s birth pill)? prepregnancy odds of having
certificate file insurance an unintended
coverage birth than
women
who know.
Bradley et al., Kenya 2003, Use episodes for Kenya ¼ 2,597, Type of method Traditional, pill, Age and Each modern Contraceptive Range for Compared with
2009 (39) Zimbabwe women aged Zimbabwe ¼ used injectable, parity at method vs. failure adjusted traditional
2005, Armenia 15–49 years 4,692, Armenia ¼ condom, IUD, discontinuation, traditional (accidental hazard ratios: method use,
2005, Egypt using 2,386, Egypt ¼ and other worked in the method pregnancy pill, 0.19 in in general, use
2005, reversible 15,025, modern method past year, years while using Zimbabwe to of modern
Bangladesh contraception Bangladesh ¼ of education, method) 1.24 in contraceptives,
2004, Indonesia in the 5 years 10,359, aware of Indonesia; particularly the
2002, Colombia before survey Indonesia ¼ contraception, injectable, IUD and
2005, 17,563, partner’s 0.05 in injectable

Dominican Colombia ¼ desired fertility, Bangladesh to methods, is
Republic 2002; 20,714, media 0.42 in significantly
multilevel Dominican exposure, Colombia; associated
hazard Republic ¼ community- condom, 0.22 with lower
regression 11,935 level in the odds of
analysis of contraceptive Dominican accidental
calendar data prevalence rate, Republic to pregnancy.
on residence, 0.94 in
contraceptive wealth status, Indonesia;
use and region, interval IUD, 0.04 in
pregnancy of use, and Egypt to 0.26
events cluster-level in Indonesia;
variance other: modern,
0.03 in
Bangladesh to
0.71 in
Colombia
Goodman Northern Aspiration 2,019: 673 cases Immediate Same-day Cases matched Controls who Rates of Hazard ratio ¼ 6.1% of
et al., California, case- abortion and 1,346 postabortal insertion postabortal on date of accepted repeat 0.37 (95% CI: postabortal
2008 (45) control, 2002– clients at 8 controls of IUD insertion of IUD abortion; non-IUD abortion per 0.26, 0.52), IUD insertion
2005 northern multivariate contraception; 1,000 P < 0.001 cases had
California hazards model women who woman- a repeat
Planned controlled for declined years in the abortion
Parenthood woman’s age, contraception first year compared with
clinics over a marital status, or received following 15.3% of
3-year period race/ethnicity, emergency the index controls;
family size contraception abortion repeat abortion
only excluded and in the rates ¼ 34.6

from controls entire (95% CI: 24.8,
follow-up 47.0) and 91.3
period (at (95% CI: 79.4,
the agency 104.9),
clinic) respectively,
per 1,000
woman-years
(P < 0.001).
Immediate
postabortion
IUD
contraception
significantly
reduces
repeat
abortion
incidence.
Table continues
169
170 Tsui et al.
care would be met (Mother-Baby Package). The study con-
abortions were
method. Odds
women at risk
of unintended
contraception
contraception
31% used an
of having 2
used reliable
of Findings
Had 2 or more Unreliable: 1.83 Nearly 60% of
intercourse;
significantly
Summary
cluded that increasing family planning prevalence from 59%
pregnancy
elevated if
was used.
at the last
unreliable
unreliable
to 74% among women older than age 20 years and from 18%
or no
to 33% among women younger than age 20 years (both
Mother-Baby Package goals) would avert 1,324 disability
events per 100,000 women annually—a 32% reduction com-
(95% CI: 1.28,
(95% CI: 1.23,

method: 1.71
Multivariate
pared with the current level of 4,149 disabling events.

Analysis
Resultsa
2.62); no
2.38)
Although high-quality published evidence is limited, the
conclusions are consistent at the population level. Optimi-
zation of family planning can prevent maternal disability.
Beyond physical disability, one study suggests that
Abbreviations: CI, confidence interval; IUD, intrauterine device; OR, odds ratio; PRAMS, Pregnancy Risk Assessment Monitoring System; STD, sexually transmitted disease.
Outcome(s)
Assessed
unintended pregnancy can adversely impact women’s qual-

abortions
Fertility
ity of life, with 94% of those surveyed saying they would

experience negative health effects. In the same study, 16%
of women also stated they would accept an ‘‘immediate risk
method used
of death’’ to avoid an unintended pregnancy (79).

