Reductions in HIV transmission risk behaviour following diagnosis of primary HIV infection: a cohort of high-risk men who have sex with men


Dr Julie Fox, Clinical Trials Centre, Winston Churchill Building, Imperial College London, St Mary's Hospital Campus, London W2 1PG, UK. Tel: +44 2078861466; fax: +44 2078866123; e-mail:



Risk-reduction counselling is a standard preventive intervention, but behaviour change is difficult to sustain over the duration of HIV infection. However, primary HIV infection (PHI) is highly infectious and plays a key role in transmission – especially through dense sexual networks – but is short term, so even transient risk reduction can mitigate its high infectivity. Targeting behaviour-change interventions at recently infected individuals may be highly effective, particularly in higher risk groups. We explored the potential impact on HIV transmission-risk behaviour of PHI diagnosis in men who have sex with men (MSM).


MSM with PHI were interviewed at diagnosis and after 3 months of follow-up about their sexual behaviour in the 12-week periods before and after diagnosis and standard counselling.


A total of 98 of 104 eligible MSM (94%) participated in the study, with 100% follow-up. PHI was associated with high levels of recreational drug use, low levels of condom use, high numbers of sexual partners and a history of sex work. In the 12 weeks post-diagnosis, 76% of participants eliminated risk of onward transmission entirely and, overall, there was a significant reduction in transmission-risk behaviour, with patients reporting greater condom use and fewer sexual partners. Those with continued transmission-risk behaviour were more likely to have another sexually transmitted infection (STI), use ketamine and have more sexual partners at baseline.


Most MSM recently diagnosed with PHI changed their behaviour to substantially reduce the risk of onward HIV transmission. Strategies are needed to (a) increase diagnoses of PHI to target prevention efforts effectively and (b) further reduce risk behaviours by targeting enhanced counselling to those most likely to continue with risk behaviours.


HIV transmission rates in men who have sex with men (MSM) remain high in the United States of America, the United Kingdom and elsewhere in Western Europe [1–4]. In both the United States of America and the United Kingdom, more than half of new HIV infections are occurring in MSM [4,5]. The major HIV prevention intervention has been the promotion of risk-reduction behaviour, which was traditionally focused on uninfected persons. More recently, the focus has been changing towards earlier diagnosis of infection through ‘routinized’ HIV testing in multiple healthcare settings [6,7], coupled with behaviour-change counselling of those diagnosed as HIV positive [8–10]. However, HIV prevention resources are limited, behaviour-change counselling interventions are costly (preventing many successful interventions from being routinely implemented) and their effects can be transient [10,11], so resources need to be focused where they will be most effective.

Prevention of onward transmission is pertinent to all stages of HIV infection. However, targeting of primary HIV infection (PHI) may be highly cost-effective in populations where it accounts for a large proportion of HIV transmission events, as it is short-lived [12–14], meaning that even transient reductions in risk behaviour can mitigate its high infectivity. In phylogenetic studies, it was found that a-quarter-to-a-half [14,15] of HIV transmission may occur during PHI, as a result of high viral load [12], common coinfection with sexually transmitted infections (STIs) that promote HIV transmission [13] and typically high-risk behaviour [13]. However, effective diagnosis of PHI is resource-intensive, requiring targeted, frequent testing of high-risk groups.

Increased HIV testing through ‘routinizing’ of testing in healthcare settings has been advocated to reduce the prevalence of undiagnosed HIV infection [5] and to facilitate behaviour change to reduce transmission [7,8,16]. However, the frequency of testing is likely to be too low to increase the detection of PHI, because of its short duration. Therefore, additional targeted efforts to detect PHI in high-risk groups through high-frequency testing may be an effective component of HIV prevention – depending upon the impact of diagnosis on transmission-risk behaviour.

To assess the prevention-effectiveness of diagnosing PHI in high-risk groups, we need to consider (i) how effectively PHI cases can be identified and (ii) how much of the PHI transmission risk can be averted. We address the latter question in this study in order to understand the impact of routine counselling and care, and to identify characteristics of patients for whom this alone is insufficient.

