www.thelancet.com Vol 368 November 18, 2006
Sexual abstinence, contraception, and condom use by
young African women: a secondary analysis of survey data
John Cleland, Mohamed M Ali
Background Drug therapy for people with AIDS is a humanitarian priority but prevention of HIV infection remains
essential. Focusing on young single African women, we aimed to assess trends in a set of behaviours—sexual
abstinence, contraceptive use, and condom use—that are known to aff ect the rates of HIV transmission.
Methods We did a secondary analysis of public-access data sets in 18 African countries (132 800 women), and
calculated changes in a set of behavioural indicators over time. We standardised these trends from nationally
representative surveys to adjust for within-country changes in age, education, and type of residential location.
Findings Between about 1993 and 2001, the percentage of women reporting no sexual experience changed little.
During the same period, the percentage of sexually experienced women who reported no sexual intercourse in the
previous 3 months (secondary abstinence) rose signifi cantly in seven of 18 countries and the median for all
18 countries increased from 43·8% to 49·2%. Use of condoms for pregnancy prevention rose signifi cantly in 13 of
18 countries and the median proportion increased from 5·3% to 18·8%. The median rate of annual increase of
condom use was 1·41 percentage points (95% CI 1·12–2·25). In the 13 countries with available data, condom use
at most recent coitus rose from a median of 19·3% to 28·4%. Over half (58·5%) of condom users were motivated,
at least in part, by a wish to avoid pregnancy.
Interpretation Condom promotion campaigns in sub-Saharan Africa have aff ected the behaviour of young single
women; the pace of change has matched the rise in contraceptive use by married couples in developing countries
over recent decades. Thus continuing eff orts to promote condom use with emphasis on pregnancy prevention are
An aura of disappointment and frustration surrounds
global eff orts to check the spread of HIV in most
low-income and middle-income countries. Despite
substantial investment in the promotion of safe sex
and marketing of condoms, only Thailand and Uganda
have clearly succeeded in stemming epidemics in the
general population, though other countries in Africa
and the Caribbean also show some signs of progress.1
Even the most intensive interventions, targeted at
young people or at specifi c localities, have not achieved
all of their aims.2–4 Two responses to this perceived
impasse have pre dominated. The fi rst is to conclude
that the necessary changes in sexual behaviour will
only occur as a result of fundamental changes that
reduce poverty and gender inequality.5–8 The other
response is to hope that increased access to drug
therapy for AIDS will present new opportunities for
eff ective prevention.9,10
Abstinence and condom use are two of the three
elements promoted by major preventive programmes
that emphasise the ABC (abstain, be faithful, and use
condoms) approach. The relative importance of
abstinence and condom use for HIV control has been
controversial. In the USA the issue is further
complicated by moral arguments surrounding the
President’s Emergency Plan for AIDS Relief (PEPFAR),
which emphasises abstinence.11–14 We do not seek to en-
gage with this debate; rather, we examine modifi cation
of behaviour in view of the threat of AIDS and the
inseparable risk of unwanted pregnancies.
We aimed to assess the evidence for behavioural
change by young women in sub-Saharan Africa, which
is the region most aff ected by HIV/AIDS. Our report
complements detailed analyses of change in specifi c
countries or small groups of countries, especially those
of the MEASURE (Monitoring and Evaluation to Assess
and Use Results) project,15 by examining changes in a
few behavioural indicators in a large sample of African
countries. Specifi cally, we analysed trends in sexual
abstinence, contraception, and condom use by single
women aged 15–24 years, in countries that had done
two or more comparable surveys between 1990 and
We analysed data from Demographic and Health
Surveys (DHS), which use nationally representative
samples, and standard instruments and procedures for
collection and processing of data. The survey design of
DHS is generally stratifi ed, with a fi rst-stage selection
of geographical clusters, followed by random selection
of households within each cluster to identify women
aged 15–49 years. These women, the primary
respondents, were interviewed by trained female staff
using a structured instrument that had been translated
Lancet 2006; 368: 1788–93
See Comment page 1749
Centre for Population Studies,
London School of Hygiene and
Tropical Medicine, London, UK
(Prof J Cleland MA); Department
of Reproductive Health and
Research, WHO, Geneva,
Switzerland (M M Ali PhD)
Prof John Cleland,
Centre for Population Studies,
London School of Hygiene and
49–51 Bedford Square,
London WC1B 3DP, UK
www.thelancet.com Vol 368 November 18, 2006 1789
into local languages. Response rates were typically 90%
We examined individual-record data fi les from DHS
surveys for single women (ie, never married) aged
15–24 years. Single women comprised 49% of all those
in this age band. With these restrictions on age and
marital status, the sample size of each national survey
ranged from 656 to 3220 women. We focused on the
following behavioural indicators: primary abstinence
(or virginity); secondary or temporary abstinence in
sexually experienced women (defi ned as no intercourse
in the 3 months preceding the survey); and use of
specifi c contraceptive methods, including condoms, in
women who had been sexually active in the 3 months
preceding the survey. Secondary abstinence was
included as an indicator of more cautious selection of
sexual partners, reduced coital frequency, or both. We
also analysed information on condom use at most
recent coitus and the source of any condoms used.
