www.thelancet.com/infection Vol 9 February 2009
Heterosexual risk of HIV-1 infection per sexual act:
systematic review and meta-analysis of observational
Marie-Claude Boily, Rebecca F Baggaley, Lei Wang, Benoit Masse, Richard G White, Richard J Hayes, Michel Alary
We did a systematic review and meta-analysis of observational studies of the risk of HIV-1 transmission per
heterosexual contact. 43 publications comprising 25 diff erent study populations were identifi ed. Pooled female-to-
male (0∙04% per act [95% CI 0∙01–0∙14]) and male-to-female (0∙08% per act [95% CI 0∙06–0∙11]) transmission
estimates in high-income countries indicated a low risk of infection in the absence of antiretrovirals. Low-income
country female-to-male (0∙38% per act [95% CI 0∙13–1∙10]) and male-to-female (0∙30% per act [95% CI 0∙14–0∙63])
estimates in the absence of commercial sex exposure (CSE) were higher. In meta-regression analysis, the infectivity
across estimates in the absence of CSE was signifi cantly associated with sex, setting, the interaction between setting
and sex, and antenatal HIV prevalence. The pooled receptive anal intercourse estimate was much higher (1∙7% per
act [95% CI 0∙3–8∙9]). Estimates for the early and late phases of HIV infection were 9∙2 (95% CI 4∙5–18∙8) and
7∙3 (95% CI 4∙5–11∙9) times larger, respectively, than for the asymptomatic phase. After adjusting for CSE, presence
or history of genital ulcers in either couple member increased per-act infectivity 5∙3 (95% CI 1∙4–19∙5) times versus
no sexually transmitted infection. Study estimates among non-circumcised men were at least twice those among
circumcised men. Low-income country estimates were more heterogeneous than high-income country estimates,
which indicates poorer study quality, greater heterogeneity of risk factors, or under-reporting of high-risk behaviour.
Eff orts are needed to better understand these diff erences and to quantify infectivity in low-income countries.
Since the beginning of the HIV epidemic, mother-to-
child transmission and iatrogenic transmission through
contaminated blood products and unsafe injections have
decreased because of improved health procedures and
treatment options, particularly
countries.1–5 However, the notion that diff erent patterns
of sexual behaviours and/or biological factors such as
male circumcision and genital ulcer disease (GUD) can
explain worldwide diff erences in heterosexual epidemic
size has been questioned.6–9 Some believe that sexual
transmission has been overestimated, whereas iatrogenic
transmission has been underestimated.10–12 Quantifi cation
of the risk of HIV infection after sexual intercourse with
an infected partner is needed to better understand the
epidemiology of HIV infection worldwide and to enable
appropriate public-health decisions to be taken.
Sexual transmission estimates fall broadly into two
categories: per-act transmission probabilities,13–23 which
quantify the risk of infection per sexual contact, and per-
partner transmission probabilities,13,24–27 which measure
the cumulative risk of infection over many sex acts during
a partnership. In both cases, transmission probabilities
depend on the infectiousness of the HIV-infected partner
and the susceptibility of the HIV-uninfected partner.
Infectiousness and susceptibility depend on behavioural,
biological, genetic, and immunological risk factors of the
host and the virus.5,6,21–24,28–42 Per-act transmission
probabilities are methodologically diffi cult to measure.43
The time of seroconversion of the index case and the
transmission to his or her partner, the number of
unprotected sex acts, duration of exposure to HIV, and
potential HIV cofactors among the index cases and the
susceptible partners at the time of transmission are rarely
known precisely, especially for time-varying cofactors,
such as recurrent sexually transmitted infections
Early narrative or methodological reviews have reported
a limited selection of per-act estimates.10–12,42,46–48 More
recently, Powers and colleagues49 published a systematic
review of per-act HIV-1 transmission probabilities of
27 studies based on 15 unique study populations. Our
systematic review extends this work by including
43 publications based on 25 diff erent study populations.
Our objectives were to provide summary estimates of
HIV-1 transmission probabilities per heterosexual
contact, to do in-depth univariate and multivariate meta-
regression analyses to explore the variation across study
estimates, and to estimate the infl uence of key risk factors
on infectivity. The review focuses on HIV-1, which is
more pathogenic and prevalent than HIV-2.50,51
The literature search (up to Sept 6, 2008) was done in
three stages. First, PubMed, Science Direct, and NLM
Gateway online databases were searched to September,
2006, by use of the following search terms: “HIV
transmission probability” OR “HIV transmission
probabilities” OR “HIV
infectiousness” NOT “perinatal” NOT “mother to child”
NOT “mother-to-child”, and by replacing “HIV” by the
terms “LAV”, “HTLV-III” and “HTLV III”. PubMed was
searched by titles. Science Direct and NLM Gateway were
searched by abstracts, titles, keywords, and authors. The
PubMed search was updated twice (to June 29, 2007, and
infectivity” OR “HIV
Lancet Infect Dis 2009;
Department of Infectious
Disease Epidemiology, Faculty
of Medicine, Imperial College,
London, UK (M-C Boily PhD,
R F Baggaley PhD); Department
of Epidemiology and
Population Health, London
School of Hygiene and Tropical
Medicine, London, UK
(R G White PhD,
Prof R J Hayes DSc); Population
Health Research Unit, Hôpital
du Saint-Sacrement, Centre
Hospitalier, University of
Quebec, and Department of
Social and Preventive
Medicine, Faculty of Medicine,
Laval University, Quebec City,
Quebec, Canada (M-C Boily,
Prof M Alary PhD); and
Statistical Center for HIV/AIDS
Research and Prevention,
Fred Hutchinson Cancer
Research Center, Seattle, WA,
USA (L Wang PhD, B Masse PhD)
Dr Marie-Claude Boily,
Department of Infectious Disease
Epidemiology, Imperial College,
Norfolk Place, London N1 3LG,
www.thelancet.com/infection Vol 9 February 2009 119
again to Sept 6, 2008) by use of more effi cient search
terms and Boolean operators, for matches under any
fi eld: (HIV OR LAV OR HTLV III OR HTLV-III OR AIDS
OR human immunodefi ciency virus OR human
T-lymphotropic virus III OR acquired immunodefi ciency)
AND (infectiousness OR infectivity OR probability OR
contact OR contacts OR partner OR partners OR wives
OR spouses OR husbands OR couples OR discordant OR
[transmission AND (heterosexual OR homosexual OR
risk OR female OR male OR anal)]). Bibliographies of
relevant articles were examined for additional references.
Four of six authors contacted provided complementary
Selection criteria and data extraction
Publications that reported empirical per-act heterosexual
HIV-1 transmission probability estimates, or suffi cient
information to derive these estimates, were included.
Indirect estimates from mathematical modelling studies,
reviews, pre-1990 abstracts, and studies with sample sizes
fewer than ten were excluded. No other restrictions were
put on language, location, study design, or type of
exposure. Each publication was examined by two
reviewers (RFB, MCB) to extract information on per-act
estimates, 95% CIs, and study and participant
characteristics, which were used to defi ne covariates.
Male-to-female and female-to-male estimates were
extracted in preference to combined estimates. Per-act
estimates stratifi ed by anal intercourse, genital ulcers,
disease stage of the index cases, male circumcision
status, and viral load were also extracted.
Pooled transmission probability estimates and 95% CI
were derived using a random-eff ects model based on
the inverse-variance method.52–54 Natural log (ln)-
transformed study estimates were used to avoid
problems associated with heteroscedasticity.55 To deal
with zero values a small value of 0·000001 was used. If
not explicitly stated in the publication, per-act
transmission probabilities were derived using reports of
total or frequency of sexual contacts. To improve
consistency across studies, infectivity estimates reported
as rates were converted into per-act transmission
probabilities (see webappendix for details).56 Hetero-
geneity across study estimates was explored by use of
the Q statistic, subgroup and sensitivity analyses, and
meta-regression techniques.52–54 Random-eff ects meta-
regression models were fi tted on ln-transformed study
estimates with the procedure “proc Mixed” in SAS
version 9.13. Pooled estimates were exponentiated to
obtain estimates on the original scale.
The main meta-analysis was done using the crude sex-
specifi c estimates from each publication. If multiple
publications reported estimates based on the same study
population, the estimate from the largest or most recent
sample was included—of these estimates, the largest took
precedence. We then did sensitivity analyses by calculating
pooled estimates for diff erent subgroups of studies (eg,
for women only, with and without commercial-sex
exposure [CSE]). We also used univariate and multivariate
meta-regression techniques to explore potential sources
of heterogeneity across estimates with the following
Figure 1: Selection of studies on heterosexual per-act HIV-1 transmission probabilities
The 43 publications included 26 articles that were included in the main meta-analysis and seven articles only
included in the sub-analyses by risk factor.17,20,46,57–60 The remaining articles were duplicates and were not included in
any analysis, but are shown in webtable 1 for completeness.