Studying PHI is difficult, because its short duration and many missed opportunities for diagnosis limit the number of subjects that can be identified [17]. Data on the risk behaviour of newly HIV-infected individuals before and after HIV diagnosis are limited and conflicting [18,19], so more studies are required. It is important to assess risk behaviour and quantify changes post-diagnosis, to determine the potential impact of increased detection of PHI, and to identify those most in need of assistance in reducing their transmission-risk behaviour so that counselling efforts can be targeted most effectively. Our objectives were (i) to characterize MSM diagnosed with PHI at an HIV clinic in London; (ii) to investigate whether HIV diagnosis and standard transmission risk-reduction counselling changed their sexual behaviour; and (iii) to examine risk factors associated with behaviours that pose a continued transmission risk to others, post-diagnosis.


MSM diagnosed with PHI were recruited from the sexual health services between January 2002 and January 2004, and enrolled in an antiretroviral intervention study, a median of 7 days post-diagnosis. PHI was defined by (i) a positive HIV antibody test within 6 months of a documented seronegative test; (ii) a positive ‘de-tuned’ (optical density<0.6) Abbott enzyme-linked immunosorbent assay (Abbott Laboratories, North Chicago, IL); (iii) a positive HIV DNA polymerase chain reaction result in the absence of an HIV antibody response; or (iv) an evolving-titre positive HIV antibody test. Local Regional Ethics Committee approval was received. Participants gave written informed consent. All received the same standard safe-sex counselling post-test and at follow-up visits for care [20]. Sexual health screens were performed at baseline and at week 12.

Sexual behaviour questionnaire

Data were collected using a structured electronic questionnaire at baseline and at 12-week follow-up (±5 days). We collected data on general characteristics including age, ethnicity, likely country of HIV acquisition, sexual health, circumcision status and history of receiving payment for sex.

Risk-behaviour questions covered the frequency of sexual acts and condom use, with both regular and casual partners, and drug use in the previous 12 weeks. Condom use was categorized as never, sometimes or always (analysed as always vs. never or sometimes). HIV disclosure to regular partners and knowledge of their HIV status was also assessed. (We did not ask about the HIV status of casual partners, as this is often not known reliably by respondents [21].)


Risk of onward transmission was defined as unprotected anal intercourse (UAI) with a regular partner of unknown or negative HIV status, UAI with casual male partners or incident STI.

Risk-reduction behaviour was defined as reducing UAI with a regular partner of unknown or negative HIV status or reducing the number of casual sexual partners.

Statistical analysis

Data were stored and analysed using SPSS v14.0 (SPSS, Chicago, IL). Odds ratios were calculated in univariate analysis, and we used a χ2 test for categorical variables, with continuity correction where indicated, and a Mann–Whitney test for continuous variables. To assess behaviour changes of individuals we used the Wilcoxon sign rank test to compare baseline and follow-up. Where data were missing, the observations were excluded from the specific analysis affected.


Of 104 MSM identified as having PHI, 98 (94%) were recruited; all attended the follow-up interview.

Baseline characteristics

The baseline characteristics are shown in Table 1. The median age of the patients was 33 years (range 20–59 years) and the majority (94; 96%) were of white ethnicity. Eighty-six experienced seroconversion-like symptoms; the median estimated time from infection to HIV diagnosis was 78 days [20].