Comparison
Group(s)
Reliable
Cost-effectiveness of family planning
International studies confirm that family planning is

intercourse, use
of contraception
among the most cost-effective of all health interventions

income, parity,
Confounders
employment,
Controlled
(80, 81). The cost savings stem from a reduction in unin-

intercourse
age at first
education,
Age, marital
personal
tended pregnancy, as well as a reduction in transmission of

status,
at first
sexually transmitted infections, including HIV. It has been

consistently documented that all contraceptive methods are
cost-effective in comparison to no method (82, 83). An
method did you use
injectable, implant,
at your last sexual
analysis of a publicly funded family planning program cal-

Reliable ¼ IUD,
reported use of
contraception,
Definition
patch, or ring)
withdrawal, or
culated that long-acting contraceptives (implants and IUDs),

contraceptive
douching (no
pill, condom;
intercourse?
unreliable ¼
spermicide,
emergency
What kind of
in particular, save US $7 in costs from unintended preg-

calendar,
nancy for every US $1 spent (84). A recent study examining

the cost effectiveness of contraception over 5 years in the
United States showed the copper-T IUD, the levonorgestrel-
containing IUD, and vasectomy to be the most cost-effective
Contraceptive
options (83). Although data show differences among indi-

method at last
Behavior
Measure
contraceptive
vidual developing countries, the measured savings are sub-

intercourse
1,147 participants Reliability of
stantial everywhere. One US dollar spent on family planning

can avert from US $2 (in Ethiopia) to US $9 (in Bolivia) in
health costs, with an average of US $8 annually for all
women using all methods of modern contraception (14,
2,501 contacted
18–44 years in (67% of 1,718
81). The previously cited cost-effectiveness models for

reachable by
phone of the
Analytic
Sample
Mexico calculate lifetime savings of US $10.5 million

Size
by mail)
with increased contraceptive prevalence (77). However, dis-

continuation of contraception, which often results from
dissatisfaction, negatively impacts cost-effectiveness. Thus,
health services
2 districts who
having many contraceptive choices available is likely to

consented to
participate in
reproductive
reproductive
Population
Women aged
an in-clinic
health and
increase overall cost-effectiveness (79, 84).

Study
survey of
DISCUSSION
Location/Design/
sectional data;
This review has focused on recent empirical studies of

Russia; cross-
St. Petersburg,
2003–2004
associations between pregnancy intentions and pregnancy

Year(s)
Study
and maternal outcomes and then examined the intermediate

role of contraception as a health intervention. In the path-
Table 4. Continued
ways of the behavioral epidemiology that link coital activ-

ity, conception, viable pregnancy, fetal growth, parturition,
Regushevskaya
(Reference
Author(s),
and the puerperium, protected sex is an important early

2009 (46)
Year
No.)
juncture for preventing unhealthy sequelae, such as sexually

et al.,
transmitted infection, unintended pregnancy, fetal wastage,

stillbirth, and maternal and neonatal morbidity and mortal-
a
ity. Modern contraceptive use and consistent condom use

are highly effective means of preventing unplanned preg- reduction in unintended pregnancies reduces the number of
nancies and sexually transmitted infections. events exposed to poor pregnancy outcomes and can make
We located and reviewed 21 eligible studies in the liter- planned events healthier. An integrated understanding of
ature between 2004 and 2009, a time frame not covered by family planning–attributable change in pregnancy events
recent reviews. All involved individual-level, multivariate and change in epidemiologic risk-associated ratios deserves
analyses; nearly all were observational; and 16 were US priority in future research efforts.
based. Three examined the association between pregnancy Understanding the effect of pregnancy intentions on con-
intentions and birth outcomes and 3 with maternal health traceptive and reproductive behaviors is also requisite for
behaviors and outcomes. Evidence of the effect of unin- strengthening the evidence base that informs maternal and
tended pregnancies was inconclusive. We examined 5 other child health policies and programs. Implicitly, it requires
studies, again all US based, of pregnancy intentions’ relation improving the measurement of fertility intentions both in
with contraceptive use, one that should have been straight- the United States and abroad. Santelli et al. (85) identified
forward and substantial but was not. The bivariate results affective, cognitive, and partner-specific dimensions as

