Modelling the impact of chlamydia screening on the transmission of HIV among men who have sex with men
- 3k Downloads
Recent studies have found high prevalences of asymptomatic rectal chlamydia among HIV-infected men who have sex with men (MSM). Chlamydia could increase the infectivity of HIV and the susceptibility to HIV infection. We investigate the role of chlamydia in the spread of HIV among MSM and the possible impact of routine chlamydia screening among HIV-infected MSM at HIV treatment centres on the incidence of chlamydia and HIV in the overall MSM population.
A mathematical model was developed to describe the transmission of HIV and chlamydia among MSM. Parameters relating to sexual behaviour were estimated from data from the Amsterdam Cohort Study among MSM. Uncertainty analysis was carried out for model parameters without confident estimates. The effects of different screening strategies for chlamydia were investigated.
Among all new HIV infections in MSM, 15% can be attributed to chlamydia infection. Introduction of routine chlamydia screening every six months among HIV-infected MSM during regular HIV consultations can reduce the incidence of both infections among MSM: after 10 years, the relative percentage reduction in chlamydia incidence would be 15% and in HIV incidence 4%, compared to the current situation. Chlamydia screening is more effective in reducing HIV incidence with more frequent screening and with higher participation of the most risky MSM in the screening program.
Chlamydia infection could contribute to the transmission of HIV among MSM. Preventive measures reducing chlamydia prevalence, such as routine chlamydia screening of HIV-infected MSM, can result in a decline in the incidence of chlamydia and HIV.
KeywordsHIV Chlamydia trachomatis Men who have sex with men Mathematical model Chlamydia screening
Human immunodeficiency virus
Men who have sex with men
Sexually transmitted infection
Unprotected anal intercourse.
In several countries, high prevalences of asymptomatic rectal chlamydia have been found among HIV-infected men who have sex with men (MSM) [1, 2, 3, 4]. Most of these infections remain undetected and untreated. Therefore, screening for chlamydia among HIV-infected MSM may considerably reduce the prevalence of chlamydia in this specific subpopulation. In addition, certain studies suggest that chlamydia screening among MSM may result in a decline in HIV transmission, because infection with chlamydia may increase the transmissibility of HIV (for HIV-infected individuals) and the susceptibility to HIV infection (for HIV-negative individuals) [5, 6, 7, 8, 9, 10, 11]. However, the benefits of chlamydia screening are still unclear; for MSM, this is because of the perceived mildness of complications of chlamydia infection in men.
In this study, we investigated the role of rectal chlamydia in the spread of HIV among MSM, using a dynamic transmission model. An increase in HIV infectivity and susceptibility in individuals with chlamydia was included in the model. Also, we accounted for the impact of antiretroviral therapy (ART) in reducing HIV infectivity and in eliminating the increase in HIV infectivity due to co-infection with chlamydia. We examined how chlamydia affects the spread of HIV in the population, by calculating the fraction of new HIV infections attributable to chlamydia infection. Subsequently, we studied the impact of chlamydia screening among HIV-infected MSM at HIV treatment centres on the incidence of chlamydia and HIV in the MSM population. To reveal the dependence of the results on the assumption of increased HIV infectivity and susceptibility due to chlamydia, the results were calculated with different levels of this increase.
We developed a deterministic compartmental model that describes the transmission of HIV and chlamydia among sexually active MSM. In the model, infection with HIV or with chlamydia occurs only via unprotected anal intercourse (UAI) between men. Three types of partnerships are distinguished in the model: steady partners, single-act casual partners (with whom they have only one sexual contact and that is UAI), and multiple-acts casual partners (with whom they have multiple sexual contacts of which at least one is UAI). The population is stratified into four sexual risk groups, with increasing level of sexual risk behaviour: low, fairly high, very high, and extremely high (the last three are referred to as high risk groups). The level of risk behaviour is determined by the total number of sexual partners with whom men have UAI. Low risk MSM have no UAI with casual partners. High risk MSM have UAI with casual partners; the total number of partners is highest in the extremely high risk group and lowest in the fairly high risk group (Additional file 1: Table S2).
The model is defined by a system of differential equations and the parameters are defined in Additional file 1: Tables S1–S3. The model was parameterized and calibrated to reflect the current situation among MSM in the Netherlands (see section about the uncertainty analysis).