In 18 countries in sub-Saharan Africa, we had two or
more surveys from which to collate the relevant
information, and in seven of these countries there were
three surveys. The median dates of the earliest and
most recent sampling rounds were February, 1993, and
January, 2001. During this period these 18 countries
accounted for about 56% of the population of
sub-Saharan Africa. Data on condom use at most recent
coitus were available for 13 of the 18 countries. For
these 13 countries, the median dates of the earliest and
most recent survey rounds were October, 1996, and
February, 2002. The list of countries and country-
specifi c results is shown in webtables 1, 2, and 3.
To take account of changes in the composition of
successive samples from each country that might distort
trends, estimates from the more recent surveys for each
country were standardised by the composition of the
earliest survey with respect to age (in 3 year bands),
education (whether up to primary school level or
higher), and urban or rural location. All assessments
were based on standardised estimates. The median date
of fi eldwork was calculated for each survey and used to
estimate the rate of annual change in the prevalence of
condom use. All rates are expressed as absolute
percentage points. Exact 95% CIs for the median values
were calculated from the binomial distribution. We
assessed the signifi cance of within-country changes
using logistic regression by including a binary indicator
for the survey round (0=earliest survey and 1=most
recent survey). We also took the cluster design of
surveys into account. We applied survey-normalised
weights to take account of variations in probabilities of
selection and non-response. All analyses were done
with Stata statistical software (version 8).
Role of the funding source
The sponsor of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. The corresponding author had
full access to all the data in the study and had fi nal
responsibility for the decision to submit for
Figure 1 shows the standardised trend in the percentage
of young single women who reported no experience of
sexual intercourse. The median value of all survey
estimates remained essentially unchanged, at about
60%, though the dispersion of values contracted over
time. The proportion of young women declaring
themselves to be virgins rose signifi cantly in seven of
18 countries and fell signifi cantly in six countries. The
changes in proportion exceeded 10% only in Cameroon
and Ghana (webtable 1).
The median value for abstinence over the 3 months
preceding the survey in sexually experienced single
women increased from 43·8% to 49·2% (fi gure 2). An
Percentage with no sexual experience
Figure 1: Box and whisker plot showing trend (1993–2001) in the percentage
of single women aged 15–24 years who reported no sexual experience
Data standardised for 18 African countries.
Percentage with no sexual experience
Figure 2: Box and whisker plot showing trend (1993–2001) in the percentage
of sexually experienced single women aged 15–24 years who reported no
sex in the 3 months before the survey
Data standardised for 18 African countries.
See Online for
webtables 1, 2, and 3
www.thelancet.com Vol 368 November 18, 2006
increase in temporary or secondary abstinence was
signifi cant in seven of the 18 countries and exceeded 10%
in fi ve of these countries (webtable 1). A decrease was
signifi cant in Burkina Faso (p=0·003). In most of the
seven countries with three surveys, trends were erratic.