62 643 abstracts identified
from PubMed and title
788 abstracts examined
61 855 titles excluded due
570 abstracts excluded due
163 abstracts excluded for
relating to oral/homosexual
studies with insufficient
information to derive
14 per-act studies excluded
due to failure to meet
inclusion criteria (reviews,
methodological studies, etc.)
11 studies excluded due to
failure on specific criteria:
4 mathematical modelling
2 obsolete abstracts
1 study identified and
obtained on 4 studies
13 studies identified through
(7 abstracts, 3 book
chapters, 3 journal articles)
218 studies retrieved for
more detailed assessment
54 potentially appropriate
43 publications provided
crude estimates or
by risk factors
26 publications included in
42 per-act studies plus
13 per-partner studies
reporting number or
frequency of sexual
acts retrieved for more
See Online for webappendix and
www.thelancet.com/infection Vol 9 February 2009
covariates: study design, setting, year of publication, sex,
exposure, condom use, STI, contamination, and antenatal
HIV prevalence. Finally, we did a series of secondary
analyses using transmission estimates stratifi ed by risk
Covariates were defi ned by available information from
each study. The covariate setting was used as a marker
of unmeasured risk factors (eg, viral subtype, co-
infection).49 The exposure covariate diff erentiated
between studies done among partners after commercial
sex, as clients or female sex workers (FSWs), or among
partners of index cases infected after blood transfusion,
or those exposed to various other sources of HIV
(including intravenous drug use or those infected
heterosexually). The contamination covariate was
defi ned to indicate the likelihood of exposure to HIV via
sources (sexual or blood) other than sex with the main
index partner. The condom-use covariate characterised
studies in which condom use was rare or somewhat
controlled for. The STI covariate was defi ned to capture
the prevalence of ulcerative STI reported in each study.
HIV prevalence from antenatal clinics (ANC) at the time
and study location reported from independent sources
(eg, WHO’s Global Health Atlas) was used as a marker
of potential unmeasured parenteral or extramarital
exposure, assuming that the risk would increase with
HIV prevalence. Further details are provided in the
Figure 1 shows details of the study selection procedure.
Most studies were excluded because they were risk-factor
analyses, reported non-sexual
transmission, per-partner estimates, or did not provide
enough information to derive an estimate. 42 studies
reporting at least one per-act heterosexual HIV-1
transmission estimate and 13 studies reporting suffi cient
information to derive an estimate were identifi ed from
the PubMed search and in one case by personal
communication. 14 publications, mainly reviews or
methodo logical studies, were rejected. 13 additional
publications were identifi ed by perusing the biblio-
graphies of relevant articles. 11 publications were rejected
based on our pre-defi ned criteria. 43 publications that
reported crude per-act estimates or estimates stratifi ed by
risk factors were found,14–21,56–89 based on 25 study
populations (webtable 1).6,24–28,30,58,61,62,66,70–72,76,77,79,80,84,88–93
Many studies reported results from the same study
population (eg, fi ve studies14,16,81–83 were based on a US
Centers for Disease Control and Prevention [CDC]
study25) and estimates from the most recent or largest
sample were included. Two studies reported on the same
study population,62,63 although we assumed them to be
independent because they
subpopulations over a diff erent period.
analysed diff erent
Overall, four study designs were used by the included
studies: retrospective partner, prospective discordant-
couple, and simple prospective (longitudinal cohort) and
retrospective (cross-sectional) studies. In retrospective
partner studies, the infection status of each partner
becomes known only at the time of the study. The index
case and time of infection are determined on the basis of
exposure to a salient risk factor.15,16,43,60,85 For example,
in transfusion studies, the infection time of index
cases can be determined more precisely from the date of
the transfusion.16,25,27,43,76,79,80 Otherwise, infection time is
estimated by exploring possible dates of infection or by
defi ning a distribution of possible infection times by use
of information from questionnaires and local epidemic
curves or CD4-cell counts.15,16,45,60,81,82,85 In prospective
discordant-couple studies, stable (preferably mono-
gamous) HIV-serodiscordant couples are followed up
after diagnosis of the index partner,19,20,70,72 and the sexual
history and seroconversion of the partner are assessed
prospectively. With simple prospective or retrospective
studies, susceptible or infected and susceptible
individuals (not necessarily monogamous), respectively,
probability* (95% CI)
Q† p value
All study estimates15,16,18,19,21,61–64,66,67,70–73,75–81,84,87,88,89
Stratifi ed by sex
350·182% (0·110–0·299) 1590·5<0·0001
50·179% (0·020–1·572) 91·4 <0·0001
Stratifi ed by sex and CSE
Stratifi ed by sex and setting
Female-to-male without CSE‡ 19,62–63,70,72
Male-to-female without CSE‡19,62–63,67,70–72
8 0·164% (0·056–0·481)
*Random-eff ects models. †Calculated on the ln scale. ‡Studies that included CSE were removed to assess their
infl uence. §Estimates of CSE were all from low-income countries and were the only non-partner studies.
Table 1: Pooled estimates for subsets of crude study estimates stratifi ed by setting, sex, and
commercial-sex exposure (CSE)
For more on WHO’s Global
Health Atlas see http://www.
www.thelancet.com/infection Vol 9 February 2009 121
are recruited after sexual contact with potentially infected,
high-risk partners. Because index cases are not
recruited, exposure to HIV is estimated by use of HIV
prevalence in the pool of potential partners and the
reported coital frequency.21,64,65,73,74
To avoid duplication, 26 of 43 studies were included in
the main meta-analysis of crude (unstratifi ed by risk
factors) estimates (webtable 1). We included male-to-
female and female-to-male estimates were in preference
to combined estimateds where possible. All but one75 of
these 26 studies reported on data collected before 2001
from high-income (Europe, North America) or low-income
(Africa, Asia, Haiti) country settings. Seven studies from
low-income countries were prospective discordant-couple
studies,19,62,63,66,70–72 fi ve were simple retrospective or pro-
spective studies,21,61,64,73,75 and one was a retrospective partner
study.67 High-income country estimates were all derived
from prospective discordant-couple studies18,76,80,84,87–89 or
retrospective partner studies.15,16,77–79,81,87
The reported information on study quality and potential
sources of biases varied across studies. For example, in
retrospective-partner studies, the identifi cation of index
cases and time of infection may be more precise if index
cases have been infected through contaminated blood
products rather than intravenous drug use, or bisexual or
casual sex. Partners of index cases infected through high-
risk behaviour (drug use, sexual promiscuity) may also
have higher-risk activities, and therefore higher rates
of STIs and/or additional sources of exposure other
than sex with the index case. Six retrospective-partner
studies included index cases who were transfusion
recipients;16,76,77,79–81 seven studies included index cases
infected through various sources,15,18,78,84,87–89 including
mainly intravenous drug use;88,89 and eight studies included
index cases probably infected heterosexually.19,62,63,66,67,70–72 All
fi ve non-partner (ie, simple prospective or retrospective)
studies were done in low-income countries among
participants after CSE, as clients,61,64 FSWs,73,75 or men with
multiple partners (including sex with FSWs),21 also with
high rates of STIs.21,61,64
Many retrospective partner or discordant couple studies
attempted to exclude partners with additional sources of
HIV exposure other than sex with the index partner by
use of various exclusion criteria.15,27,30,31,77,88,89,93 For example,
Marincovich and collegues89 excluded partners who
reported parenteral exposure, blood transfusion,
tattooing, and multiple partners, whereas Pedraza and
colleagues88 excluded intravenous drug users and
promiscuous participants. Infrequent exposure of
partners to blood through injections from traditional
healers or multiple sexual partners was reported by a few
participants in two studies.62,63 Based on reported
information, we judged that contamination was possible
in ten studies because of occasional reports of extramarital
sex15,21,62–64,78,81,84 and/or potential exposure to blood.61,63,64,73
Because of the high risk associated with CSE, it was
generally assumed to be the source of infection, which
may not always be the case.61,91,73 Contamination was
thought unlikely in two studies,20,72 because HIV
transmissions within couples were matched by
epidemiological linkage. Failure to control for condom
use may lead to over-estimation of unprotected sex acts
and underestimation of infectivity. Only three studies did
not report any attempt to control for condom use or did
not provide suffi cient information (webtable 1).61,64,78
Figure 2: Crude sex specifi c per-act study estimates
For reference, a vertical dotted line is shown at 0·1% because this has previously been a commonly cited value for
HIV-1 per-act transmission probability.49 Pooled data estimates were calculated by random-eff ects meta-analyses.