showed a surprisingly high percentage of study participants promising directions for future improvements. The modest
not using contraception despite intentions to defer or limit empirical connections observed here among pregnancy in-
further childbearing. We next reviewed findings of 9 studies, tentions, contraceptive use, and health outcomes challenge
4 based in the United States, of contraceptive use and preg- the assumed reliability of unmet contraceptive need,
nancy incidence, particularly unplanned and electively ter- a widely used measure that depends on a woman’s reported
minated pregnancies. Here, we found more consistent desire to time future pregnancies. The gap between inten-
results, generally of the order observed for 1-year contra- tions and behavior, likely due to a combination of individual
ceptive efficacy under typical use conditions. preferences and contextual factors of service access and of
The limited number of rigorous studies, particularly out- cultural and personal relationships, demands appropriate re-
side the United States and beyond individual risk factors, search designs to assess the causal relevance of pregnancy
prompted us to look at studies adopting a population or intentions to reproductive behavior and more focused re-
demographic methods approach. Many of these analyses search on family planning’s health benefits when used be-
are cross-national, using standardized data and measures fore, after, and between pregnancies. The potential of
from the DHS. The identified studies often applied statisti- contraception-facilitated birth spacing for preventing pre-
cal models or forecasting methods with multiple country term and low birth weight infants to avert chronic disease
surveys to generate aggregate estimates of health benefits, in later adulthood is intriguing and warrants robust investi-
such as averted unplanned pregnancies, pregnancy termina- gation with cohort data.
tions, and maternal and infant deaths. In contrast to risk Demographic growth in the developing world will continue
ratios from individual-level studies, the population-level to exert upward pressure on the population base of women of
studies provide counts of contraception-averted events that reproductive age for several decades. Considerable momen-
affect the pregnancy denominators of maternal and infant tum is built into population age structures as a consequence of
morbidity and mortality rates. Reviewing the demographic past high fertility. At current rates, the number of unintended
evidence of the contraceptive use–attributable impact on the pregnancies will rise to 92 million globally by 2015. In many
burden of unintended pregnancies offers a complementary countries, where contraceptive prevalence is low and 40% to
perspective and a more comprehensive understanding of the 50% of the population is under age 15 years, the entry of these
underlying structure of behavioral linkages. cohorts into sexual activity and reproductive age will expose
By assessing study findings with population-level rates large numbers to the risk of unintended pregnancy. Access to
and ratios, we observed the scale and recurring probability quality contraceptive services will need to be expanded to
of enabling and disabling sexual and reproductive health avoid a rise in the volume of unplanned births and improve
practices and events. Global contraceptive use prevents preconceptional health. Perhaps even more critical is to mount
more than 200 million unintended births annually, which a substantial research effort that can address the implications
lowers rates of both unintended pregnancy and abortion. of elevated sexual and reproductive health risks among ado-
Some studies found that it significantly impacted maternal lescents and youth and the demographic import for future
and infant mortality rates as well. Other studies have mea- generations’ health and well-being.
sured contraception’s benefits in lowering the number of
vertical HIV transmission cases among infected mothers
interested in postponing or delaying future pregnancies.
More research is needed to differentiate contraception’s di- ACKNOWLEDGMENTS
rect effect in reducing the number of (unintended) pregnan-
cies from its indirect effect on the prevalence and incidence Author affiliations: Department of Population, Family
of unfavorable outcomes. Rising contraceptive practice in and Reproductive Health, Johns Hopkins Bloomberg School
a population can coincide with favorable shifts in the distri- of Public Health, Baltimore, Maryland (Amy O. Tsui); and
bution of pregnancy-related risks because of such common Department of Gynecology and Obstetrics, The Johns Hop-
influences as gains in female education or household in- kins University School of Medicine, Baltimore, Maryland
come, but prolonged contraceptive practice in the interpreg- (Raegan McDonald-Mosley, Anne E. Burke).
nancy interval can also confer health benefits of maternal This review was supported in part by funding from the
nutrition repletion on fetal growth and newborn survival. A Bill and Melinda Gates Foundation to the Bill and Melinda