For women who reported coitus in the 3 months
preceding the DHS surveys (n=9100), use of any
contraceptive method rose slightly during the study
period from a median of 32·6% to 36·5% (fi gure 3). Use
of highly eff ective non-barrier methods (predominantly
oral contraceptives) changed little, whereas use of less
eff ective methods (mainly periodic abstinence) fell
slightly. However, use of condoms rose substantially
during the study period, from a median of 5·3% to
18·8%. This rise in condom use was signifi cant in 13 of
18 countries and exceeded 10% in nine countries
(webtable 2). Condom use was not related to the severity
of national HIV epidemics—some of the largest increases
in condom use were recorded in west African countries
where HIV prevalence is lower than in eastern or
The annual rate of increase in use of condoms for
pregnancy prevention between the earliest and most
recent survey was 2% or more in seven countries. The
median rate of annual increase for all 18 countries was
1·41% (95% CI 1·12–2·25). Trends for condom use were
monotonic in all countries that had three surveys during
the study period (webtable 2).
The median rate of condom use at most recent coitus
also rose, from 19·3% to 28·4% (fi gure 4). In seven of the
13 countries for which these data were available, the rise
in condom use was signifi cant and in six of the seven
countries it exceeded 10%. The median rate of annual
increase was 2·10% (95% CI 0·77–3·06). The median
rate of annual increase in condom use for contraception,
recalculated for the same 13 pairs of surveys, was 1·99%
(95% CI 1·02–2·38).
Information about contraceptive use and condom
use at most recent coitus was ascertained at diff erent
times during the DHS interviews, and the question on
condom use at most recent coitus made no mention of
the motive for their use. Therefore, we examined the
statements of women who used a condom at most
recent coitus to gain insights into dual protection and
the probable importance of contraception as a motive
for condom use. In the most recent survey round of
the 13 countries, between 34% and 79% of women who
had used a condom at most recent coitus had stated
earlier in the interview that they were using condoms
for pregnancy prevention; the median was 58·8%
(webtable 3). In the same survey round, the median
proportion of women who reported use of an eff ective
contraceptive method was 5·8%, the proportion who
used a less eff ective non-barrier method was 2·2%,
and 24·5% reported that they did not use
Women who reported using condoms for pregnancy
prevention were asked about the source of their supply.
In seven of the 18 countries surveyed in the most
recent round, 90% or more of women reported that
pharmacies and shops were the main sources of supply
and in a further eight countries this proportion was
Self-reported sexual behaviour is commonly regarded
with scepticism, which is partly justifi ed by the risk that
survey respondents censor their answers, especially as
public-health messages intensify. We cannot validate
the trends we have identifi ed in reported behaviour,
because there is no evidence on the incidence of sexually
transmitted infections that is disaggregated by marital
status, and because of the diffi culty of comparing
condom sales with reported use.17 Thus an extreme
20011993 20011993 20011993 2001
Percentage currently using contraception
Figure 3: Box and whisker plot showing trend (1993–2001) in current
contraceptive use by single women aged 15–24 who were sexually active in
the last 3 months before the survey
Data standardised for 18 African countries.
Percentage condom use at most recent coitus
Figure 4: Box and whisker plot showing trend in percentage of single women
aged 15–24 who reported condom use at most recent coitus and who were
sexually active in the last 3 months
Data standardised for 13 African countries.
www.thelancet.com Vol 368 November 18, 2006 1791
interpretation of our fi ndings is that they indicate no
more than a growing tendency for young women in
Africa to give socially desirable responses.
We accept that willingness to disclose premarital sex
or use of condoms probably varies between societies
and can be aff ected by survey design, the sequence and
wording of questions, and the mode of data capture.18,19
Because of this, comparisons between countries and
of results from diff erent types of inquiry can be
hazardous. We circumvented these pitfalls by focusing
on trends within countries and by using information
from the DHS rounds, which are regarded as the gold
standard for collection of quantitative data from
large-scale interview-based surveys in developing
countries. We cannot claim that this study is without
fl aws but we are cautiously confi dent that the direction
and approximate magnitude of trends in reported
behaviour are valid.
A more complete picture of changes in behaviour of
young single African women in response to the threat
of HIV infection would have included evidence on
multiple sexual partners. Some commentators have
argued that the B (Be faithful) element of ABC health
messages, involving reduction in sexual contact with
multiple partners, is crucial to stemming HIV
epidemics.20 We decided against inclusion of this factor
for two reasons. First, there is convincing evidence that
women under-report multiple partners because of social
disapproval.18 Second, the DHS programme changed
the relevant sequence of questions in the late 1990s,
jeopardising comparability between results from early
and more recent survey rounds.
