You don't need to be signed in to read BMJ Group Blogs, but you can register here to receive updates about other BMJ Group products and services via our Group site.


Partners PrEP sub-study finds no evidence that PrEP use is associated with risk compensation behaviour

21 Nov, 13 | by Leslie Goode, Blogmaster

How useful is pre-exposure prophylaxis (PrEP)?  The Partners PrEP randomized control study of daily pre-exposure prophylaxis among HIV-uninfected partners of heterosexual HIV-discordant couples in Uganda and Kenya has indicated that, given adequate adherence, PrEP has high biologic efficacy.  The study itself (Baeten & Celum) demonstrates levels of risk reduction of 75%; while a spin-off sub-study from the original trial, monitoring adherence (STI blog/Haberer & Bangsberg ), has established that, with high adherence (c. 97%), levels of protection are even higher than the study might suggest (none of the 14/1,147 sub-study participants who sero-converted were from the intervention arm of the study).

These results have fed into recent attempts to model the likely effectiveness of PrEP.  The consensus hitherto seems to be that PrEP is a relatively high-cost intervention most likely to be cost-effective as an addition to ART in countries where the burden of HIV is high, and rates of male circumcision low – such as southern Africa (Verguet & WalshYing and Barnabas).

Last month saw the online publication of a second interesting spin-off sub-study of Partners PrEP (Mugwanya & Baeten).  It undertakes a longitudinal analysis of data from the original trial to address what is perhaps the greatest concern affecting the implementation of PrEP as a public health intervention (STI blog/Sugarman & Mayer), that of risk compensation – i.e. the possibility that the security promised by PrEP will itself encourage sexual risk-taking behaviour.  M&B’s results, along with those of any future studies of this issue, will no doubt serve to inform the assumptions of future models.

In July 2011 the Partners PrEP study reported its findings, and the placebo arm of the trial was closed.  However, follow-up of 3,024 participants continued.  This allowed data on their sexual activity to be collected over a period which spanned the 12 months preceding – and the 12 months following – the disclosure to participants of the results of the study.  These data included unprotected sex acts and total sex acts over this period, both within the primary relationship as registered by the study and outside that relationship.

Within the primary relationship (i.e. with the registered HIV-infected partner), the crude average frequency of unprotected/total sex acts within the primary relationship was 59/414 per 100 person- months prior to, and 53/361 following, unmasking.  Outside that primary relationship the frequency of unprotected total sex acts was 49/62 as opposed to 67/84. So there appears to be a fall in unprotected/total sex acts with the primary sex partner as between the two halves of the period, and a small, but significant, rise in unprotected/total sex acts outside this relationship.  The latter rise (unprotected sex acts outside the primary relationship) is helpfully quantified at an average of 6.8 sex acts per year following unmasking as against an average of 6.2 acts in a predicted counterfactual scenario had patients remained unmasked.  M&B place this figure in the context of the estimated doubling of risk-behaviour, which modellers suggest would be required in order to see any attenuation in the effectiveness of PrEP.

The authors suggest – optimistically perhaps – that the small decrease in unprotected sex following enrolment indicates that PrEP delivered in the context of an HIV prevention package may be synergistic for risk reduction.  However, they also point to an earlier study (Ndase & Thomas) that finds this overall pattern of decline in sex acts with primary partner and rise in sex acts outside the primary relationship to be indicative of dissolution of the primary relationship and formation of new relationships.  They also observe that unprotected sex without outside partners was “high among the few participants who reported sex outside the primary partnership” – an observation that accords with recent study findings that a quarter of HIV infections in sero-discordant partnerships arise from non-primary partners.

The findings of the study seem reassuring.  They raise two questions, however.  The first concerns the impact of the probable gap between the HIV prevention package accompanying any real-life PrEP implementation and the package made available to the participants in Partners-PrEP.  The second concerns the impact of new partnership formation on unprotected sex over the longer term.  It is here that the findings of M&B would lead us to expect the greatest challenge to PrEP; yet the true extent of such a challenge is something of which their study can offer us only the most preliminary impression.

Are bisexuals well served by HIV interventions that assume gay identity?

30 Oct, 13 | by Leslie Goode, Blogmaster

Studies published in STI journal have examined the impact of bisexual concurrency on HIV epidemiology in South Africa (Behrer & Baral) (B&B) and China (Yun & Shang) (Y&S), where it is reckoned at 53.7% and 31.2% of the MSM population, respectively.  However, a recent randomized control study of an educational intervention in Los Angeles (Harawa & Cunningham) claims to be among the first studies addressing bisexual concurrency among black MSM in the US.