172 Tsui et al.
Gates Institute for Population and Reproductive Health at 17. Klerman LV. The intendedness of pregnancy: a concept in
the Johns Hopkins Bloomberg School of Public Health. transition. Matern Child Health J. 2000;4(3):155–162.
Conflict of interest: none declared. 18. Bankole A, Westoff CF. The consistency and validity of re-
productive attitudes: evidence from Morocco. J Biosoc Sci.
1998;30(4):439–455.
19. Westoff C. New Estimates of Unmet Need and the Demand for
Family Planning. Calverton, MD: Macro International Inc;
REFERENCES 2006. (DHS comparative reports no. 14).
1. Ten great public health achievements—United States, 1900– 20. Westoff C, Bankole A. Unmet Need: 1990–1994. Calverton,
1999. MMWR Morb Mortal Wkly Rep. 1999;48(12):241–243. MD: Macro International Inc; 1995. (DHS Comparative
2. United Nations. Levels and Trends of Contraceptive Use as Studies no. 16).
Assessed in 1988. New York, NY: Department of International 21. Sonfield A. Working to eliminate the world’s unmet need for
Economic and Social Affairs, United Nations; 1989. (ST/ESA/ contraception. Guttmacher Policy Rev. 2006;9(1):10–13.
SER.A/110). 22. Trussell J. Contraceptive failure in the United States. Con-