We chose our study population, of single women aged
15–24 years, because their behaviour, together with that
of young men, will largely shape epidemic pathways in
the next decade. In the 18 countries we studied these
women constitute a sizeable fraction (about 20%) of all
women aged 15–49 years. Despite the fact that in mature
HIV epidemics about 50% of HIV infections occur
within marriage, sexually active single people remain at
higher risk than those who are married.21,22 And, fi nally,
patterns of sexual behaviour established at a young age
probably persist into adulthood.23,24
We noted a substantial rise in the use of condoms
reported by young, sexually active, single women in
sub-Saharan Africa. Regrettably, the DHS data did not
contain information that would allow us to estimate the
consistency of condom use; however, condom use at
most recent coitus is highly correlated with consistent
Some will argue than an annual increase in condom
use of 1·4% per year—the median for all 18 countries—
is far too slow a pace of change to justify any claim that
condom promotion campaigns have been eff ective in
aff ecting the behaviour of young women in Africa.
What pace of change might reasonably have been
expected? One appropriate yardstick might be the rise
in contraceptive practice by married couples in
developing countries. Between 1965 and 1998 the
prevalence of contraceptive use in developing countries
rose from about 10% to nearly 60%, corresponding to
an average annual increase of 1·5%.27 Like condom use,
adoption of contraception
adjustments in attitudes towards procreation and sex.
This example suggests that a 1·4% annual increase in
the reported use of condoms by young single women in
18 African countries can be regarded as at least a
Of course, in view of the severity of AIDS in eastern
and southern Africa, a faster pace of behavioural
modifi cation would be highly desirable. However, to
expect abrupt, major changes in behaviour was probably
unrealistic, especially since warnings of the threat
posed by AIDS and methods for prevention and
treatment of the disease came largely from more
developed countries and were initially received with
denial and suspicion in much of Africa. Behavioural
change on a large scale tends to take time, since it needs
to be preceded by a period in which unfamiliar messages
become assimilated into local social networks.28
Although not all premarital conceptions are
unwelcome in African communities,29 in nearly all
African DHS rounds, fewer than 10% of sexually active,
single women stated that they would like to have a child
in the next 12 months. Thus the need for contraception
was high, as was the likelihood of unwanted pregnancy
if contraception was not used. Induced abortions are
illegal in most African countries, but are nevertheless
thought to be common. The proportion of all unsafe
abortions that occur in young women and the number
of abortion-related deaths are both higher in Africa than
in other regions.30,31 For example, in Ghana, unsafe
abortion probably kills more women than AIDS does.32
Thus, the fear of unwanted pregnancy might be as great
as the fear of AIDS for many young women in western
and central Africa.
Against this background, mixed messages can be
drawn from our fi ndings. As shown in fi gure 3, the rise
in overall contraceptive use has been negligible. The
shift away from periodic abstinence is welcome because
of the high failure rates of this method caused by
widespread misunderstanding of the safe period.33 The
absence of evidence that use of oral contraceptives or
other modern non-barrier methods has increased is
perhaps surprising but might indicate diffi culties with
access or health concerns, coupled with anxiety that
future childbearing might be jeopardised by such
methods.34 The fi nding that condoms have now become
the dominant method of contraception is to be
welcomed, of course, because of their dual protection
against HIV transmission and pregnancy. In our view,
this advantage is more important than the condom’s
higher rate of failure in preventing pregnancy than
other modern methods. The same shift in choice of
www.thelancet.com Vol 368 November 18, 2006
contraceptive method has been documented in young
single women in Latin America.35
Our data on the source of condom supplies suggest
that the rise in their use might be partly due to greater
familiarity with the method, and to social marketing
campaigns that have improved the availability of
condoms through commercial outlets. The growing
popularity of condom use by single women could also
indicate in part a low frequency of intercourse, as
suggested by the fi nding that about half of sexually
experienced women reported no coitus in the 3 months
preceding the survey. When intercourse is infrequent,
use of a method that provides continuous protection,
such as oral contra ception, might seem unnecessary—
The relative importance of the two motives for
condom use—pregnancy prevention and disease
prevention—is very diffi cult to establish. The uptake of
condoms did not seem to be related to the severity of
national HIV epidemics: some of the sharpest increases
in usage were recorded in west African countries where
HIV prevalence is lower than in eastern or southern
Africa. Our fi ndings suggest that at least 60% of single
women who used a condom at most recent coitus did
so mainly, or partly, to avoid pregnancy. Though this
proportion should be interpreted with great caution, it
emphasises that programmes promoting family
planning and HIV prevention have common interests.