US black MSM who have sex with women (MSMW) are less likely than white MSM to identify gay, and more concerned to fulfil traditional gender expectations.  Consequently, say the authors, they may be less well served by interventions based on contemporary conceptualizations of sexual behaviour in terms of fixed sexual identities.  This raises a number of interesting questions, among them, how far the situation of US black MSM resembles that of MSM in traditional societies (e.g. South Africa or China), and how far influential contemporary conceptualizations about sexual identity, based on the cultural experience of white MSM, constitute an appropriate model for interventions outside that specific social context.   The authors see potential benefits for MSMW of interventions based on more fluid and context-dependent models of sexual behaviour.

The intervention that is the object of this study – Men of African American Legacy Empowering Self (MAALES) – is an HIV education and risk-reduction course addressed specifically to the needs of this group (Williams & Harawa).  It consists in six two-hour sessions delivered over a three week period (with booster sessions at six and 18 weeks), and aims, above all, to be “culturally congruent”.  It is conducted by black MSM facilitators, and its content is theoretically grounded in a teaching model developed in African American communities (the critical thinking model), as well as in reasoned action theory and empowerment theory.

The study itself compares sexual behaviours, at base-line and three and six months after the intervention, of 437 black MSMW randomly assigned to either the MAALES intervention or a twenty-minute HIV education and risk-reduction session based on a standard HIV test counselling approach. Adjusted results indicate the achievement of significantly less unprotected sex acts with male or female over prior ninety days at six months in the intervention arm as against the control arm (RR 0.61), significantly less unprotected sex acts with females (RR 0.5), and a near-significant reduction in sex acts with males (RR 0.63).  Given the time and resources dedicated to the MAALES as compared to the control intervention, one wonders how much of this behavioural modification is owed to the greater investment in the patients in the intervention arm of the study, and how much to the superiority of its culturally congruent methodology over the conventional alternative.

In their conclusions Harawa and Cunningham tend to corroborate the emphasis of Behrer and Baral, Yun and Shang and others on the influence of societal and cultural factors.  There have even been attempts to quantify the impact of psychosocial constructs on MSM sexual behaviour (Konda & Kegeles).  Yet, Harawa and Cunningham differ somewhat from these other studies in the greater stress they place on the role of participants as responsible agents.  They also recommend on the basis of the more frequent reporting of unprotected sex with females in their study, and the relatively low levels of sex with males, that interventions aiming to “responsibilise” MSMW should prioritize the reduction of risky behaviour which involves females, rather than males.

Was the “sexual revolution” triggered by the decline of syphilis?

26 Mar, 13 | by Leslie Goode, Blogmaster

The year 1939 saw total US syphilis deaths at 15 per 100,000 and syphilis deaths of black males at 72.5 per 100,000: this is a death rate comparable to that for HIV/AIDS at the height of the epidemic in 1995 when total deaths and deaths of black males stood, respectively, at 16.2 and 80.2 per 100,000.  Subsequently, in the late 40s and early 50s, incidence and mortality from syphilis were to fall precipitously – thanks to penicillin.  A recent paper in the Archives for Sexual Behavior by an economist, Andrew Francis, argues for the importance of this collapse in the “cost” of syphilis in spurring the sexual revolution ( His exploration of this hypothesis prompts general reflection on the link between the “cost” of disease (which he equates with absence of an effective treatment) and sexual behaviour.

The paper correlates data across US states on syphilis incidence and mortality with measures of “risky non-traditional sex” – which, in the context of the poverty of relevant data for the period, is evaluated on the basis of gonorrhoea rate, illegitimate birth ratio and teen birth share.  Coefficients are given from regressions of measures of sexual behaviour on indicators for the number of years since syphilis collapse (which varies by state).  As regards illegitimate birth ratio and teen birth share, a positive correlation emerges which goes back as early as three years or less from syphilis collapse; gonorrhoea, however, continues to decline from its WW2 peak, as we might expect.