3. Population Reference Bureau. Family Planning Worldwide traception. 2004;70(2):89–96.
2008 Data Sheet. Washington, DC: Population Reference 23. Seiber EE, Bertrand JT, Sullivan TM. Changes in contracep-
Bureau; 2008. tive method mix in developing countries. Int Fam Plan Per-
4. Robinson WC, Ross JA. The Global Family Planning Revo- spect. 2007;33(3):117–123.
lution: Three Decades of Population Policies and Programs. 24. Sullivan TM, Bertrand JT, Rice J, et al. Skewed contraceptive
Washington, DC: World Bank; 2007. method mix: why it happens, why it matters. J Biosoc Sci.
5. Wu J, Meldrum S, Dozier A, et al. Contraceptive nonuse 2006;38(4):501–521.
among US women at risk for unplanned pregnancy. Contra- 25. Gipson JD, Koenig MA, Hindin MJ. The effects of unintended
ception. 2008;78(4):284–289. pregnancy on infant, child, and parental health: a review of the
6. United Nations Population Fund. How the ICPD Programme literature. Stud Fam Plann. 2008;39(1):18–38.
of Action supports the MDGs. New York, NY: UNFPA 26. Cleland J, Bernstein S, Ezeh A, et al. Family planning: the
(www.unfpa.org/public/cache/offonce/icpd/pid/5072#mdg5). unfinished agenda. Lancet. 2006;368(9549):1810–1827.
(Accessed January 11, 2010). 27. Shah PS, Balkhair T, Ohlsson A, et al. Intention to become
7. United Nations. World population policies 2009. New York, pregnant and low birth weight and preterm birth: a systematic
NY: United Nations Department of Economic and Social review [published online December 10, 2009]. Matern Child
Affairs/Population Division, 2009. (http://www.un.org/esa/ Health J. (doi:10.1007/s10995-009-0546-2).
population/publications/wpp2009/Publication_highlights. 28. Afable-Munsuz A, Braveman P. Pregnancy intention and pre-
pdf). (Accessed January 13, 2010). term birth: differential associations among a diverse population
8. Financial resource flows for population activities in 2007. The of women. Perspect Sex Reprod Health. 2008;40(2):66–73.
Hague, the Netherlands: Netherlands Interdisciplinary De- 29. Keeton K, Hayward RA. Pregnancy intention and birth out-
mographic Institute; 2009. (http://www.resourceflows.org/ comes: does the relationship differ by age or race? J Womens
index.php/articles/c87/). (Accessed December 15, 2009). Health (Larchmt). 2007;16(4):510–516.
9. United Nations Population Fund. Financing the ICPD pro- 30. Mohllajee AP, Curtis KM, Morrow B, et al. Pregnancy inten-
gramme of action: data for 2006. New York, NY. tion and its relationship to birth and maternal outcomes. Obstet
(www.unfpa.org/public/cache/offonce/publications/pid/3935). Gynecol. 2007;109(3):678–686.
(Accessed December 15, 2009). 31. Marston C, Cleland J. The Effects of Contraception on Obstetric
10. Tietze C. Probability of pregnancy resulting from a single Outcomes. Geneva, Switzerland: Department of Reproductive
unprotected coitus. Fertil Steril. 1960;11:485–488. Health and Research, World Health Organization; 2004.
11. Wilcox AJ, Dunson DB, Weinberg CR, et al. Likelihood of 32. Cheng D, Schwarz EB, Douglas E, et al. Unintended pregnancy
conception with a single act of intercourse: providing bench- and associated maternal preconception, prenatal and postpartum
mark rates for assessment of post-coital contraceptives. Con- behaviors. Contraception. 2009;79(3):194–198.
traception. 2001;6(4):211–215. 33. Shapiro-Mendoza CK, Selwyn BJ, Smith DP, et al. The impact
12. Wawer MJ, Gray RH, Sewankambo NK, et al. Rates of HIV-1 of pregnancy intention on breastfeeding duration in Bolivia
transmission per coital act, by stage of HIV-1 infection, in and Paraguay. Stud Fam Plann. 2007;38(3):198–205.
Rakai, Uganda. J Infect Dis. 2005;191(9):1403–1409. 34. Bartz D, Shew M, Ofner S, et al. Pregnancy intentions and
13. King JC. The risk of maternal nutritional depletion and poor contraceptive behaviors among adolescent women: a coital
outcomes increases in early or closely spaced pregnancies. event level analysis. J Adolesc Health. 2007;41(3):271–276.
J Nutr. 2003;133(5 suppl 2):1732S–1736S. 35. O’Rourke K, Richman A, Roddy M, et al. Does pregnancy/
14. Singh S, Darroch JE, Ashford L, et al. Meeting the need for paternity intention predict contraception use? A study among
modern family planning services. In: Singh S, Darroch JE, US soldiers who have completed initial entry training. J Fam
Ashford L, et al, eds. Adding It Up: The Costs and Benefits of Plann Reprod Health Care. 2008;34(3):165–168.
Investing in Family Planning and Maternal and Newborn 36. Frost JJ, Darroch JE. Factors associated with contraceptive
Health. New York, NY: Guttmacher Institute and United choice and inconsistent method use, United States, 2004.
Nations Population Fund; 2009:16–20. Perspect Sex Reprod Health. 2008;40(2):94–104.
15. Santelli J, Rochat R, Hatfield-Timajchy K, et al. The mea- 37. Vaughan B, Trussell J, Kost K, et al. Discontinuation and re-
surement and meaning of unintended pregnancy. Perspect Sex sumption of contraceptive use: results from the 2002 National
Reprod Health. 2003;35(2):94–101. Survey of Family Growth. Contraception. 2008;78(4):
16. Koenig MA, Acharya R, Singh S, et al. Do current measure- 271–283.
ment approaches underestimate levels of unwanted child- 38. Barden-O’Fallon JL, Speizer IS, White JS. Association between
bearing? Evidence from rural India. Popul Stud (Camb). contraceptive discontinuation and pregnancy intentions in
2006;60(3):243–256. Guatemala. Rev Panam Salud Publica. 2008;23(6):410–417.