A young woman might fi nd it easier to negotiate use of
condoms with a partner for prevention of pregnancy
than for protection against HIV transmission. There-
fore, we suggest that condoms might be promoted
more eff ectively if the emphasis was on pregnancy
prevention rather than prevention of sexually trans-
mitted disease. In our view, moves to dissociate the aim
of HIV prevention from that of contraception are
regrettable—for example, the creation of the Global
Fund to fi ght AIDS, Tuberculosis and Malaria and the
prohibition on buying family planning commodities
from PEPFAR funds.
Other behaviours, such as reported abstinence, have
changed less than use of condoms. The median
proportion of reported virgins changed little over the
study period. Secondary abstinence did increase in
some countries, which could indicate more cautious
selection of sexual partners, or reduced coital frequency,
or both. A change in selection of sexual partners could
perhaps be attributed to HIV control programmes.
Most of the survey data reviewed here predate new
HIV prevention programmes such as PEPFAR, that
emphasise sexual abstinence, and thus any judgments
about the eff ectiveness of such strategies would be
premature. The very wide national variations in the
prevalence of virginity in young women imply that
societal attitudes towards premarital sex vary greatly
between African countries and therefore that advocacy
for abstinence might resonate in some ethnic and
religious groups but prove counter-productive in
Our central conclusion is that the sense of failure
pervading HIV prevention eff orts in Africa is unjustifi ed
and that investments in condom promotion and
marketing have had an appreciable eff ect, at least for
young single women. We hope that this paper will
contribute to a more balanced perspective on progress
towards control of HIV infection in Africa, and will
underline the need for greater attention to the issue of
J Cleland and M M Ali jointly planned this paper. M M Ali was
primarily responsible for the statistical analysis and J Cleland for
drafting and interpretation.
Confl ict of interest statement
We declare that we have no confl ict of interest.
Helpful comments on earlier drafts of this paper were made by
Marge Berer, Iqbal H Shah, James Shelton, and Paul Van Look.
1 Joint United Nations Programme on HIV/AIDS and World Health
Organization. AIDS Epidemic Update. Geneva: UNAIDS/WHO,
2 Jemmott JB, Jemmott LS. HIV risk reduction behavioral
interventions with heterosexual adolescents. AIDS 2000;
14 (suppl 2): 40–52.
3 Oakley A, Fullerton D, Holland J. Behavioural interventions for
HIV/AIDS prevention. AIDS 1995; 9: 479–86.
4 Stephenson JM, Imrie J, Sutton SR. Rigorous trials of sexual
behaviour interventions in STD/HIV prevention: what can we
learn from them? AIDS 2000; 14 (suppl 3): 115–24.
5 Dunkle K, Jewkes, R, Brown H, Gray G, McIntryre J, Harlow S.
Gender-based violence, relationship power, and the risk of HIV
infection in women attending antenatal clinics in South Africa.
Lancet 2004; 363: 1415–20.
6 Jewkes R, Levin JB, Penn-Kekera LA. Gender inequalities,
intimate partner violence and HIV prevention practices: fi ndings
of a South Africa cross-sectional study. Soc Sci Med 2003; 56:
7 Kim JC, Watts CH. Gaining a foothold: tackling poverty, gender
inequality, and HIV in Africa. BMJ 2005; 331: 769–72.
8 Parker RG, Easton D, Klein CH. Structural barriers and
facilitators in HIV prevention: a review of international research.
AIDS 2000; 14 (suppl 1): 22–32.
9 Global HIV Prevention Working Group. HIV prevention in the
era of expanded treatment access. Seattle and Menlo Park: Kaiser
Family Foundation, 2004
10 UNAIDS. Intensifying HIV prevention. Geneva: UNAIDS, 2005.
11 Bass E. The two sides of PEPFAR in Uganda. Lancet 2005; 365:
12 Blum RW. Uganda AIDS prevention: A, B, C, and politics.
J Adolesc Health 2004; 34: 428–32.