Increased HIV infectivity and susceptibility due to chlamydia infection
We assumed that for individuals infected with both HIV and chlamydia, the level of HIV infectivity is increased by a factor v (compared to HIV infectivity for individuals without chlamydia), due to co-infection with chlamydia [7, 8, 9]. For those infected with chlamydia but not HIV, the susceptibility to HIV infection is increased by a factor φ (compared to HIV susceptibility for those without chlamydia), due to chlamydia infection [10, 11]. The number of new HIV infections that can be attributed to infection with chlamydia was calculated as the difference between the total number of new HIV infections and the number of new HIV infections if the two factors were both equal to one. Dividing this number by the total number of new HIV infections resulted in the fraction of new HIV infections attributable to infection with chlamydia.
Uncertainty analysis and model calibration
To account for uncertainty in model parameters, a range of possible values was assigned to each uncertain parameter. Using Latin Hypercube Sampling , 10,000 sets of values were sampled from the uniform distribution on these ranges. The model equations were then solved numerically with each of the 10,000 parameter sets until the system reached a stable state; this is assumed to represent the current situation in the Netherlands (results with the current opportunistic screening rates for chlamydia – see details on screening scenarios below).
Then we calibrated the model to the current number of HIV-infected MSM in care at HIV treatment centres in the Netherlands (N = 8,523 in 2011 ). Let Z i denote the number of HIV-infected MSM in care calculated from the model with the i-th set of parameter values, for i = 1,2,…,10,000. We calculated the Poisson likelihood of each estimate Z i by assuming that the number of HIV-infected MSM in care follows the Poisson distribution with mean Z i . This likelihood represents the probability of observing the value 8,523, if the true expectation were Z i . In this way, each of the 10,000 estimates Z i is assigned a likelihood representing how likely this estimate is, given our current knowledge. Subsequently, we selected the parameter sets with high likelihood: those with likelihood higher than 1/8 of the saturated likelihood (which is the likelihood of the value 8,523 from the Poisson distribution with mean 8,523): 0.0043212/8 = 0.00054. In this way, we selected the parameter combinations resulting in likely estimates of the number of HIV-infected MSM in care and excluded those resulting in unlikely estimates. In the remaining of the manuscript, we present results only from the selected parameter sets.
Next, the screening rates were modified to reflect changes in the testing frequency of MSM. The model equations were solved with the new screening rates and the selected parameter sets for the subsequent thirty years. For each outcome of interest (for example, prevalence of HIV) we report the median and the interquartile range (IQR: from the 25th to the 75th value) of the values calculated only with the selected parameters.
Scenarios for chlamydia screening
Current opportunistic chlamydia screening
Currently in the Netherlands, there is no routine screening for chlamydia, but only opportunistic screening: individuals without chlamydia symptoms may be tested for chlamydia and other sexually transmitted infections (STIs) at their own initiative, at STI clinics or general practitioners. The current frequency of opportunistic screening was estimated as follows: low risk men are tested every 2.5-3.5 years; high risk men in care every 1–1.5 years; high risk men not infected with HIV or HIV-infected not in care every 1.5-2.5 years [1, 15] (Additional file 1: Table S3).
Reduction in the frequency of the current opportunistic screening among MSM
To reveal the impact of the current opportunistic chlamydia screening on the incidence of chlamydia and HIV, we examined first a hypothetical scenario where the frequency of opportunistic screening is reduced. This could happen, for instance, if campaigns promoting chlamydia screening are relaxed or if the need to treat asymptomatic chlamydia is perceived as less important than it is now. This scenario was implemented by increasing the time interval between chlamydia tests by 50%: low risk men are tested every 3.75-5.25 years; high risk men in care every 1.5-2.25 years; high risk men who are not infected with HIV or HIV-infected not in care every 2.25-3.75 years.
Introduction of routine chlamydia screening of HIV-infected MSM in care at HIV treatment centres
Chlamydia screening among HIV-infected MSM in care could have an impact on reducing HIV transmission, because chlamydia infection may increase HIV infectivity. Moreover, chlamydia screening could be implemented during the existing visits of HIV-infected men at HIV treatment centres. To investigate the impact of such a program, we examined three hypothetical scenarios with different screening frequencies: every twelve months, every six months, or every four months. In these scenarios, it is assumed that: only HIV-infected MSM in care are screened; routine screening is added to (and not replacing) the current opportunistic screening; participation in the routine screening program is 100%, which means that all HIV-infected MSM in care are screened (and treated if positive) every twelve, six, or four months. The three scenarios are referred to as routine screening at HIV treatment centres; we present its impact on reducing HIV incidence in the whole MSM population.