Table 1.   Demographics and baseline characteristics
Ethnicity [n (%)]
 Caucasian94 (96)
 Black African0 (0)
 Black Caribbean1 (1)
 Asian2 (2)
 Other1 (1)
Age (years) [median (range)]33 (20–59)
Seroconversion symptoms [n (%)]86 (88)
HIV infection acquired in United Kingdom [n (%)]96 (98)
 B96 (98)
 Non-B2 (2)
Days from seroconversion symptoms to HIV diagnosis [median (range)]78 (12–135)
Sexually transmitted infection at diagnosis (other than HIV) [n (%)]25 (26)
Regular partner [n (%)]
 HIV negative24 (25)
 HIV positive18 (18)
 HIV status unknown10 (10)
 No regular partner46 (47)
Number of casual sexual partners in previous 3 months [n (%)]
 010 (10)
 1–537 (38)
 6–1017 (17)
 >1034 (35)
Unprotected insertive anal sex with casual sexual partner [n (%)]50 (51)
Unprotected receptive anal sex with casual sexual partner [n (%)]63 (64)
Ever received payment for sex [n (%)]37 (38)

Fifty-two men (53%) had a regular partner. Of these regular partners, 18 (35%) were known to be HIV positive and 24 (46%) HIV negative, and 10 were of unknown status. Ten of the 52 men with a regular partner reported monogamous relationships. Just over half of all respondents (51 of 98; 52%) reported six or more casual partners in the previous 12 weeks and 52 (53%) reported >100 sexual partners in their lifetime. Twenty-five (26%) had another STI at HIV diagnosis.

Risks for HIV transmission in the 12 weeks prior to diagnosis

Prior to HIV seroconversion, 24 men reported UAI with their regular partner: 15 were known to be HIV-infected, and the other nine were of unknown HIV status. Of these 24 men, 12 (50%) – including nine of the 15 men with a regular partner who was known to be HIV-infected – also had UAI with casual partners. A further 61 men reported UAI with at least one casual partner; 50 (51%) reported unprotected insertive anal intercourse (UIAI) and 61 (62%) reported unprotected receptive anal intercourse (URAI). This included 40 individuals who reported both types of unprotected anal sex.

Thirty-seven men (38%) had ever been paid for sex, including three who were sex workers at the time of HIV acquisition.

Seventy men (71%) reported recreational drug use, including cocaine (59), ecstasy (57), gamma-hydroxybutyrate (GHB; 28), ketamine (42) and crystal methamphetamine (8).

Risks for HIV transmission in the 12 weeks post-diagnosis

There were significant changes in risk behaviour in the 12 weeks following HIV diagnosis, with 74 men (76%) posing no risk of onward transmission during that period. In the whole study cohort there was an overall shift towards fewer sexual partners (see below), a 44% reduction in the number of patients with a recent STI (from 25 to 14) and a 29% reduction in the number reporting recreational drug use (from 70 to 50).

Sexual partners

Paired analysis for individuals showed a significant shift towards fewer casual partners in the 12-week period following PHI diagnosis compared with the preceding 12 weeks (Fig. 1), with 65 men reducing their numbers of casual partners, 26 staying the same and seven reporting more partners (Wilcoxon sign rank test, Z=−6.302, P<0.001). Thirteen men reported no sexual contact and 29 had one casual partner, but 22 (22%) reported six or more casual partners.

Figure 1.

 Comparison of numbers of casual sexual partners in the 12 weeks before and after diagnosis of primary HIV infection. Bars indicate the frequency distribution of participants reporting various numbers of casual sexual partners at follow-up, grouped by numbers reported at baseline. The majority of study participants greatly reduced their numbers of casual partners in the 12 weeks post-diagnosis, with many reporting no casual partners, although a few reported increased numbers.

Condom use

Condom use increased for anal sex with casual partners: the proportion always using condoms during RAI increased from 13 of 76 (17%) to 29 of 45 (64%) (P<0.001), and for IAI this proportion increased from 22 of 72 (31%) to 28 of 46 (61%) (P<0.01). Paired analysis for RAI showed 23 men increasing condom use, 16 staying the same and two using condoms less (Wilcoxon sign rank test, Z=−4.097, P<0.001). For IAI, 15 men increased condom use, 19 stayed the same and five reduced condom use (Wilcoxon sign rank test, Z=−2.294, P=0.024).