Ultimately, however, Francis’s argument rests also on the claim that his own explanation of sexual revolution fits the facts better than the alternative explanations, of which the most familiar is that “anti-conception technology”, especially the pill, played a decisive role.  Francis admits that, from the late sixties, conception technologies and measures of risky behaviour increase simultaneously.  Yet measures of risky sexual behaviour, he claims, had already been rising sharply for a decade before anti-conception technology began to make its impact.  His own data show that the two changes do not coincide.   Nor do measures of permissive values or religious observance show any discontinuous change that would coincide with the increase in risky behaviour.  So if Francis’ hypothesis is wrong, then the precise timing of the sexually revolution remains mysterious.  Francis gestures vaguely towards the possibility of the change in sexual behaviour being triggered by still less definable “economic, social and cultural changes”.

The obvious recent parallel to the syphilis collapse in our own times would be impact of the introduction of HAART on HIV/AIDS mortality.  The concern that an effective therapy might lead to disinhibition has been widely discussed in STI journal and elsewhere.  An increase in risk-taking behaviour among MSM during the late 1990s has been established: (Stolte & Coutinho: Amsterdam); (Elford & Hart: London); & Allard: Montreal); (Stephensen & Williams: London).  But opinions on the contribution of HAART to changing sexual behaviour seem divided, with Stolte & Coutinho supporting the hypothesis, and the others inclined to attribute it to other factors.  The probable future emergence of multi-resistant gonorrhoea ( represents the inverse case (loss of a therapy leading to potential inhibition).  If the diminished “cost” of an STI (syphilis) can spur an increase in risky sexual behaviour, should we not expect an increased cost to have the opposite result?  Time will tell.

Does Muslim religion have an impact on HIV transmission?

7 Nov, 12 | by Leslie Goode, Blogmaster

How is it that the nations of the Middle East and North Africa appear to have relatively low rates of HIV (see A recent paper in the American Sociological Review – Adamczyk & Hayes  – seeks to demonstrate that the predominance of the Muslim religion in a country may have a macro-level, cultural impact on the sexual behaviour (pre- and extra-marital sex) of its residents.  What do they mean by qualifying religious impact as, to some extent, macro-level and cultural?   They are claiming that the impact of religion is not reducible either to individual affiliation, or to formal restrictions.  In their words, it is, in their words sui generis – unique of its kind.  Behind this is a Durkheimian understanding of the religious group as something more than the sum of its parts (

To test their hypothesis the authors use data from the Demographic and Health Survey (DHS).  Their most interesting findings relate to pre-marital sex.  Muslims are far less likely to report pre-marital sex than Christians/Jews (0.61: 0.77).  This effect is not explained wholly by age of first marriage.  As the proportion of Muslims in a nation increases likelihood of all residents (including non-Muslims) reporting pre-marital sex decreases substantially.  So, for a non-working rural woman, likelihood of premarital sex  is 0.71 where 1% residents are Muslim, 0.61 where 23% are Muslim, 0.28 where 90% are Muslim.  Muslims, however, are not more likely to report pre-marital sex as the proportion of Muslims decreases.  Interestingly the relationship between likelihood of pre-marital sex and percentage Muslim does not seem to be mediated by formal restrictions (exemplified here by restrictions on women’s mobility).

The tentative conclusion of the study is, therefore, that the striking relationship of pre-marital sex and percentage Muslim may not be a matter of individual affiliation (or Muslims would be more likely to report pre-marital sex where percentage Muslim decreases).  It may also not be a matter of formal restriction, for in that case the relationship of pre-maritcal sex to percentage Muslim would be mediated by women’s mobility).  Hence the relationship is likely to be explained by something other than individual affiliation, nor formal restriction – i.e. probably macro-level cultural effect.

To non-sociologists this argument from premises to conclusion might seem tenuous.  However hypothesis of a relationship between religion and behaviour is interesting, and the authors claim that this is the first serious attempt to test such a hypothesis on the basis of international data. The interest of the paper for STI journal readers may go beyond the intriguing question of why certain populations should be less susceptible to HIV epidemics than others. From the angle of sexual health policy, the claims of Adamczyk & Hayes matter because they suggest the existence of other potentially refractory cultural factors (i.e. religion in the sense of Adamczyk & Hayes ) – a “black box”, if you like – that, depending on the policy, could turn out to have negative as well as positive effects. On the face of it, this seems at least plausible, and, if it is the case, would certainly be important for health policy. To the extent it militates against “one-size-fits-all” approaches to health policy, the argument of Adamczyk & Hayes may be additional ammunition in the “armentarium” of the “programme science” based approach  that is a topic of ongoing comment in this journal (see for example Aral & Blanchard: STI journal readers will be interested to know that STI journal continues to run a series on Programme Science, edited by Dr Sevgi Aral.