13 Das P. Is abstinence-only threatening Uganda’s HIV success
story? Lancet Infect Dis 2005; 5: 263–64.
14 Hearst N, Chen S. Condom promotion for AIDS prevention in
the developing world: is it working? Stud Fam Plann 2004; 35:
15 US AID. MEASURE Evaluation Project. http://www.cpc.unc.edu/
measure/publications/index.php (accessed Oct 17, 2005).
16 ORC Macro. MEASURE DHS+ STATcompiler. http://www.
measuredhs.com. (accessed Oct 11, 2005).
17 Meekers D, Van Rossem R. Explaining inconsistencies between
data on condom use and condom sales. BMC Health Serv Res
2005, 5: 5.
18 Cleland J, Boerma JT, Carael M, Weir SS. Monitoring sexual
behaviour in general populations: a synthesis of lessons of the
past decade. Sex Transm Infect 2004; 80 (suppl 2): 1–7.
www.thelancet.com Vol 368 November 18, 2006 1793
19 Slaymaker E. A critique of international indicators of sexual risk
behaviour. Sex Transm Infect 2004; 80 (suppl 2): 13–21.
20 Shelton JD, Halperin DT, Nantulya V, Potts M, Gayle HD,
Holmes KK. Partner reduction is crucial for balanced “ABC”
approach to HIV prevention. BMJ 2004; 328: 891–93.
21 Huygonnet S, Mosha F, Todd J, et al. Incidence of HIV infection
in stable sexual partnerships: a retrospective cohort study of 1802
couples in Mwanza Region, Tanzania. J Acquir Immun Defi c Syndr
2002; 30: 73–80.
22 Quigley M, Munguti K, Grosskurth H, et al. Sexual behaviour
patterns and other risk factors for HIV infection in rural
Tanzania: a case-control study. AIDS 1997; 11: 237–48.
23 Genuis SJ, Genuis SK. Adolescent behaviour should be priority.
BMJ 2004; 328: 894.
24 White R, Cleland J, Carael M. Links between premarital sexual
behaviour and extramarital intercourse: a multi-site analysis.
AIDS 2000; 14: 2323–31.
25 Lagarde E, Auvert B, Chege J, et al. Condom use and its
association with HIV/sexually transmitted diseases in four urban
communities of sub-Saharan Africa. AIDS 2001; 15 (suppl 4):
26 Myer L, Mathews C, Little F. Measuring consistent condom use: a
comparison of cross-sectional and prospective measurements in
South Africa. Int J STD AIDS. 2002; 13: 62–63.
27 United Nations Population Division. Levels and trends of
contraceptive use as assessed in 1998. New York: UN, 1999.
28 Cleland J, Watkins SC. The key lesson of family planning
programmes for HIV/AIDS control. AIDS 2006; 20: 1–3
29 Bledsoe CH, Pison G, eds. Nuptiality in sub-Saharan Africa:
Contemporary anthropological and demographic perspectives.
Oxford: Clarendon Press, 1994.
30 Shah I, Åhman E. Age patterns of unsafe abortion in developing
country regions. Reprod Health Matters 2004: 24: 9–17.
31 Åhman E, Shah I. Unsafe abortion: global and regional estimates
of the incidence of unsafe abortion and associated mortality in
2000. Geneva: WHO, 2004.
32 Mayhew SM, Adjei S. Sexual and reproductive health: challenges
for priority-setting in Ghana’s health reforms. Health Policy Plan
2004; 19 (suppl 1): 50–61.
33 Che Y, Cleland JG, Ali MM. Periodic abstinence in developing
countries: an assessment of failure rates and consequences.
Contraception 2004; 69: 15–21.
34 Castle S. Factors infl uencing young Malians’ reluctance to use
hormonal contraceptives. Stud Fam Plann 2003; 34: 186–99.
35 Ali MM, Cleland J. Sexual and reproductive behaviour among
single women aged 15–24 in eight Latin American countries: a
comparative analysis. Soc Sci Med 2005; 60: 1175–85.