39. Bradley SEK, Schwandt HM, Khan S. Levels, Trends, and 58. Westoff C. Recent Trends in Abortion and Contraception in 12
Reasons for Contraceptive Discontinuation. Calverton, MD: Countries. Calverton, MD: Macro International Inc; 2005.
ICF Macro; 2009. (DHS analytical studies 20). (DHS analytic studies no. 8).
40. Frank-Herrmann P, Heil J, Gnoth C, et al. The effectiveness of 59. Schünmann C, Glasier A. Measuring pregnancy intention and its
a fertility awareness based method to avoid pregnancy in re- relationship with contraceptive use among women undergoing
lation to a couple’s sexual behaviour during the fertile time: therapeutic abortion. Contraception. 2006;73(5):520–524.
a prospective longitudinal study. Hum Reprod. 2007;22(5): 60. Sedgh G, Henshaw S, Singh S, et al. Induced abortion: esti-
1310–1319. mated rates and trends worldwide. Lancet. 2007;370(9595):
41. Goodman S, Hendlish SK, Reeves MF, et al. Impact of 1338–1345.
immediate postabortal insertion of intrauterine contraception 61. Rutstein SO. Effects of preceding birth intervals on neonatal,
on repeat abortion. Contraception. 2008;78(2):143–148. infant and under-five years mortality and nutritional status in
42. Kuroki LM, Allsworth JE, Redding CA, et al. Is a pre- developing countries: evidence from the demographic and health
vious unplanned pregnancy a risk factor for a subsequent surveys. Int J Gynaecol Obstet. 2005;89(suppl 1):S7–S24.
unplanned pregnancy? Am J Obstet Gynecol. 2008;199(5): 62. Conde-Agudelo A, Belizán JM, Norton MH, et al. Effect of the