Suboptimal participation in routine chlamydia screening of HIV-infected MSM in care at HIV treatment centres
Although HIV-infected MSM in care have regular consultations at HIV treatment centres, some of them may not participate in the routine screening program (for instance, because they have been recently tested at STI clinics, due to symptoms or known risk of exposure) or they may participate but not return to receive their medication, if they are found positive (mostly because of the lack of symptoms or hindrance ). In addition, recent data suggest that the current treatment may not be 100% effective . To model these conditions, we recalculated the impact of one of the scenarios for routine screening assuming that participation is suboptimal (less than 100%); this was done for the scenario with screening every six months, also referred to as semi-annual screening. In order to examine the role of the different sexual risk groups, we studied four scenarios: in each scenario, participation was 80% in one of the four sexual risk groups and 100% in the other three risk groups:
80% participation of low risk HIV-infected MSM in care;
80% participation of fairly high risk HIV-infected MSM in care;
80% participation of very high risk HIV-infected MSM in care;
80% participation of extremely high risk HIV-infected MSM in care.
Percentage change in incidence
where, “current” is the incidence with the current opportunistic screening and “new” is the incidence with the new screening scenario.
The interaction between HIV and chlamydia
First, we calculated from the model the prevalence and the incidence of HIV and of chlamydia infection, for the current situation in the Netherlands (with only opportunistic screening). The calculated median prevalence of HIV is 4.26% (IQR, 4.19-4.33%) and the median prevalence of chlamydia 2.69% (IQR, 1.68-3.71%) in the MSM population. However, large variations are observed in the prevalences of HIV and chlamydia within the population. Specifically, the prevalence of HIV varies from 0.5% in the lowest risk group to 72.8% in the highest risk group; the prevalence of chlamydia infections varies from 0.1% in the lowest risk group to 57.6% in the highest risk group (Additional file 1: Figure S1).
The median prevalence of chlamydia is 1.8% among MSM not infected with HIV, but 22.7% among HIV-infected MSM (Additional file 1: Figure S2). The high prevalence of chlamydia among HIV-infected MSM is mostly attributable to the high risk behaviour in this specific subgroup. From the model, it was calculated that MSM who have UAI with casual partners (high risk MSM) comprise 90.8% of the HIV-infected MSM population, but only 24.5% of the MSM not infected with HIV. This implies that the men who get infected with HIV are mostly those with high sexual risk behaviour (having UAI with casual partners) and because of their high risk behaviour they are also more likely to get infected with chlamydia.
The impact of chlamydia screening on the incidence of HIV and of chlamydia
In this modelling study, we showed that routine chlamydia screening among HIV-infected MSM in care can result in reductions in chlamydia and HIV incidence in the overall MSM population. The impact of chlamydia screening on HIV incidence is a result of the increased HIV infectivity and susceptibility in individuals infected with chlamydia. The impact of chlamydia screening specifically among HIV-infected MSM in care can be explained by the fact that chlamydia incidence among HIV-infected MSM in care is much higher than among other MSM, due to the high density of high risk MSM within the HIV-infected population. Therefore, reducing chlamydia transmission in this core group will result in health gains not only for themselves, but also for the rest of the MSM population by preventing new chlamydia and HIV infections.
It should be stressed that the impact of chlamydia on HIV transmission shown here depends on the assumption of increased HIV infectivity and susceptibility due to chlamydia infection. This impact declines as the increase in HIV infectivity and susceptibility becomes smaller (Figures 2a,b and 4c,d); and if HIV infectivity and susceptibility are not altered at all in those with chlamydia infection, then chlamydia has no effect on HIV transmission. On the other hand, the decline in HIV incidence that we found due to chlamydia screening of HIV-infected MSM in care is not very high, as is the case with other interventions, such as widespread use of ART [18, 19, 20], or as previous studies have suggested [5, 6]. This is a result of two factors. First, we assumed that HIV infectivity and HIV susceptibility are increased due to chlamydia infection by a factor of two, at most. Other modelling studies for heterosexuals have assumed higher increases [5, 19], but we found no studies with significant evidence of such high increases [7, 8, 9, 10, 11]. Second, the increase in HIV infectivity due to chlamydia is diminished in individuals with undetectable viral load [9, 21]; these individuals comprise the majority of the target group, since most of the HIV-infected MSM in care receive ART and most of them have undetectable viral load .