Heterogeneity statistics were calculated on the ln scale. Arrow indicates zero value of estimate and/or lower
confi dence limit. See webtable 1 and webappendix for details of individual estimate derivation.
van der Ende (1988)76
Test for heterogeneity: Q=91·7 (p<0·0001)
Test for heterogeneity: Q=426·9 (p<0·0001)
Test for heterogeneity: Q=559·0 (p<0·0001)
0·001% 0·01% 0·1% 1·0% 10%100%
probability per sex act (%)
NEstimate (95% CI)
www.thelancet.com/infection Vol 9 February 2009
The meta-analysis included 35 crude sex-specifi c (male-
to-female, female-to-male, combined) transmission
probability estimates (webtable 1, fi gure 2). One
publication reported independent estimates from both
the prospective discordant and retrospective partner
study components, which were both included.87 Per-act
estimates ranged from zero76,84,89 to 8∙2%,61 and showed
highly signifi cant heterogeneity (table 1, fi gure 2). The
highest (>0·1%) estimates were mostly from low-income
countries. The heterogeneity across estimates remained
signifi cant even after stratifi cation by sex (table 1). With
further stratifi cation by setting (high-income vs low-
income countries), the heterogeneity across sex-specifi c
study estimates was no longer signifi cant for high-
income countries only. The pooled combined, female-to-
male, and male-to-female high-income country estimates
were 0∙08% per act (95% CI 0∙04–0∙16), 0∙04% per act
(95% CI 0∙01–0∙14), and 0∙08% per act (95% CI
0∙06–0∙11), respectively. By contrast, the pooled female-
to-male and male-to-female estimates for low-income
countries were 0∙867% per act (95% CI 0∙279–2∙601)
and 0∙193% per act (95% CI 0∙086–0∙433), respectively.
The pooled male-to-female estimate with CSE only was
much lower than the female-to-male estimates, indicating
the relatively lower Senegalese and recent Kenyan
estimates (webtable 1).73,75 Interestingly, by excluding
estimates after CSE, which were the only estimates from
simple prospective and retrospective studies and were
exclusively from low-income countries, the pooled male-
to-female estimates increased, whereas the female-to-
male estimates decreased (table 1). The heterogeneity
between low-income country estimates remained.
In univariate meta-regression analyses, a substantial
fraction of the variability across all 35 study estimates
could be explained by either exposure, setting, STI
prevalence, condom use, design, or ANC prevalence
(webtable 2). Greater infectivity was associated with CSE,
low-income country setting, studies that did not control
for condom use, non-partner studies, and higher STI or
higher ANC HIV prevalence. The covariates condom and
STI (borderline) were no longer signifi cant after excluding
estimates with CSE (webtable 2). Among all low-income
country estimates, only sex, condom use, and year of
publication (negative association) were signifi cantly
associated with infectivity; no association was found after
removing the estimates with CSE (webtable 2).
The multivariate meta-regression analyses aimed to
explain the heterogeneity across the 30 high-income and
low-income country estimates without CSE, which were
all based on discordant-couple or retrospective partner
studies. In models that controlled for sex (p>0·23) and
study design (p>0·49), only setting, ANC prevalence or
exposure were independently associated with infectivity
(p<0∙0001) and explained 62–68% of the variability (not
shown). In models that included design (p>0·10), sex
(p<0·015), setting (p<0·0001), and the interaction
between setting and sex (p<0·036), only contamination
(p=0·009) or ANC prevalence (p=0·006) remained
signifi cant and together explained 83–85% of the
variability (not shown). Lower infectivity estimates were
associated with the contamination category “no
information” compared with the categories “possible” or
“unlikely”, which were not statistically diff erent (p=0∙45).
Thus, our fi nal model excluded design and included ANC
prevalence (table 2).
Only two studies reported male-to-female estimates for
receptive anal intercourse (pooled estimate 1∙69% per act
[95% CI 0∙32–8∙91]),59,60 and fi ve studies explicitly
reported male-to-female estimates for vaginal sex only
(pooled estimate 0∙076%
0∙052–0∙111]).16,18,60,70,79 Additional information on these
estimates is available from the authors on request.
Six publications reported low-income country estimates
stratifi ed by GUD status of HIV-1-susceptible part-
ners,21,57,61,65,69 which indicated increased HIV sus cep tibility
caused by GUD, or by GUD status of the index case,19
which indicated increased HIV infectivity. One study was
excluded because the confi dence interval could not be
derived.69 The only study19 among stable couples, reported
lower infectivity in the presence of GUD than those
reported in the presence of CSE.21,57,61,65 An additional
eight study estimates in the absence of STI were also inc-
luded.18,59,65,71,77,80,81,87 Because of the small number of
per act [95% CI
RR (95% CI) p value
ANC HIV prevalence†
Fraction of the variance explained=82% (calculated on the ln scale). Final model:
sex (p=0·014)+setting (p=0·0001)+setting×sex (p=0·021)+antenatal clinic (ANC)
HIV prevalence (p=0·007). *Includes only one study.66 †The natural logarithm of
the infectivity estimate was increased linearly by 0·046 times for each 1% increase
in ANC HIV prevalence. The covariates explored in the diff erent multi-regression
models included sex, design, setting, exposure, condom, sexually transmitted
infections, contamination, ANC HIV prevalence.
Table 2: Main multivariate analysis: fi nal meta-regression model for the
subset of crude study estimates from partner studies (n=30) without
See Online for webtable 2
www.thelancet.com/infection Vol 9 February 2009 123
estimates, simple explanatory meta-regression analyses
were done. We classifi ed the estimates into three
categories: study participants without STI; without GUD
but potentially other STI; and with GUD and potentially
other STI (fi gure 3). The covariate GUD status alone
explained 57% of the variability across study estimates.
The meta-regression model with the covariates CSE and
GUD status explained a larger fraction of the variability
(81%) than GUD status with either covariates setting
(77%) or sex (70%; details not shown). Estimates in the
presence of GUD were fi ve times larger than estimates in
absence of STI, whereas CSE was associated with an
11-times increase in infectivity compared to estimates
without CSE (table 3).
Seven studies reported estimates by disease stage of
index partners from partner studies (fi gure 4).17,20,56,58,60,78,81
We only included one20 of the two studies reporting on
the same low income country population.20,56 Wawer and
colleagues20 reported many estimates from diff erent
subsamples of discordant couples in which index cases
had been infected for diff erent lengths of time. We used
the pooled estimate from couples for which index cases
had seroconverted for less than 5 months (1∙07% per
act), which was larger than from couples 6–15 months
(0·17% per act) and 16–35 months (0·10% per act) after
the index cases had seroconverted (fi gure 4). The study
estimate from all couples with prevalent index cases
(0∙08% per act) was used for the asymptomatic stage.
The late-stage estimate used corresponded to 6–15 months
before death of index cases (0∙49% per act).20 Per-act
estimates were 0∙29–1∙07%,
0∙13–5∙67% for the early, asymptomatic, and late-stage
disease, respectively. Disease stage alone explained 95%
of the variability across estimates. After adjusting for
disease stages, the addition of the setting covariate was
not signifi cant (table 3). The impact of mode of sexual
transmission could not be explored owing to lack of data.
The risk in the early (risk ratio [RR] 9∙2 [95% CI
4∙5–18∙8]) and late (RR 7∙3 [95% CI 4∙5–11∙9]) disease
stages, adjusted for setting, were signifi cantly larger than
for the asymptomatic phase (table 3).
Only two studies reported empirical estimates
stratifi ed by level of either semen or serum viral load on
the same study population (not shown).19,20 Partners of
index cases who had a median serum viral load of
approximately 30 000 HIV RNA copies per mL (range
<400–3·1×10⁶ copies per mL <5 months after
seroconversion) had a higher infectivity (1∙07% per act)
than those with a median serum viral load of
approximately 2600 copies/mL by 15 months,20 and even
higher than Gray’s estimate (0∙23% per act) if viral load
exceeded 38 500 copies per mL.19 Wawer and colleagues’20
estimate from couples for which prevalent index cases
were followed up for 0–10 months was higher (0∙09%
per act at a median of ~10 300 copies per mL), albeit not
signifi cantly, than when followed up for more than
30 months (0∙04% per act at a median of ~15 000 copies
per mL). Gray and colleagues’19 combined per-act esti-
mates at high (>38 500 serum viral load copies per mL),
medium (~1700–12 499 or 12 500–38 499 copies per mL),
and low (<1700 copies per mL) viral loads were 0∙23%,
0∙13% or 0∙14%, and 0∙01%, respectively. Two studies
also reported higher infectivity at higher viral load, but
their estimates were not directly comparable because
they were derived from theoretical studies based on
measurement of HIV-1 viral load by volume of
Only two studies reported female-to-male estimates by
circumcision status (not shown).21,61 Baeten and
colleagues’21 study female-to-male transmission estimate
among uncircumcised men was approximately 2∙6 times
that among circumcised men (1∙3% per act [95% CI
0∙5–2∙0] vs 0∙5% per act [95% CI 0∙3–0∙7]) and 4∙5 times
larger in non-circumcised than circumcised men in the
presence of GUD (1∙8% per act [95% CI 0∙0–3∙7] vs
0∙4% per act [95% CI 0∙0–0∙9]). In Cameron and
No STI in partner
Test for heterogeneity: Q=86·7 (p<0·0001)
Pooled, without Mastro (1994)65
Test for heterogeneity: Q=38·0 (p<0·0001)
No GUD in partner
Test for heterogeneity: Q=102·5 (p<0·0001)
GUD in partner
Test for heterogeneity: Q=51·6 (p<0·0001)
Setting and CSE status
Low-income, no CSE
High-income, no CSE
Sex transmission mode
0·001% 0·01%0·1% 1·0% 10%100%
probability per sex act (%)
N Estimate (95% CI)
Figure 3: Per-act and pooled estimates for sub-analyses of estimates stratifi ed by genital ulcer disease (GUD)
status in HIV-1 susceptible partner†
Pooled data estimates were calculated by random-eff ects meta-analyses. Heterogeneity statistics were calculated
on the ln scale. Number (N) of participants in the subgroup sample (*total sample size). Arrow indicates zero value
of estimate and/or lower confi dence limit. See webappendix for details of individual estimate derivation. †Only
one study19 reported GUD status for the index cases rather than for HIV-susceptible partners. Estimate for Halperin
et al59 was adjusted for anal intercourse, condom use, and history of sexually transmitted infection. Hayes et al’s57
estimate was obtained during episodes of GUD, rather than during follow-up (which included periods with and
without GUD episodes). Details for this analysis are available from the authors on request.