Ongoing transmission risk

Detailed analysis of the changes in sexual behaviour relevant to reducing onward transmission needs to take into account not only numbers of sexual partners and behaviour with those partners, but also whether they are known to have HIV infection or not. Risk of onward transmission was defined as UAI with a regular partner of unknown or negative HIV status, UAI with casual male partners or incident STI. Seventy-four men (76%) reported no risk to others, while 24 (24%) reported one or more risks: four had UAI with an HIV-negative partner, four had UAI with a partner of unknown HIV status and 16 had UAI with a casual partner(s).

We compared the characteristics of the 24 men who reported behaviours that posed a continuing transmission risk after their diagnosis with those of the 74 men who did not (Table 2). The two groups of men had similar demographic characteristics, but the men with a continuing transmission risk were significantly more likely at diagnosis to have had another STI, reported higher numbers of sexual partners and have taken ketamine. However, men with a continuing transmission risk did significantly reduce their numbers of sexual partners, with 14 of 24 reducing their numbers of partners, eight of 24 staying the same and only two of 24 increasing their numbers of partners (Wilcoxon sign rank test, Z=−2.610, P<0.009).

Table 2.   Comparison of men who posed a potential risk of onward transmission during the 12 weeks after PHI diagnosis with those who did not (see text for definitions)
Variable*Ongoing transmission risk
Any risk (n=24)No risk (n=74)Odds ratio (95% CI)Test statistic, P value
  • *

    Data from baseline visit.

  • Age was compared using a Mann–Whitney test; odds ratios and 95% confidence intervals (CIs) were calculated for the association between ongoing transmission risk and the binary categorical explanatory variables listed, and a χ2 test with continuity correction was used to determine significance; number of partners was categorized and a χ2 test for trend carried out.

  • PHI, primary HIV infection; STI, sexually transmitted infection.

Age (years) [median (range)] 31.0 (20–59)32.5 (21–59) Z=–0.327, P=0.744
Previous sex work [n (%)] 11 (45.8)26 (35.1)1.562 (0.614, 3.975)χ2=0.883, P=0.347
STI at baseline [n (%)] 10 (41.7)15 (20.5)2.762 (1.026, 7.436)χ2=4.211, P=0.040
Number of casual partners 12 weeks before diagnosis [n (%)]01 (4.2)9 (12.2) χ2=12.697, P=0.005
1–53 (12.5)34 (45.9)  
6–106 (25.0)11 (14.9)  
>1014 (58.3)20 (27.0)  
Any drug use [n (%)] 19 (79.2)51 (68.9)1.714 (0.570, 5.155)χ2=0.993, P=0.334
Ketamine use [n (%)] 15 (62.5)27 (36.5)2.901 (1.119, 7.519)χ2=5.008, P=0.025

Three individuals transmitted HIV to their regular partner during the study period, which was confirmed through sequence analysis of the polymerase (pol) gene.

Disclosure of HIV status to regular sexual partner

Most men with a regular partner (46 of 52; 88%) disclosed their new HIV diagnosis to their partner, but six did not. Of those men who did not disclose their diagnosis, one had an HIV-positive regular partner at baseline and they continued to have UAI; another had an HIV-negative partner at baseline and they continued to have UAI. Four men acquired a new regular partner during the 12 weeks: of these, one was known to be HIV positive and three were of unknown status, one of whom was at risk from UAI with the index case.


In this study of 98 men with PHI (94% participation rate), acquisition of HIV was associated with high levels of recreational drug use, commercial sex (as a client or worker), high numbers of sexual partners and low levels of condom use, which is in agreement with other studies [18, 19, 21]. For at least the 3 months following diagnosis and standard counselling, three-quarters of participants eliminated their risk of onward transmission entirely, through changing their behaviour. Overall, there was a significant reduction in transmission-risk behaviour, with greater condom use and fewer sexual partners, lasting for at least the duration of PHI. Consistent with this, incident STIs also decreased. This behaviour change occurred following standard counselling from a physician. Further reductions may well be achieved if more intensive behaviour-change interventions were delivered to those whose behaviour continued to pose a transmission risk; these individuals were more likely to have another STI, use ketamine and have more sexual partners at baseline. The substantial reduction in transmission risk following diagnosis is an important finding and shows the need to improve the early diagnosis of HIV infection. Currently, only a minority of people are diagnosed early, as such most individuals do not have the opportunity to reduce their onward transmission risk during the highly infectious PHI.