Impact on sexual behaviour of “Don’t Ask, Don’t Tell” policy in US navy

6 Nov, 12 | by Leslie Goode, Blogmaster

Epidemiological research has sometimes addressed the impact on men who have sex with men (MSM) sexual behaviour of being “non-gay identifying” (NGI) (Yun, Wang et al. ( ); Mercer & Cassell ( or of belonging to a culture in which openness about sexuality by MSM is sometimes difficult and personally costly (Lane, Kegeles et al. (

The position of MSM in the US military under the recently abolished “Don’t Ask, Don’t Tell” (DADT) policy was an extreme case – possibly a limit-case – of this situation:  up until September 2011, an admission of sexual orientation by MSM, or evidence such as hand-holding, could result in ejection from the military.

How does this kind of situation influence patterns of HIV transmission among MSM, and what is the impact on the sexual behaviour of those MSM who engage in relationships with men regardless?  Would we expect the repressive effects of DADT to result in a relatively lower proportion of total HIV infection due to MSM sexual contact than in the general population – or the reverse?

Results of a recent online survey of the sexual behaviour prior to forced HIV testing of  US Navy and Marine Corps personnel who sero-converted under DADT intriguingly lifts the lid on this formerly closed epidemiological world – and perhaps, to some degree, on other similarly closed worlds.  Of course, the survey itself has major limitations: most importantly its restriction to a minority (64 (524): 26%), and an apparently not very representative minority, who responded to the survey; also the often considerable lapse of time since the behaviours reported.  Despite all these limitations the forced imposition of an HIV test on the whole group allows the survey to capture HIV prevalence at a moment in time.

Among the men who became HIV-infected  84% had had sex with men in the 3-year period prior to diagnosis: 55% reporting sex with just men, and 30% reporting sex with both men and women.  DADT would not then appear to have had much impact on reducing the burden of MSM infection as a proportion of total burden.  This higher figure relative to earlier surveys of the US military (84% as opposed to 59% reporting sex with men) probably reflects the liberalizing effects of DADT repeal.  The frequency of inconsistent condom use with anal sex was 65%, and more than three quarters expressed surprise at their HIV diagnosis.

The story these figures tell mirrors other “repressive” settings such as those with which we began our blog.  On the whole, a culture of repression drives the unwelcome sexual behaviour underground rather than eliminating it, while, at the same time, discouraging responsible behaviour and the adoption of risk-reduction strategies.  As the authors note,  “Several opportunities for primary prevention messaging now possible after DADT repeal are evident”.

Non-disclosure of HIV sero-status by Indian female sex workers

25 Sep, 12 | by Leslie Goode, Blogmaster

With the roll-out of the Bill and Melinda Gates initiated Avahan interventions in India over the last decade, a growing body of evidence has accumulated on the contribution of commercial sex-work to the spread of the HIV epidemic, and the effectiveness of behavioural interventions focussed on this sector.  With the international effort concentrating elsewhere primarily on the deployment of ART, the epidemiological and preventative emphasis of Avahan has broken new ground: attention has been increasingly focussed on influencing the behaviour of key populations ( ).  Commercial male and female sex workers (FSW/MSW) seem to be an important one of these in the Indian context, due to the importance of male sex activity outside marriage as a factor in Indian HIV epidemiology (  One neglected aspect of this behaviour has apparently been the disclosure of HIV status.  This is a gap in the research that Saggurti, Samat et al. seek to fill a recent paper (

Their headline finding is that 58% of the 211 women (FSM) surveyed by the study, and 41% of the 205 men (clients) had not disclosed their sero-status to any sexual partner.  This would seem to have implications that are more interesting for the relationships of FSM to their non-commercial partners than to their commercial ones.  After all, non-disclosure between FSM and clients is surely what we would expect.  However, non-disclosure in the context of non-commercial and married partners, where consistent condom use may not be the norm ( indicates the potential vulnerability of “bridging populations”.  Here Saggurti et al. give further data on the 18 married women and 76 married men included in the study:  39% of the women and 1% of the men had not disclosed HIV status to their married partner; while 78% of the women and 36% of the men claimed not to know the sero-status of the partner.  These findings suggest a marked imbalance – though an imbalance that might reflect behavioural differences between sex-workers and sex-workers’ clients, rather than gender differences per se.