www.thelancet.com/infection Vol 9 February 2009
colleagues’ study,61 study estimates were higher among
uncircumcised men (18∙5% per act [95% CI 2∙3–34∙8])
than among circumcised men (2∙2% per act [95% CI
0∙0–6∙4]). Among those with GUD, estimates were six
times higher among uncircumcised men (42∙8% per act
[95% CI 1∙26–73∙0]) than among circumcised men (6∙7%
per act [95% CI 0∙0–19∙2]). In the absence of GUD, no
HIV transmission occurred
in circumcised or
Our systematic review and meta-analysis of HIV-1
transmission probabilities per heterosexual act updates
and extends the fi ndings of a recent similar review.49 We
confi rmed the earlier observation of substantial
heterogeneity in per-act estimates,49 provided sex-specifi c
transmission estimates, and identifi ed additional sources
of heterogeneity by exploring interactions between
covariates. We also reported the infl uence of key risk
factors on infectivity in terms of relative risk (risk ratios),
instead of risk diff erence, which is easier to interpret.
Heterogeneity across crude study estimates could be
mostly explained by CSE as FSWs or clients, setting, sex,
and ANC HIV prevalence at the time and location of the
study. Although a previous review only found a weak
association between sex and infectivity,49 our results
suggested that this may vary by settings. In the subset of
estimates without CSE, the pooled female-to-male
transmission estimate for high-income countries,
adjusted for HIV prevalence, was about half the male-to-
female or combined estimates (RR about 0∙5), although
the diff erence failed to reach signifi cance. By contrast,
the adjusted low-income country female-to-male and
male-to-female estimates were very similar (RR about
1∙0), and the female-to-male low-income country estimate
(RR about 3∙3) was signifi cantly larger than the female-
to-male high-income country estimate. The male-to-
female or combined pooled estimates in our sub-analyses
in the absence of receptive anal intercourse, GUD, CSE,
or for the asymptomatic phase were of similar magnitude
(about 0∙07% per act) to the male-to-female and combined
pooled estimates from high-income countries (about
0∙08% per act), which would suggest that they represent
the average per-vaginal-sex-act transmission in absence
of cofactors, during the asymptomatic phase.
Despite diff erences in some selection criteria and the
strategy adopted for the analysis, we confi rmed the
fi ndings of previous reviews on the weak infl uence of
study quality,49 and the importance of key risk factors on
infectivity.16,39,42,49,69,94 In agreement with studies among
homosexual men,5,95–97 our pooled estimate by receptive
anal intercourse supports evidence that it is a more risky
practice than receptive vaginal sex. Two studies reporting
per-act estimates by circumcision status suggested a
three to eight times increase in HIV infection among
uncircumcised men overall or in presence of GUD.21,61
This is consistent, yet somewhat higher, with the results
Sex transmission mode
probability per sex act (%)
NEstimate (95% CI)
Test for heterogeneity: Q=4·2 (p=0·04)
Test for heterogeneity: Q=7·5 (p=0·186)
Test for heterogeneity: Q=18·5 (p=0·002)
0·001% 0·01%0·1%1·0% 10%100%
Figure 4: Per-act and pooled estimates for sub-analyses of study estimates stratifi ed by HIV-1 disease stage
Pooled data estimates were calculated by random-eff ects meta-analyses. Heterogeneity statistics were calculated
on the ln scale. Number (N) of participants in the subgroup sample (*total sample size). Arrow indicates zero value
of estimate and/or lower confi dence limit. See webappendix for details of individual estimate derivation. †For
early-stage disease, we used the data from Wawer et al20 at <5 months since seroconversion of the index case, in
preference to the data provided by Pinkerton et al56 (not shown). *For late-stage disease, we used the estimates
obtained 6–15 months before death for Wawer et al’s20 study. Details on selection of estimates for this analysis is
available from the authors on request.
RR (95%CI)p value
Analysis by GUD status (17 study estimates)*
Analysis by disease stage (14 study estimates)†
*Fraction of the total variance explained=81% (calculated on ln scale). †Fraction of
the total variance explained=96% (calculated on ln scale).
Table 3: Multivariate meta-regression models for sub-analysis by genital
ulcer disease (GUD) status and disease stage
www.thelancet.com/infection Vol 9 February 2009 125
of two previous meta-analyses,94,98 and three recent
randomised controlled trials of male circumcision.99–101
We found that the presence of GUD and CSE were
independently associated with increased infectivity. Our
GUD cofactor estimate (RR 5∙3) was intermediate
between previous study estimates for high-risk groups
(10–50 and 50–300 for male-to-female and female-to-
male transmission per act, respectively),57 and those from
a meta-analysis of observational studies that reported a
2∙8 times (95% CI 2∙0–4∙0) and 4∙4 times (95% CI
2∙9–6∙6) increase in female and male susceptibility
caused by GUD, respectively.102
Our RRs and estimates from observational studies may
be biased because of misclassifi cation, undiagnosed STI,
or misreporting of symptoms. Additionally, the
intermittent nature of GUD means that it is unlikely to
have been present throughout the at-risk period, and per-
contact cofactor eff ects may therefore be underestimated.44,57
Our cofactor estimate predominantly captured the
increased HIV susceptibility caused by GUD (only one
study reported estimates stratifi ed by GUD status of index
cases19). Thus, the increased risk associated with CSE may
partly indicate increases in HIV infectivity because high-
risk index cases (FSW, clients) would probably have also
been infected with GUD or other STIs. Baeten and
colleagues’21 female-to-male study estimate from men
with multiple partners (31% monogamous, 57% sex with
FSW) was higher than from the subsample of men who
only reported sex with their wives (0∙63% [0·35–0·91] vs
0∙38% [<0·01–69·73] per act, p>0∙10).
Interestingly, the early Kenyan61 and Thai64 FSW-to-
client estimates were substantially larger than the
Senegalese and the recent Kenyan client-to-FSW
estimates.73–75 Although these estimates are probably
imprecise because they were based on simple retrospective
or cross-sectional study design, the large diff erence
(>35 times) could also be caused by other cofactors. STI
prevalence may have been lower among Senegalese FSWs
because of an early governmental public-health
programme, whereby self-identifi ed FSW regularly
attended health clinics providing free STI treatment.103,104
By contrast, the client studies were done in east Africa
(early in the epidemic) and Thailand, where male
circumcision prevalence is lower than in west Africa,105
and at a time when STI and GUD were virtually
ubiquitous, FSW were experiencing an explosive HIV
epidemic, and index partners were more likely to be in
the primary phase of HIV infection.46,61,90,106 For example,
one group reported prevalences of 21% Haemophilus
ducreyi, 80% herpes simplex virus 2, and 9% GUD among
Thai FSWs.65,106 Additionally, Kimani and colleagues75
suggested that the decline in per-act infectivity observed
over calendar time in their study correlated with a decline
in STI prevalence among FSWs.