517.e1–517.e7. (doi:10.1016/j.ajog.2008.03.049). interpregnancy interval on perinatal outcomes in Latin
43. Sedgh G, Bankole A, Oye-Adeniran B, et al. Unwanted America. Obstet Gynecol. 2005;106(2):359–366.
pregnancy and associated factors among Nigerian women. Int 63. Conde-Agudelo A, Belizán JM. Maternal morbidity and
Fam Plan Perspect. 2006;32(4):175–184. mortality associated with interpregnancy interval: cross sec-
44. Shlay JC, Zolot L, Bell D, et al. Association between provision tional study. BMJ. 2000;321(7271):1255–1259.
of initial family planning services and unintended pregnancy 64. Razzaque A, Da Vanzo J, Rahman M, et al. Pregnancy spacing
among women attending an STD clinic. J Womens Health and maternal morbidity in Matlab, Bangladesh. Int J Gynaecol
(Larchmt). 2009;18(10):1693–1699. Obstet. 2005;89(suppl 1):S41–S49.
45. Goldsmith KA, Kasehagen LJ, Rosenberg KD, et al. Unin- 65. Yeakey MP, Muntifering CJ, Ramachandran DV, et al. How
tended childbearing and knowledge of emergency contracep- contraceptive use affects birth intervals: results of a literature
tion in a population-based survey of postpartum women. review. Stud Fam Plann. 2009;40(3):205–214.
Matern Child Health J. 2008;12(3):332–341. 66. Tsui A, Creanga A. Does contraceptive use reduce
46. Regushevskaya E, Dubikaytis T, Nikula M, et al. Contraceptive neonatal and infant mortality? Findings from a multi-country
use and abortion among women of reproductive age in St. Pe- analysis. Paper presented at the 2009 Annual Meeting of the
tersburg, Russia. Perspect Sex Reprod Health. 2009;41(1):51–58. Population Association of America, Detroit, Michigan, April
47. MEASURE Demographic and Health Surveys. StatCompiler. 30–May 2.
Calverton, MD: Measure DHS, ICF Macro (http://www.stat- 67. Liu L, Becker S, Tsui A, et al. Three methods of estimating
compiler.com). (Accessed January 12, 2010). births averted nationally by contraception. Popul Stud (Camb).
48. Wilcher R, Cates W Jr, Gregson S. Family planning and HIV: 2008;62(2):191–210.
strange bedfellows no longer. AIDS. 2009;23(suppl 1):S1–S6. 68. Ross J, Blanc A. The contribution of family planning to re-
49. Halperin DT, Stover J, Reynolds HW. Benefits and costs of ducing maternal deaths. Presentation at the International
expanding access to family planning programs to women liv- Family Planning Conference: Research and Best Practices,
ing with HIV. AIDS. 2009;23(suppl 1):S123–S130. Kampala, Uganda, November 15–18, 2009.
50. Bradley H, Bedada A, Tsui A, et al. HIV and family planning 69. Chowdhury ME, Ahmed A, Kalim N, et al. Causes of maternal
service integration and voluntary HIV counselling and mortality decline in Matlab, Bangladesh. J Health Popul Nutr.
testing client composition in Ethiopia. AIDS Care. 2008;20(1): 2009;27(2):108–123.
61–71. 70. Khadr Z. Monitoring socioeconomic inequity in maternal
51. Hoffman IF, Martinson FE, Powers KA, et al. The year-long health indicators in Egypt: 1995–2005. Int J Equity Health.
effect of HIV-positive test results on pregnancy inten- 2009;8:38.
tions, contraceptive use, and pregnancy incidence among 71. Stover J, Ross J. How increased contraceptive use has reduced
Malawian women. J Acquir Immune Defic Syndr. 2008; maternal mortality [published online September 17, 2009].
47(4):477–483. Mater Child Health J. (doi:10.1007/s10995-009-0521-y).
52. Hladik W, Stover J, Esiru G, et al. The contribution of family 72. Sciarra JJ. Global issues in women’s health. Int J Gynaecol
planning towards the prevention of vertical HIV transmission Obstet. 2009;104(1):77–79.
in Uganda. PLoS One. 2009;4(11):e7691. (doi:10.1371/journal. 73. Levine R, Langer A, Birdsall N, et al. Contraception. In:
pone.0007691). Jamieson DT, Breman JG, Measham AR, et al, eds. Disease
53. Grimes DA, Benson J, Singh S, et al. Unsafe abortion: the Control Priorities in Developing Countries. 2nd ed. Wash-
preventable pandemic. Lancet. 2006;368(9550):1908–1919. ington, DC: World Bank; 2006:1075–1090.
54. World Health Organization. Unsafe Abortion: Global and 74. Ezzati M, Vander Hoom S, Lopez AD, et al. Comparative
Regional Estimates of the Incidence of Unsafe Abortion and quantification of mortality and burden of disease attributable
Associated Mortality in 2003. 5th ed. Geneva, Switzerland: to selected risk factors. Global Burden of Disease and Risk
WHO; 2007. Factors. Washington, DC: World Bank; 2006. (http://
55. Prata N, Sreenivas A, Vahidnia F, et al. Saving maternal lives www.ncbi.nlm.nih.gov/bookshelf/br.fcgi?book¼gbd&part¼
in resource-poor settings: facing reality. Health Policy. 2009; A592).
89(2):131–148. 75. Chersich MF, Luchters SM, Yard E, et al. Morbidity in the first
56. Vlassoff M, Sundaram A, Bankole A, et al. Benefits of meeting year postpartum among HIV-infected women in Kenya. Int J
the contraceptive needs of Ugandan women. Issues Brief (Alan Gynaecol Obstet. 2008;100(1):45–51.
Guttmacher Inst). 2009;4:1–8. 76. Chersich MF, Kley N, Luchters SM, et al. Maternal morbidity
57. Singh S, Prada E, Kestler E. Induced abortion and unintended in the first year after childbirth in Mombasa, Kenya: a needs
pregnancy in Guatemala. Int Fam Plan Perspect. 2006; assessment. BMC Pregnancy Childbirth. 2009;9:51.
32(3):136–145. (doi:10.1186/1471-2393-9-51).

174 Tsui et al.
77. Biddlecom A. Unsafe Abortion in Kenya. In brief, 2008 series, 86. Brown MS. Coitus, the proximate determinant of conception:
no. 4. New York, NY: Guttmacher Institute; 2008. inter-country variance in sub-Saharan Africa. J Biosoc Sci.
78. Hu D, Bertozzi SM, Gakidou E, et al. The costs, benefits, and 2000;32(2):145–159.
cost-effectiveness of interventions to reduce maternal mor- 87. Beck S, Wojdyla D, Say L, et al. The worldwide incidence
bidity and mortality in Mexico. PLoS One. 2007;2(1):e750. of preterm birth: a systematic review of maternal mortality
(doi:10.1371/journal.pone.0000750.g002). and morbidity. Bull World Health Organ. 2010;88(1):
79. Schwarz EB, Smith R, Steinauer J, et al. Measuring the effects 31–38.
of unintended pregnancy on women’s quality of life. Contra- 88. Ventura SJ, Abma JC, Mosher WD. Estimated pregnancy rates
ception. 2008;78(3):204–210. for the United States, 1990–2005: an update. Natl Vital Stat
80. Burkman RT, Sonnenberg FA. Health economics of contra- Rep. 2009;58(4):1–14.
ception. Obstet Gynecol Clin North Am. 2000;27(4): 89. Alan Guttmacher Institute. Sharing Responsibility: Women,
917–931viii. Society and Abortion Worldwide. New York, NY: Alan Gutt-
81. Smith R, Ashford L, Gribble J, et al. Family Planning Saves Lives. macher Institute; 1999.
4th ed. Washington, DC: Population Reference Bureau; 2009. 90. World Health Organization. Maternal Mortality in 2005: Es-
82. Trussell J, Leveque JA, Koenig JD, et al. The economic value timates Developed by WHO, UNICEF, UNFPA and the World