Literature on the epidemiology of HIV transmission stresses the importance of “bridging populations” – which in the case of FSM presumably includes the wives of male clients, on the one hand, and the non-commercial partners of the FSM themselves, on the other.  As regards the former group, a recent study of FSW in Karnataka (Shaw, Deering, Blanchard et al., 2011) indicates that clients of FSW with NCP are less likely (OR 1.8) than those without NCP to use a condom, and more likely (OR 1.5) to be infected with HSV-2.  ( But with this group the findings of Saggurti et al. suggest high levels of disclosure.  The case is different with the NCP of FSW, who represent the other bridging group (though here the bridge could well be from NCP to FSW rather than from FSW to NCP).  One of the few studies of the latter group, also from Karnataka (Deering, Bhattacharjee, Alary et al., 2011) finds that levels of consistent condom use (CCU) of FSW with husbands and cohabiting partners is low (22.6% and 40.3% respectively) (  Thus, here, low disclosure would appear to go with low CCU.  An interesting recent study of FSW and their NCP in Vietnam that would seem to bear out this general picture, Hoffman & Niccolai 2011, examines the relationship between intra-couple “communication divergence” and low CCU, and discovers a relatively strong correlation of OR 0.54 (

HIV testing gives rise to behaviour change

10 May, 12 | by Leslie Goode, Blogmaster

What sort of impact does HIV testing (Voluntary Testing and Counselling: VCT) have on an individual’s subsequent sexual behaviour?

A number of studies of VCT clients in sub-Saharan Africa, and published in STI journal, suggest a significant preventative effect.  A two-year cohort study of 401 clients of Kenyan government health centres (Arthur & Gilks, 2007) points to a significant reduction at 7.5 months both in the proportion of clients with multiple partners (from 16% to 6%) and in the numbers reporting symptoms of STIs (16% to 6%).  A more recent study of attendees at health-centre affiliated VCTs in two Zambian mining villages involving serial interviews (Sikasote & Murray, 2011) serve to consolidate pre-test decisions concerning sexual risk behaviour and reflect clients’ desire to “regain control of their lives”.

In the view of the support these studies lend to the preventative aspect of VCT, it is encouraging to note the comparable behavioural changes reported in a recent study of a very different client group of 1,038 cocaine and heroin users taking up base-line VCT at a US Emergency Department (Bernstein and Madico, 2012).  The study involved secondary analysis of data from a randomized control trial of a motivational intervention (ineffective, as it turned out), which included behaviour change at six and twelve months, notably: a sustained decrease in the proportion of unprotected sex acts (OR 0.7 at 6 months; 0.69 at 12 months), and a sustained drop off – from a high base-line, admittedly – in the number of sex acts (OR 3.1; 0.25).  Given the screening took place during visits to an ED, this effect cannot, as in the African cases, be attributed to a pre-test decision.

These recent findings seem to endorse once again, albeit in the relation to a very different client group, the preventative impact of VCT.  It is interesting, in this regard, that the Zambian study (Sikasote and Murray, 2011) raises the issue of VCT impact for those testing negative, and the importance of post-test support for this group.  They even recommend the de-linking of counselling from testing to encourage earlier attendance.

G. Arthur, C. Gilks et al., “Sexual behaviour change in clients of health centre-based voluntary HIV counselling and testing services in Kenya”, Sexually Transmitted Infections 2007: 83;7

[Abstract][Full text][PDF]

Janet Sikasote, Scott A. Murray et al., “Voluntary counselling and testing for HIV in a Zambian mining community: serial interviews with people testing negative”, Sexually Transmitted Infections 2011:87;5

[Abstract][Full text][PDF]

Edward Bernstein, Guillermo Madico et al., “Long-term Follow-up After Voluntary Human Immunodeficiency Virus/Sexually Transmitted Infection Counseling, Point-of-service Testing, and Referral to Substance Abuse Treatment From the Emergency Department, Academic Emergency Medicine, Volume 19, Issue 4, April 2012;jsessionid=7D68FDAA3B7B33CFC869B379642BA484.d03t02

How infectious are HIV+ MSM on ART?