Previous individual-based studies showed an association
between HIV infectivity and viral load or time since
infection.22,24,28,107–110 Our risk-factor analysis also suggested
increased infectivity for index cases in the early and late
phase of infection compared with the asymptomatic
phase. The diff erence between estimates for the early and
late stages was not signifi cant, which may indicate similar
infectivity, under-sampling of couples with most recently
infected and highly infectious index cases, imprecise
defi nition of the duration of the early phase, or lack of
statistical power. A recent re-analysis of Wawer and
colleagues’ data20 suggested that primary infection and
late-stage infection were 26 and seven times higher than
asymptomatic infection, respectively, and that the high
infectiousness during primary
approximately 3 months.111
We initially did not impose any inclusion criteria based
on study design because each design has intrinsic biases,
even prospective discordant-partner studies, which are
seen as the most appropriate design to estimate
transmission probabilities. Although discordant-partner
studies are likely to reduce recall biases regarding type
and frequency of unprotected contacts and HIV cofactors,
the reporting of sensitive behaviour is still subject to
social desirability biases. Frailty selection, whereby the
most vulnerable couples of so-called “high and fast
transmitters” rapidly become seroconcordant,16,45 may also
result in over-sampling of less susceptible partners and/
or less infective index cases who remain uninfected
longer and become more likely to be enrolled in such
studies. Frailty selection would result in under-estimation
of infectivity. Shiboski and colleagues16,45 have also
suggested that heterogeneity in infectivity was not well
shown by the US CDC data25 and CDC Heterosexual
AIDS Transmission Study27 retrospective-partner studies
because the duration of many relationships was too short
compared with the time since infection of index cases.
Results from our risk-factor analyses are mainly
explanatory. The estimates of the magnitude of the cofactor
eff ects may not be very precise because of the small
number of studies and covariates that could be explored,
the heterogeneity across study estimates, diff erences in
risk-factor exposure defi nitions across studies, and
because study estimates were based on subgroups of the
study sample. Publication biases may also be present
because estimates by risk factor may not be reported from
studies that did not fi nd a signifi cant association.
The independent positive association between infectivity
and setting or ANC HIV prevalence for studies without
CSE is diffi cult to interpret, but is unlikely to be caused by
study design or analytic methods. As reported previously,49
study design was only weakly associated with infectivity.
Additionally, we converted estimates reported as rates into
probabilities (webappendix), which improved comparability
across studies. Larger transmission probabilities may lead
to higher HIV prevalence in the general population, as
estimated in low-income countries. Alternatively, higher
HIV prevalence may increase the likelihood of
contamination resulting from exposure to additional
sources of infection other than sex with the main index
www.thelancet.com/infection Vol 9 February 2009
partner and thus bias estimates upward. Low-income
country estimates displayed greater heterogeneity than
high-income country estimates. Sex, date of publication, or
the covariate condom (confounded with CSE) only
explained a signifi cant fraction of the variation across low-
income country estimates (if estimates with CSE were
included). This is not entirely surprising given the limited
number of studies and that the STI, contamination, and
condom-use covariates could only be defi ned broadly,
leading to potential misclassifi cation. Thus, the
heterogeneity may indicate uncaptured contamination or
variation in the prevalence of key risk factors. For example,
the larger female-to-male than male-to-female estimates in
three discordant-partner studies19,62,70 in low-income
countries may indicate contamination, because men are
more likely than women to report extramarital sex before
or during the study period.28,62,63,66,71,112 Interestingly, in Fideli
and colleagues’ study,72 in which transmission events
within couples could be epidemiologically linked, female-
to-male transmission was lower than male-to-female
transmission. However, their estimates were larger than
Wawer and colleagues’ study,20 whereby infections within
couples were also confi rmed by epidemiological linkage,
which reduces the risk of misclassifi cation, but does not
reduce biases caused by misreporting of number of
unprotected sex acts or unmeasured risk factors.113
Because many studies in low-income countries were
done within the context of interventions involving an
important counselling component,19,20,62,70,71 condom use
may have been over-reported by study participants, leading
to higher infectivity estimates. Nevertheless, reported
condom use remained low or even decreased in some
studies.19,70 Other studies tried to minimise misreporting
biases on sexual behaviour by checking for concordance
between both members in the couple or using sexual
diaries.19,66 In Roth and colleagues’ study,63 because men
reported more protected sex acts then women, we used the
sexual activity reported by women to minimise over-
estimation of infectivity. Confl icting evidence remains
about unmeasured exposure to contaminated equipment
or blood transfusion that may have increased low-income
country estimates.7,8,10–12,114–116 An early cohort study of
registered Senegalese FSWs reported high prevalence of
transfusion, scarifi cation, excision, or tattoos, yet HIV
prevalence in west Africa and the reported transmission
probability estimate for this population are low.73,74,91,103,104
We cannot exclude the possibility that our high and
heterogeneous low-income country estimates are a result
of unmeasured heterogeneity in the prevalence of risk
factors. To assert that a 3∙5-times diff erence in female-to-
male pooled estimates between low-income and high-
income countries is solely caused by contamination would
imply that approximately 70% of infections are acquired
outside the main relationship. Although this seems
inconsistent with the relatively low proportion of unlinked
infections reported in at least two studies,20,72,112 this
remains a subject of debate.114–117 Powers and colleagues49
reported a weak association between region and infectivity,
which they assumed was a proxy for viral subtypes.
However, they also found greater heterogeneity across
estimates from Africa. The reason for the diff erences by
setting is likely to be multifactorial. Lack of male
circumcision may be more important in low-income
countries than in Europe (where circumcision is rare)
because of interacting cofactors such as ulcerative
STI.39,118–121 Between-settings diff erences may never be
completely understood because risk factors such as STI
prevalences may have changed since the beginning of the
epidemic.75,122 Greater heterogeneity in risk factors or
median viral loads in low-income countries may exacerbate
frailty selection over time. Median plasma viral loads as
high as 1·26×10⁶ copies per mL have been observed
among acutely infected men in Malawi, and presence of
STI was the stronger risk factor associated with high viral
load.42,121 Thus, intermittent interaction between risk
factors may result in very high peaks of infectivity during
the incubation period and results in frailty selection at the
population level.42,121 This possibility may also explain why
estimates tended to be lower (albeit not signifi cantly) for
couples with prevalent index cases with 31–40 months of
follow-up (0∙04% per act), compared with 0–10 months
(0∙09% per act), despite the higher median viral load
reported after 30 months.20 However, unmeasured
reduction in prevalence of risk factors resulting from
longer exposure to the study or other intervention is also
Heterogeneity across estimates may also be caused by
population-level declines in infectivity over calendar time
as the fraction of recent seroconverters is expected to
decrease in maturing epidemics.57,62,63,75 Nonoxinol 9
spermicide, which has been associated with increased
susceptibility to GUD and HIV infection,36,123,124 was also
reported in at least four early African studies.62,63,70,72
However, in the study by Allen and colleagues,62 only 12%
of women reported the use of nonoxinol 9 without
condoms, 6% and 19% reported a history of STI in the
past year and past 2 years, respectively, which were similar
to estimates reported in the Rakai study.19 Most studies
were done before wide-scale use of antiretroviral therapy
and is therefore unlikely to have infl uenced results.29,32
Our results indicated higher transmission probabilities
for low-income than for high-income country studies. The
greater heterogeneity of low-income country estimates is
itself interesting and may suggest poorer study quality,
greater heterogeneity in risk factors, or greater under-
reporting of high-risk behaviour in these studies. More
research is needed to better understand these diff erences,
and particularly the low estimates from Rakai.19,20 Greater
heterogeneity may also be caused by diff erential infectivity
of the diff erent viral subtypes, mutation of chemokine-
receptor genes, contraception method, genetic, biological,
and virological host factors, and interaction with other
These are described
in detail in the
www.thelancet.com/infection Vol 9 February 2009 127
infectious diseases,5,33–41,50,108–111,118,123–125 Better quantifi cation
of per-act infectivity is important to improve understanding
of the epidemiology of HIV/AIDS worldwide, to predict
the future HIV/AIDS pandemic, and to design appropriate
prevention strategies. The methodology of discordant-
partner studies could be improved by designing and
powering them for carefully planned risk-factor analyses,
including epidemiological linkage, by use of data collection
methods to reduce social desirability biases, cross-
validating sexual history in couples, and carefully
potential sources of
Confl icts of interest
We declare that we have no confl icts of interest. RFB was supported
during part of the research by an unrestricted educational grant from
This work was supported by the Wellcome Trust (RFB [GR082623MA]
and RGW [GR078499MA]), GlaxoSmithKline (RFB), and the UK Medical
Research Council (RGW). Part of this work was supported by the
National Institute of Allergy Infectious Diseases of the US National
Institutes of Health (BM and LW: 5U01AI068615). MA is a National
Researcher of the Fonds de la Recherche en Santé du Québec, Canada
(grant #8722). Support for this study was partly provided by the Bill and
Melinda Gates Foundation (M-CB). The views expressed herein are those
of the authors and do not necessarily refl ect the offi cial policy or position
of the Bill & Melinda Gates Foundation. We thank Elizabeth Baggaley
and Kamal Desai for language translations, and Peter Gilbert and
Stephen Shiboski for personal communication. We would also like to
thank the anonymous reviewers for very useful comments.