While it is reassuring that, in three-quarters of cases, diagnosis and standard counselling led to an elimination of UAI with regular partners (of unknown and HIV-negative status) and all casual partners, a quarter of respondents nevertheless continued to pose a risk of onward transmission – although most of them did reduce their risk. Also consistent condom use for (both insertive and receptive) anal sex with HIV-negative regular partners remained below 50% overall and 16% of individuals reported UAI with casual partners. Further counselling effort should be directed to those whom we have identified as being most likely to pose an ongoing risk; risk factors included having had another STI and having had higher numbers of sexual partners.

Serosorting (choosing HIV-concordant partners to avoid transmission risk) [22–24], which was not assessed for casual partners, may further reduce transmission risk. However, at follow-up, 8% of participants reported UAI with a regular partner known to be HIV negative or of unknown HIV status, and serosorting with casual partners is known to be low [21]. (In our study 16% reported UAI with casual partners.) Importantly, if serosorting occurred while the participants wrongly believed that they were still HIV negative, then transmission would have been promoted [25]. This may explain why a recent study found that serosorting by MSM in the United States of America had only limited protective effect [26].

Our study was powerful in having a high level of participation, rapid enrolment post-diagnosis and 100% follow-up (<80% is typical), which are difficult to achieve in studying PHI [18,19]. There was little evidence of recall bias, as evidenced by the greatly reduced numbers of incident STIs corroborating the reported reduction in risk behaviour. Additionally, some participants reported increased transmission risk, and there were high rates of reported risk behaviour with regular partners. Limitations of the study are the descriptive format and the inability to distinguish whether the behaviour change observed was a consequence of counselling or knowledge of the infection per se. Standard counselling has been instituted, so this is of little relevance to the clinical situation.

As rates of HIV transmission in MSM in the United States of America, the United Kingdom and other parts of Western Europe remain high [1,2,4], improving our HIV prevention strategy is essential. Unfortunately, most HIV infections are not diagnosed at an early stage [27,28], permitting prolonged unknowing transmission risk. Clearly it is important to diagnose all HIV infections as early as possible – with ‘routinizing’ of testing in multiple healthcare settings having an important role to play [6, 7] – and to reduce transmission risk from all stages of infection through HIV care [16] and effective behavioural interventions [8].

Limited prevention resources need to be focused where they will be most effective, and this study indicates that an important component of HIV prevention should be targeted efforts to achieve rapid diagnosis of PHI in high-risk groups, which can greatly reduce transmission through (even transient) behaviour change, with extra counselling being directed to those most likely to pose an ongoing transmission risk. Identification of PHI requires a high frequency of testing, which is both feasible and cost-effective for high-risk individuals [29,30]. HIV RNA screening could identify substantially more PHI cases than antibody testing alone [31]. Effective strategies need to be developed to increase diagnoses of PHI, such as through increasing awareness of symptoms in non-GUM/HIV healthcare professionals as well as MSM to reduce missed opportunities for diagnosis [17,31], as well as increased emphasis on outbreak detection in surveillance to facilitate case finding [15]. More study is required to find ways to further reduce transmission risk in those who we found did not eliminate it completely.


NM thanks the UK Medical Research Council, and PJW thanks UNAIDS and the UK Medical Research Council, for funding. The funding sources had no involvement in the study design or conduct; the collection, analysis and interpretation of data; the preparation, review or approval of the manuscript; or the decision to submit the manuscript for publication.

Conflicts of interest: J Fox, P White, N Macdonald, J Weber, M McClure and S Fidler have no conflicts of interest.

H Ward is editor of the journal Sexually Transmitted Infections.