27 Apr, 12 | by Leslie Goode, Blogmaster

How effective is HAART (highly active antiretroviral therapy) in preventing HIV transmission among MSM (men who have sex with men)?  The recent HPTN (HIV Prevention Trials Network) 052 clinical trial demonstrated, to great acclaim, that with heterosexual sero-discordant couples early initiation of HAART is associated with 96% reduction in HIV transmission.  Yet there are reasons why we might expect that for MSM the preventive effect would be less.  We know for example that rectal intercourse is an especially effective route of HIV transmission due to the thinness of the rectal epithelial membrane; high prevalence of urethritis and other STDs in this group could also increase vulnerability.  There is little or no evidence, however.  Given this ignorance, the possibility that HIV- infected MSM may be using HAART status in their sexual decision-making is a matter of concern.  How important is it that sexually-active HIV-infected MSM should use condoms and other risk-reduction strategies through all stages of HIV disease regardless of HIV treatment status?

A US study (Politch, Anderson et al.) published in AIDS has attempted to cast some light on this difficult area, by assessing HIV levels in paired semen and blood samples from 101 HIV-infected MSM who had been on stable ART regimens for at least 3 months.  Given the impracticability of undertaking anything like HPTN 052 on an MSM population, prevalence of HIV viral shedding in blood may offer the best approach to determining the relative infectivity of this group.

Of the 83 men with undetectable HIV in blood plasma, 25% proved to have HIV in semen with copy numbers ranging from 80-2,560 (median 200), as against 50% of those with HIV in blood plasma.  Large recent studies of blood plasma in HIV positive men on HAART (not MSM) have reported levels of viral shedding in semen of 2-3%.  In attempting to quantify the real infectivity risk associated with this higher level of compartmentalized shedding, the authors refer to a theoretical paper (Chakraborty, Eron et al., “Viral burden in genital secretions determines male-to-female sexual transmission of HIV-1: a probabilistic empiric model”, AIDS 2001:15:621-627) and a clinical study (Baeten, Nakku-Joloba, et al., “Genital HIV-1RNA Predicts Risk of Heterosexual HIV-1 Transmission”, Sci Transl Med 2011:3:77ra29) which both conclude that <1000 copies of HIV RNA pose a low but real risk of male-to-female HIV transmission.  Thus, reckoning on a five-fold reduction in this copy number for rectal transmission among MSM, our authors place the threshold of significant risk at <200 – which is well within the range of values reported in their own study.

All this may seem speculative.  In the absence of any more reassuring evidence, however, readers will no doubt agree that “it would be prudent to advise sexually active HIV-infected MSM to use condoms, … regardless of HIV treatment status”.

Joseph A. Politch, Deborah J. Anderson et al., “Highly active antiretroviral therapy does not completely suppress HIV in semen of sexually active HIV-infected men who have sex with men”, AIDS 2012:26

Papers on related issues published in STI journal:

For a sceptical view of the influence of perceptions of ART status on sexual risk taking, see:

J. Cox, J. Beauchemin, and R. Allard, “HIV status of sexual partners is more important than antiretroviral treatment related perceptions for risk taking by HIV positive MSM in Montreal, Canada”, 2004:80:518-523



The real-life STI prevention potential of the male condom: hard to fulfil, hard to evaluate

23 Mar, 12 | by Leslie Goode, Blogmaster

In the face of 340 million incident cases of STI worldwide each year, UNAIDS in a recent statement deemed the male latex condom “… the single most efficient, available technology to reduce the sexual transmission of HIV and other sexually transmitted infections”.  Sadly, though the condom may be effective in vitro, this potential proves hard to realize use in real life.  Sustained condom use requires levels of commitment such that, even in the case of discordant heterosexual HIV couples, studies indicate fewer than half of participants report regular use.  In the light of this, the Australian journal Sexual Health has devoted a special issue to condom use – recognizing it to be a “complex behaviour embedded in the fabric” of social relationships.  Attention is given both to data regarding the behaviour itself (including errors and problems of condom use), and to the methodological problems affecting its evaluation by recent research.

Individual reviews cover China, Central and Eastern Europe and Sub-Saharan African.  There are also papers on the female condom and on the issue of “risk compensation” – i.e. where the introduction of one preventative method (e.g. vaginal microbicides) impacts on the use of another (e.g. condoms).

Two systematic review papers, reflecting the two-fold concern of this special issue with the behaviour itself and methodological issues around its evaluation, offer a global perspective on condom use.  The first of these – a review of the literature regarding condom use errors and problems (Stephanie A. Sanders et al.) – leaves the reader with a bewildering sense of the disparity in the importance assigned by studies to the various causes of “condom failure”.  Statements abound such as that “breakage rates ranged from 0.8% to 40.7% of participants across 15 studies”.  The reader may wonder what there is to learn from such data, other than that condom use is indeed a complex behaviour and difficult to evaluate scientifically.