1 Brewer TH, Hasbun J, Ryan CA, et al. Migration, ethnicity and
environment: HIV risk factors for women on the sugar cane
plantations of the Dominican Republic. AIDS 1998; 12: 1879–87.
2 Fiscus SA, Adimora AA, Schoenbach VJ, et al. Perinatal HIV
infection and the eff ect of zidovudine therapy on transmission in
rural and urban counties. JAMA 1996; 275: 1483–88.
3 Harris NS, Thomson SJ, Call R, et al. Zidovudine and perinatal
human immunodefi ciency virus type 1 transmission: a population-
based approach. Pediatrics 2002; 109: e60.
4 Thaithumyanon P, Thisyakorn U, Limpongsanurak S, et al.
Intrapartum and neonatal zidovudine treatment in reduction of
perinatal HIV-1 transmission in Bangkok. J Med Assoc Thai 2001;
5 Baggaley RF, Boily MC, White RG, Alary M. Systematic review of
HIV-1 transmission probabilities: In absence of antiretroviral
therapy. Geneva: UNAIDS Reference Group on Estimates,
Modelling and Projection, 2004.
6 Plummer FA, Simonsen JN, Cameron DW, et al. Cofactors in male-
female sexual transmission of human immunodefi ciency virus
type 1. J Infect Dis 1991; 163: 233–39.
7 Schmid GP, Buve A, Mugyenyi P, et al. Transmission of HIV-1
infection in sub-Saharan Africa and eff ect of elimination of unsafe
injections. Lancet 2004; 363: 482–88.
8 Buve A, Bishikwabo-Nsarhaza K, Mutangadura G. The spread and
eff ect of HIV-1 infection in sub-Saharan Africa. Lancet 2002;
9 Nyindo M. Complementary factors contributing to the rapid spread of
HIV-1 in sub-Saharan Africa: a review. East Afr Med J 2005; 82: 40–46.
10 Deuchert E, Brody S. Plausible and implausible parameters for
mathematical modeling of nominal heterosexual HIV transmission.
Ann Epidemiol 2007; 17: 237–44.
11 Gisselquist D, Potterat JJ, Brody S, Vachon F. Let it be sexual: how
health care transmission of AIDS in Africa was ignored.
Int J STD AIDS 2003; 14: 148–61.
12 Gisselquist D, Potterat JJ, Brody S. Running on empty: sexual co-
factors are insuffi cient to fuel Africa’s turbocharged HIV epidemic.
Int J STD AIDS 2004; 15: 442–52.
13 Garnett GP, Rottingen JA. Measuring the risk of HIV transmission.
AIDS 2001; 15: 641–43.
14 Kaplan EH. Modeling HIV infectivity: must sex acts be counted?
J Acquir Immune Defi c Syndr 1990; 3: 55–61.
15 Padian NS, Shiboski SC, Glass SO, Vittinghoff E. Heterosexual
transmission of human immunodefi ciency virus (HIV) in northern
California: results from a ten-year study. Am J Epidemiol 1997;
16 Shiboski SC, Padian NS. Epidemiologic evidence for time variation
in HIV infectivity. J Acquir Immune Defi c Syndr Hum Retrovirol 1998;
17 de Vincenzi IA, European Study Group on Heterosexual
Transmission of HIV. Longitudinal study of human
immunodefi ciency virus transmission by heterosexual partners.
N Engl J Med 1994; 331: 341–46.
18 Saracco A, Veglia F, Lazzarin A. Risk of HIV-1 transmission in
heterosexual stable and random couples. The Italian Partner Study.
J Biol Regul Homeost Agents 1997; 11: 3–6.
19 Gray RH, Wawer MJ, Brookmeyer R, et al. Probability of HIV-1
transmission per coital act in monogamous, heterosexual, HIV-1-
discordant couples in Rakai, Uganda. Lancet 2001; 357: 1149–53.
20 Wawer MJ, Gray RH, Sewankambo NK, et al. Rates of HIV-1
transmission per coital act, by stage of HIV-1 infection, in Rakai,
Uganda. J Infect Dis 2005; 191: 1403–09.
21 Baeten JM, Richardson BA, Lavreys L, et al. Female-to-male
infectivity of HIV-1 among circumcised and uncircumcised Kenyan
men. J Infect Dis 2005; 191: 546–53.
22 Pilcher CD, Tien HC, Eron JJ Jr, et al. Brief but effi cient: acute HIV
infection and the sexual transmission of HIV. J Infect Dis 2004;
23 Chakraborty H, Sen PK, Helms RW, et al. Viral burden in genital
secretions determines male-to-female sexual transmission of HIV-1:
a probabilistic empiric model. AIDS 2001; 15: 621–27.
24 Quinn TC, Wawer MJ, Sewankambo N, et al. Viral load and
heterosexual transmission of human immunodefi ciency virus
type 1. N Engl J Med 2000; 342: 921–29.
25 Peterman TA, Stoneburner RL, Allen JR, et al. Risk of human
immunodefi ciency virus transmission from heterosexual adults
with transfusion associated infections. JAMA 1988; 259: 55–58.
26 Padian N, Marquis L, Francis DP, et al. Male-to-female transmission
of human immunodefi ciency virus. JAMA 1987; 258: 788–90.
27 O’Brien TR, Busch MP, Donegan E, et al. Heterosexual
transmission of human immunodefi ciency virus type 1 from
transfusion recipients to their sex partners.
J Acquir Immune Defi c Syndr 1994; 7: 705–10.
28 Tovanabutra S, Robison V, Wongtrakul J, et al. Male viral load and
heterosexual transmission of HIV-1 subtype E in northern Thailand.
J Acquir Immune Defi c Syndr 2002; 29: 275–83.
29 Porco TC, Martin JN, Page-Shafer KA, et al. Decline in HIV
infectivity following the introduction of highly active antiretroviral
therapy. AIDS 2004; 18: 81–88.
30 European Study Group on Heterosexual Transmission of HIV.
Comparison of female to male and male to female transmission of
HIV in 563 stable couples. BMJ 1992; 304: 809–13.
31 European Study Group on Heterosexual Transmission of HIV. Risk
factors for male to female transmission of HIV. BMJ 1989;
32 Wilson DP, Law MG, Grulich AE, Cooper DA, Kaldor JM. Relation
between HIV viral load and infectiousness: a model-based analysis.
Lancet 2008; 372: 314–20.
33 Holmberg SD, Horsburgh CR Jr, Ward JW, Jaff e HW. Biologic
factors in the sexual transmission of human immunodefi ciency
virus. J Infect Dis 1989; 160: 116–25.
34 Buchacz KA, Wilkinson DA, Krowka JF, et al. Genetic and
immunological host factors associated with susceptibility to HIV-1
infection. AIDS 1998; 12 (suppl A): S87–94.
35 Samson M, Libert F, Doranz BJ, et al. Resistance to HIV-1 infection
in caucasian individuals bearing mutant alleles of the CCR-5
chemokine receptor gene. Nature 1996; 382: 722–25.
36 Van Damme L, Ramjee G, Alary M, et al. Eff ectiveness of COL-
1492, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female
sex workers: a randomised controlled trial. Lancet 2002;
www.thelancet.com/infection Vol 9 February 2009
37 Hudgens, MG, Longini IM, Vanichseni S, et al. Subtype-specifi c
transmission probabilities for human immunodefi ciency virus type
1 among injecting drug users in Bangkok, Thailand. Am J Epidemiol
2002; 155: 159–68.
38 Soto-Ramirez LE, Renjifo B, McLane MF, et al. HIV-1 Langerhans’
cell tropism associated with heterosexual transmission of HIV.
Science 1996; 271: 1291–93.
39 Wasserheit JN. Epidemiological synergy. Interrelationships
between human immunodefi ciency virus infection and other
sexually transmitted diseases. Sex Transm Dis 1992; 19: 61–77.
40 Laga M, Alary M, Nzila N, et al. Condom promotion, sexually
transmitted diseases treatment, and declining incidence of HIV-1
infection in female Zairian sex workers. Lancet 1994; 344: 246–48.
41 Sewankambo N, Gray RH, Wawer MJ, et al. HIV-1 infection
associated with abnormal vaginal fl ora morphology and bacterial
vaginosis. Lancet 1997; 350: 546–50.
42 Cohen MS, Pilcher CD. Amplifi ed HIV transmission and new
approaches to HIV prevention. J Infect Dis 2005; 191: 1391–93.
43 Shiboski S, Padian NS. Population- and individual-based
approaches to the design and analysis of epidemiologic studies of
sexually transmitted disease transmission. J Infect Dis 1996;
174 (suppl 2): S188–200.
44 Boily MC, Anderson RM. Measure human immunodefi ciency virus
transmission and the role of other sexually transmitted diseases.