of contraception: a comparison of 15 methods. Am J Public Bank. Geneva, Switzerland: WHO; 2007.
Health. 1995;85(4):494–503. 91. Stanton C, Lawn JE, Rahman H, et al. Stillbirth rates: delivering
83. Trussell J, Lalla AM, Doan QV, et al. Cost effectiveness of estimates in 190 countries. Lancet. 2006;367(9521):1487–1494.
contraceptives in the United States. Contraception. 2009; 92. Population Reference Bureau. 2009 World population data
79(1):5–14. sheet. Washington, DC: Population Reference Bureau; 2009.
84. Foster DG, Rostovtseva DP, Brindis CD, et al. Cost savings (http://www.prb.org/pdf09/09wpds_eng.pdf). (Accessed April
from the provision of specific methods of contraception in 28, 2010).
a publicly funded program. Am J Public Health. 2009; 93. World Health Organization. Neonatal and Perinatal Mortality:
99(3):446–451. Country, Regional and Global Estimates. Geneva, Switzer-
85. Santelli JS, Lindberg LD, Orr MG, et al. Toward a multidi- land: WHO; 2006.
mensional measure of pregnancy intentions: evidence from 94. Hardee K, Eggleston E, Wong EL, et al. Unintended preg-
the United States. Stud Fam Plann. 2009; 40(2):40(2): nancy and women’s psychological well-being in Indonesia.
87–100. J Biosoc Sci. 2004;36(5):617–626.

MXQ 012

Uploaded by

Copyright:

Available Formats

You might also like

MXQ 012

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MXQ 012

Uploaded by

Copyright:

Available Formats

Epidemiologic Reviews Vol.

Family Planning and the Burden of Unintended Pregnancies

Amy O. Tsui*, Raegan McDonald-Mosley, and Anne E. Burke

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

152 Epidemiol Rev 2010;32:152–174

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

Induced Induced Pregnancies 100 41 in developed Worldwide a Contraception

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

unmarried women, and especially adolescents, may not be

Terminology and deﬁnitions in family planning

Modern contraceptive methods can be categorized in sev-

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

158 Tsui et al.

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

pregnant then.’’ combination of maternal age.

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Family Planning and Unintended Pregnancies

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

162 Tsui et al.

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Family Planning and Unintended Pregnancies

Reasons for this apparent contradiction may vary from

% Using Modern Contraception

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

The scale of sexual activity, reproduction, and their po-

Epidemiol Rev 2010;32:152–174

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Epidemiol Rev 2010;32:152–174

166 Tsui et al.

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Family Planning and Unintended Pregnancies

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

Family Planning and Unintended Pregnancies

care would be met (Mother-Baby Package). The study con-

Had 2 or more Unreliable: 1.83 Nearly 60% of

(95% CI: 1.23,

pared with the current level of 4,149 disabling events.

unintended pregnancy can adversely impact women’s qual-

ity of life, with 94% of those surveyed saying they would

Downloaded from https://academic.oup.com/epirev/article/32/1/152/503735 by guest on 10 March 2024

of death’’ to avoid an unintended pregnancy (79).

Cost-effectiveness of family planning

International studies confirm that family planning is

among the most cost-effective of all health interventions

(80, 81). The cost savings stem from a reduction in unin-