The other systematic review paper (Richard A. Crosby & Sarah Bounse) is complementary to this.  It deals specifically with the methodological problems of evaluating the link between condom use and STI, focussing exclusively on prospective studies.  Forms of misclassification bias are rife, the authors claim.  The most serious derive from the difficulty of determining, when infection takes place within the recall period, whether events of condom-protected sex occurred before infection, or after infection.  Other sources of misclassification bias involve failing, where figures are given for number of infections for a given level of condom use, to control for “use errors” such as breakage, slipping, or incomplete use.

Ultimately the critical question facing policy formers is this: whether protective effect of condom use warrants the full support of public health efforts to keep condoms to the forefront of STI prevention?  It should be borne in mind that all the potential forms of study bias regarding condom use tend towards the nul hypothesis (i.e. they underestimate the effectiveness of condoms).  In the light of this, the authors affirm that the threshold for sufficient protective effect has probably already been crossed.  Yet clearly more definitive findings would greatly assist the cause of deploying condoms as part of public health efforts.  To this end, the paper seeks to explain, and thereby avert, the causes of “error variance” between studies of condom effectiveness (variance not determined by real differences of the relation between condom use and STI).

Sexual Health 9: 1, 2012

Stephanie A. Sanders, Robin R. Milhausen et al., Condom Use Errors and Problems: A Global View, pp.81-95

Richard A. Crosby and Sarah Bounse, Condom Effectiveness: Where are we Now, pp.10-17



R. Crosby, W.L. Yarber et al.,  ”Two heads are better than one: the association between condom decision-making andcondom use errors and problems”, Sex Transm Infect 2008;84:198-201 doi:10.1136/sti.2007.027755

And for a perspective on promoting condoms through religious leadership, Willms et al in Sex Transm Infect 2011;87:611-615 doi:10.1136/sextrans-2011-050045





How does the experience of sexual coercion influence subsequent sexual behaviour?

11 Oct, 11 | by Leslie Goode, Blogmaster

Does the experience of sexual coercion predispose the sufferer to the kind of sexual behaviour likely to render him/her more vulnerable to HIV?  Knowing the extent and the mechanisms of such influence could be helpful to those planning public health interventions in high risk populations.  How important to HIV prevention is it to modify coercive behaviour?  Of course, answers to such questions are likely to differ from one culture to another.

Results of a recent study on 1,200 university students at Mbarara University Uganda, based on a self-administered questionnaire, has established, for this setting at least, a strong association between coercion, measured on a validated scale of six items, and 3 out of 4 “high-risk” sexual behaviours.  Significant odds ratios (OR), adjusted for various social and other confounders, are given for: (1.) having previously had sex: OR 1.6; (2.) early sexual debut (≤18 years): OR 2.4; (3.) having had ≥3 sexual partners: OR 1.9.  There was no significant association with the fourth type of high-risk behaviour: inconsistent condom use.  The results obtained also indicate a significant correlation between coercion and 3 social factors: namely, religious background, mental health scores and capacity to trust.  The authors conclude that religion, high trust in others, and good mental health help to buffer the negative effect of sexual coercion.

These results will no doubt be of interest to anyone planning HIV interventions in settings that could be considered culturally comparable to those of a Ugandan university.   However, the study seems to fall somewhat short of demonstrating causal mechanisms, even for the setting of the study – something that the authors themselves acknowledge would be hard to achieve.

One particularly interesting aspect of this study – which, according to its authors, marks it out from previous work – is that males report coercion almost at the same levels as females (29% as against 31%), and most of the associations between coercion and high-risk behaviour (as well as the buffering effects of social and religious factors) apply, in varying measure, not just to females but to both sexes.  It will be interesting to see how far this is borne out by future studies in this area.

Agardh A, Odberg-Pettersson K, Östergren, “Experience of sexual coercion and risky sexual behavior among Ugandan university students”, BMC Public Health 2011, 11:527 (4 July 2011)

Sexually Transmitted Infections blog

Sexually Transmitted Infections

Discussion and suggestion space for readers of STIs.
Visit site

Latest from Sexually Transmitted Infections

Latest from Sexually Transmitted Infections