Measures of association and study design. Sex Transm Dis 1996;
45 Shiboski SC, Jewell NP. Statistical analysis of the time dependence
of HIV infectivity based on partner study data. J Am Stat Assoc 1992;
46 Mastro TD, de Vincenzi I. Probabilities of sexual HIV-1
transmission. AIDS 1996; 10 (suppl A): S75–82.
47 May RM, Anderson RM. Transmission dynamics of HIV infection.
Nature 1987; 326: 137–42.
48 Gisselquist D, Potterat JJ. Heterosexual transmission of HIV in
Africa: an empiric estimate. Int J STD AIDS 2003; 14: 162–73.
49 Powers KA, Poole C, Pettifor AE, Cohen MS. Rethinking the
heterosexual infectivity of HIV-1: systematic review and
meta-analysis. Lancet Infect Dis 2008; 8: 553–63.
50 Poulsen AG, Kvinesdal BB, Aaby P, et al. Lack of evidence of vertical
transmission of human immunodefi ciency virus type 2 in a sample
of the general population in Bissau. J Acquir Immune Defi c Syndr
1992; 5: 25–30.
51 De Cock KM, Adjorlolo G, Ekpini E, et al. Epidemiology and
transmission of HIV-2: why there is no HIV-2 pandemic. JAMA
1993; 270: 2083–86.
52 Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for
examining heterogeneity and combining results from several
studies in meta-analysis. In: Egger M, Davey Smith G,
Altman DG, eds. Systematic reviews in health care: meta-analysis
in context, 2nd edn. London: BMJ Publishing Group, 2001:
53 Hartung J, Knapp G. An alternative test procedure for meta-
analysis. In: Schulze R, Holling H, Bohning D, eds. Meta-analysis.
New developments and applications in medical and social sciences.
Cambridge, MA: Hogrefe & Huber, 2003.
54 van Houwelingen, HC, Arends LR, Stijnen T. Tutorial in
biostatistics. Advanced methods in meta-analysis: multivariate
apporach and meta-regression. Stat Med 2002; 21: 589–624.
55 Baggaley RF, Boily MC, White RG, Alary M. Risk of HIV-1
transmission for parenteral exposure and blood transfusion: a
systematic review and meta-analysis. AIDS 2006; 20: 805–12.
56 Pinkerton SD. Probability of HIV transmission during acute
infection in Rakai, Uganda. AIDS Behav 2008; 12: 677–84.
57 Hayes RJ, Schulz KF, Plummer FA. The cofactor eff ect of genital
ulcers on the per-exposure risk of HIV transmission in sub-Saharan
Africa. J Trop Med Hyg 1995; 98: 1–8.
58 Fischl MA, Dickinson GM, Scott GB, et al. Evaluation of
heterosexual partners, children, and household contacts of adults
with AIDS. JAMA 1987; 257: 640–44.
59 Halperin DT, Shiboski SC, Palefsky JM, Padian NS. High level of
HIV-1 infection from anal intercourse: a neglected risk factor in
heterosexual AIDS prevention. Proceedings of the XIV
International Conference on AIDS, Barcelona, Spain; July 7–12,
2002 [abstract ThPeC7438].
60 Leynaert B, Downs AM, de Vincenzi I. Heterosexual transmission
of human immunodefi ciency virus: variability of infectivity
throughout the course of infection. Am J Epidemiol 1998; 148: 88–96.
61 Cameron DW, Simonsen JN, D’Costa LJ, et al. Female to male
transmission of human immunodefi ciency virus type 1: risk factors
for seroconversion in men. Lancet 1989; 2: 403–07.
62 Allen S, Tice J, Van de Perre P, et al. Eff ect of serotesting with
counselling on condom use and seroconversion among HIV
discordant couples in Africa. BMJ 1992; 304: 1605–09.
63 Roth DL, Stewart KE, Clay OJ, et al. Sexual practices of HIV
discordant and concordant couples in Rwanda: eff ects of a testing and
counselling programme for men. Int J STD AIDS 2001; 12: 181–88.
64 Satten, GA, Mastro TD, Longini IM Jr. Modelling the female-to-
male per-act HIV transmission probability in an emerging epidemic
in Asia. Stat Med 1994; 13: 2097–106.
65 Mastro TD, Satten GA, Nopkesorn T, et al. Probability of female-to-
male transmission of HIV-1 in Thailand. Lancet 1994; 343: 204–07.
66 Deschamps MM, Pape JW, Hafner A, Johnson WD Jr. Heterosexual
transmission of HIV in Haiti. Ann Intern Med 1996; 125: 324–30.
67 Duerr A, Hsia J, Nicolosi A, et al. Probability of male-to-female
HIV transmission among couples with HIV subtype E and B.
Proceedings of the XIII International Conference on AIDS,
Durban, South Africa; Jul 9–14, 2000 [abstract WePpC1319].
68 Duerr A, Xia Z, Nagachinta T, Tovanabutra S, Tansuhaj A, Nelson K.
Probability of male-to-female HIV transmission among married
couples in Chiang Mai, Thailand. Proceedings of the
X International Conference on AIDS, Yokahama, Japan;
Aug 7–12, 1994 [abstract 105C].
69 Corey L, Wald A, Celum CL, Quinn TC. The eff ects of herpes simplex
virus-2 on HIV-1 acquisition and transmission: a review of two
overlapping epidemics. J Acquir Immune Defi c Syndr 2004; 35: 435–45.
70 Hira SK, Feldblum PJ, Kamanga J, et al. Condom and nonoxynol-9
use and the incidence of HIV infection in serodiscordant couples in
Zambia. Int J STD AIDS 1997; 8: 243–50.
71 Ryder RW, Kamenga C, Jingu M, et al. Pregnancy and HIV-1 incidence
in 178 married couples with discordant HIV-1 serostatus: additional
experience at an HIV-1 counselling centre in the Democratic Republic
of the Congo. Trop Med Int Health 2000; 5: 482–87.
72 Fideli US, Allen SA, Musonda R, et al. Virologic and immunologic
determinants of heterosexual transmission of human
immunodefi ciency virus type 1 in Africa.
AIDS Res Hum Retroviruses 2001; 17: 901–10.
73 Gilbert PB, McKeague IW, Eisen G, et al. Comparison of HIV-1 and
HIV-2 infectivity from a prospective cohort study in Senegal.
Stat Med 2003; 22: 573–93.
74 Donnelly C, Leisenring W, Kanki P, Awerbuch T, Sandberg S.
Comparison of transmission rates of HIV-1 and HIV-2 in a cohort of
prostitutes in Senegal. Bull Math Biol 1993; 55: 731–43.
75 Kimani J, Kaul R, Nagelkerke NJ, et al. Reduced rates of HIV
acquisition during unprotected sex by Kenyan female sex workers
predating population declines in HIV prevalence. AIDS 2008;
76 van der Ende ME, Rothbarth P, Stibbe J. Heterosexual transmission
of HIV by haemophiliacs. BMJ 1988; 297: 1102–03.
77 Kim HC, Raska K, Clemow L, et al. Human immunodefi ciency
virus infection in sexually active wives of infected hemophilic men.
Am J Med 1988; 85: 472–76.
78 Longini IM, Scott Clark W, Haber M, Horsburgh R Jr. The stages of
HIV infection: waiting times and infection transmission probabilities.
In: Castillo-Chavez C, ed. Mathematical and statistical approaches to
AIDS epidemiology. Berlin: Springer-Verlag, 1989: 111–37.
79 Jackson JB, Kwork SY, Hopsicker JS, et al. Absence of HIV-1
infection in antibody negative sexual partners of HIV-1 infected
hemophiliacs. Transfusion 1989; 29: 265–67.
80 Lawrence DN, Jason JM, Holman RC, Heine P, Evatt BL. Sex
practice correlates of human immunodefi ciency virus transmission
and acquired immunodefi ciency syndrome incidence in
heterosexual partners and off spring of U.S. hemophilic men.
Am J Hematol 1989; 30: 68–76.
81 Kim MY, Lagakos SW. Estimating the infecivity of HIV from
partner studies. Ann Epidemiol 1990; 1: 117–28.
82 Wiley JA, Herschkorn SJ, Padian NS. Heterogeneity in the probability
of HIV transmission per sexual contact: the case of male-to-female
transmission in penile-vaginal intercourse. Stat Med 1989; 8: 93–102.
www.thelancet.com/infection Vol 9 February 2009 129 Download full-text
83 Kramer I, Yorke JA, Yorke ED. Modelling non-monogamous
heterosexual transmission of AIDS. Math Comput Model 1990;
84 Siddiqui NS, Brown LS Jr, Phillips RY, Vargas O, Makuch RW.
No seroconversions among steady sex partners of methadone-
maintained HIV-1-seropositive injecting drug users in New York
City. AIDS 1992; 6: 1529–33.
85 Jewell NP, Shiboski SC. Statistical analysis of HIV infectivity based
on partner studies. Biometrics 1990; 46: 1133–50.
86 Jewell NP, Malani HM, Vittinghoff E. Non parametric estimation
for a form of doubly censored-data, with application to two
problems in AIDS. J Am Stat Assoc 1994; 89: 7–19.
87 Downs AM, De Vincenzi I. Probability of heterosexual transmission
of HIV: relationship to the number of unprotected sexual contacts.
J Acquir Immune Defi c Syndr Hum Retrovirol 1996; 11: 388–95.
88 Pedraza MA, del Romero J, Fernando R, et al. Heterosexual
transmission of HIV-1 associated with high plasma viral load
levels and a positive viral isolation in the infected partner.
J Acquir Immune Defi c Syndr Hum Retrovirol 1999; 21: 120–25.
89 Marincovich B, Castilla J, del Romero J, et al. Absence of hepatitis C
virus transmission in a prospective cohort of heterosexual
serodiscordant couples. Sex Transm Infect 2003; 79: 160–62.
90 Nopkesorn T, Mastro TD, Sangkharomya S, et al. HIV-1 infection in
young men in Northern Thailand. AIDS 1993; 7: 1233–39.
91 Kanki P, M’Boup S, Marlink K, et al. Prevalence and risk
determinants of human immunodefi ciency virus type-2 (HIV-2) and
human immunodefi ciency virus type-1 (HIV-1) in west African
female prostitutes. Am J Epidemiol 1992; 36: 895–907.
92 Rakwar J, Lavreys L, Thompson ML, et al. Cofactors for the
acquisition of HIV-1 among heterosexual men: prospective cohort
study of trucking company workers in Kenya. AIDS 1999; 13: 607–14.
93 Nicolosi A, Corrêa Leite ML, Musicco M, et al. The effi ciency of
male-to-female and female-to-male sexual transmission of the
human immunodefi ciency virus: a study of 730 stable couples.
Epidemiology 1994; 5: 570–75.
94 Weiss HA, Quigley MA, Hayes RJ. Male circumcision and risk of
HIV infection in sub-Saharan Africa: a systematic review and
meta-analysis. AIDS 2000; 14: 2361–70.
95 DeGruttola V, Seage GR 3rd, Mayer KH, Horsburgh CR Jr.
Infectiousness of HIV between male homosexual partners.
J Clin Epidemiol 1989; 42: 849–56.
96 Vittinghoff E, Douglas J, Judson F, et al. Per-contact risk of human
immunodefi ciency virus transmission between male sexual
partners. Am J Epidemiol 1999; 150: 306–11.
97 Seage GR, Mayer KH, Horsburgh CJR. Risk of human
immunodefi ciency virus infection from unprotected receptive anal
intercourse increases with decline in immunologic status of
infected partners. Am J Epidemiol 1993; 137: 899–908.
98 Siegfried N, Muller M, Volmink J, et al. Male circumcision for
prevention of heterosexual acquisition of HIV in men.
Cochrane Database Syst Rev 2003; 3: CD003362.
99 Auvert B, Taljaard D, Lagarde E, et al. Randomized, controlled
intervention trial of male circumcision for reduction of HIV
infection risk: the ANRS 1265 trial. PLoS Med 2005; 2: e298.
100 Gray RH, Kigozi G, Serwadda D, et al. Male circumcision for HIV
prevention in men in Rakai, Uganda: a randomised trial. Lancet
2007; 369: 657–66.
101 Bailey RC, Moses S, Parker CB, et al. Male circumcision for HIV
prevention in young men in Kisumu, Kenya: a randomised
controlled trial. Lancet 2007; 369: 643–56.
102 Rottingen JA, Cameron DW, Garnett GP. A systematic review of the
epidemiologic interactions between classic sexually transmitted
diseases and HIV: how much really is known? Sex Transm Dis 2001;
103 Laurent C, Seck K, Coumba N, et al. Prevalence of HIV and other
sexually transmitted infections, and risk beahviours in unregistered
sex workers in Dakar, Senegal. AIDS 2003; 17: 1811–16.
104 Meda N, Ndoye I, M’Boup S, et al. Low and stable HIV infection
rates in Senegal: natural course of the epidemic or evidence for
success of prevention? AIDS 1999; 13: 1397–405.
105 Drain PK, Halperin DT, Hughes JP, et al. Male circumcision,
religion, and infectious diseases: an ecologic analysis of
118 developing countries. BMC Infect Dis 2006; 6: 172.
106 Limpakarnjanarat K, Mastro TD, Yindeeyoungyeon W, et al. STDs
in female prostitues in northern Thailand. Proceedings of the IX
International Conference on AIDS, Berlin, Gernany; June 6–11,
1993 [abstract PO-C10-2820].
107 Ragni MV, Faruki H, Kingsley LA. Heterosexual HIV-1 transmission
and viral load in hemophilic patients. J Acquir Immune Defi c Syndr
Hum Retrovirol 1998; 17: 42–45.
108 Lee TH, Sakahara N, Fiebig E, et al. Correlation of HIV-1 RNA levels
in plasma and heterosexual transmission of HIV-1 from infected
transfusion recipients. J Acquir Immune Defi c Syndr Hum Retrovirol
1996; 12: 427–28.
109 Pedraza MA, del Romero J, Roldán F, et al. Heterosexual
transmission of HIV-1 is associated with high plasma viral load
levels and a positive viral isolation in an infected partner.
J Acquir Immune Defi c Syndr Hum Retrovirol 1999; 21: 120–25.
110 Osmond D, Bacchetti P, Chaisson RE, et al. Time of exposure and
risk of HIV infection in homosexual partners of men with AIDS.
Am J Public Health 1988; 78: 944–48.
111 Hollingsworth TD, Anderson RM, Fraser C. HIV-1 transmission,
by stage of infection. J Infect Dis 2008; 198: 687–93.
112 Allen S, Meinzen-Derr J, Kautzman M, et al. Sexual behaviour of
HIV discordant couples after HIV counselling and testing. AIDS
2003; 17: 733–40.
113 Guest G, Bunce A, Johnson L, et al. Fear, hope and social
desirability bias among women at high risk for HIV in West Africa.
J Fam Plann Reprod Health Care 2005; 31: 285–87.
114 White RG, Cooper BS, Kedhar A, et al. Quantifying HIV-1
transmission due to contaminated injections.
Proc Natl Acad Sci USA 2007; 104: 9794–99.
115 Gisselquist D. How much do blood exposures contribute to HIV
prevalence in female sex workers in sub-Saharan Africa, Thailand
and India? Int J STD AIDS 2007; 18: 581–88.
116 Gisselquist D, Potterat J. Questioning Wawer et al’s estimated rate
of sexual transmission from persons with early HIV infection.
J Infect Dis 2005; 192: 1497–98.
117 Wawer M, Serwadda D, Quinn TC, et al. Questioning Wawer et al’s
estimated rate of sexual transmission from persons with early HIV
infection. Reply to Gisselquist and Potterat. J Infect Dis 2005;
118 Fleming DT, Wasserheit JN. From epidemiological synergy to public
health policy and practice: the contribution of other sexually
transmitted diseases to sexual transmission of HIV infection.
Sex Transm Infect 1999; 75: 3–17.
119 Weiss HA. Male circumcision as a preventive measure against HIV
and other sexually transmitted diseases. Curr Opin Infect Dis 2007;
120 Seed J, Allen S, Mertens T, et al. Male circumcision, sexually
transmitted disease, and risk of HIV. J Acquir Immune
Defi c Syndr Hum Retrovirol 1995; 8: 83–90.
121 Pilcher CD, Price MA, Hoff man IF, et al. Amplifi ed transmission of
HIV-1: comparison of HIV-1 concentrations in semen and blood
during acute and chronic infection. AIDS 2007; 21: 1723–30.
122 Paz-Bailey G, Rahman M, Chen C, et al. Changes in the etiology of
sexually transmitted diseases in Botswana between 1993 and 2002:
implications for the clinical management of genital ulcer disease.
Clin Infect Dis 2005; 41: 1304–12.
123 Kreiss J, Ngugi E, Holmes K. Effi cacy of nonoxynol 9 contraceptive
sponge use in preventing heterosexual acquisition of HIV in
Nairobi prostitutes. JAMA 1992; 268: 477–82.
124 Bird KD. The use of spermicide containing nonoxynol-9 in the
prevention of HIV infection. AIDS 1991; 5: 791–96.
125 Nielsen NO, Friis H, Magnussen P, et al. Co-infection with
subclinical HIV and Wuchereria bancrofti, and the role of malaria
and hookworms, in adult Tanzanians: infection intensities, CD4/
CD8 counts and cytokine responses. Trans R Soc Trop Med Hyg
2007; 101: 602–12.