The impact of STIs on the HIV epidemic among heterosexuals has been investigated in several modelling studies (see, for instance, [6, 22, 23, 24, 25, 26]). For MSM populations, this is the first study that addresses this issue and examines the impact of chlamydia screening specifically among HIV-infected MSM registered at HIV treatment centres. This is a high risk population, contributing the most to HIV and chlamydia transmission. Moreover, screening HIV-infected MSM in care can be easily implemented during the regular visits of these men at HIV clinics.
Certain limitations of this study should be mentioned. A test for anogenital chlamydia, in practice, may be carried out in combination with a test for gonorrhoea. Therefore, gonorrhoea infections may also be detected and treated, resulting in extra reduction in HIV transmission, since gonorrhoea also increases HIV infectivity and susceptibility [7, 11]. From this point, our model may have underestimated the impact of routine chlamydia screening, as it does not account for the indirect effect of combined screening for chlamydia and gonorrhoea. Furthermore, some individuals treated for gonorrhoea, may also receive treatment for chlamydia; in that case the prevalence of chlamydia is reduced and consequently the impact of chlamydia screening could be lower.
Studies among individuals with recently acquired HIV infection have found high prevalences of chlamydia, suggesting that co-transmission of HIV and chlamydia could be possible and even frequent [27, 28, 29]. However, there are no data to verify that the two infections were acquired via the same sexual contact or to inform parameters relating to co-transmission. Therefore transmission of HIV and of chlamydia was modelled separately, such that co-transmission of the two infections is not possible and hence its potential impact is not accounted for in the model. Moreover, we did not account for orogenital transmission of HIV and chlamydia due to the lack of data. To avoid bias, we consistently restricted all related parameters and data to anogenital transmission, as much as possible. Nevertheless, for some parameters, data on heterosexual transmission were used, since there were no data for transmission between MSM.
Earlier modelling studies have shown that a substantial proportion of new HIV infections can be attributed to individuals with acute HIV infection [30, 31], due to the high level of HIV infectivity during this short stage. This means that the impact of chlamydia might be overestimated in this study, since the impact is reduced with higher HIV infectivity (Figure 2d). Due to the complexity of the model, acute HIV infection was not included. Moreover, we did not account for differences in the duration of asymptomatic chlamydia between HIV-infected and HIV-negative MSM, because there are hardly any data on asymptomatic infections. Finally, in our calculations, we assumed a moderately assortative pattern of sexual mixing [32, 33]; the role of mixing on the spread of STIs has been investigated in previous modelling studies (see, for instance, [34, 35, 36, 37]).
After the introduction of routine chlamydia screening at HIV treatment centres, HIV-infected MSM in care may be less likely to be tested opportunistically outside the routine screening program. This means that HIV-infected MSM in care may not seek STI testing after symptoms or known risk of exposure, awaiting for the arranged regular visit at the HIV treatment centre. Moreover, opportunistic STI screening is mostly implemented at STI clinics, where also safe sex counselling and partner notification are offered. This means that introduction of routine chlamydia screening at HIV treatment centres may result in a reduction in opportunistic testing for other STIs, a reduction in partner notification, and, consequently, in increases in transmission of HIV and chlamydia. From this point, it is important that HIV-infected MSM in care who test positive for chlamydia during routine screening will still be offered a full STI consultation including safe sex counselling and partner notification.
These results have important implications for the design of public health policy interventions. Measures that can reduce chlamydia prevalence, such as chlamydia screening among HIV-infected MSM in care, should be promoted because they may also contribute to reducing HIV transmission. However, they should be considered as an addition to (and not as a substitute of) other measures to control HIV. Furthermore, it is essential that in particular high risk MSM participate in screening programs. Finally, it should be stressed that although data from the Netherlands were used for population characteristics in our model, such as sexual risk behaviour, our results were robust to changes in these characteristics. Hence, our findings apply also to other countries with considerable HIV and STI transmission among MSM.
In conclusion, this analysis shows that chlamydia infection could have a considerable contribution to the population spread of HIV among MSM. Routine chlamydia screening of HIV-infected MSM at HIV treatment centres has the potential to reduce HIV and chlamydia incidence not only in the screened population of HIV-infected MSM in care, but among all MSM. Chlamydia screening will be more effective in reducing HIV incidence with more frequent testing or with higher participation of high-risk MSM in the screening program.
The authors would like to thank Gerben Rienk Visser for processing the data on the duration of steady relations and for providing the data on sexual behaviour from the Amsterdam Cohort Study. Hanneke de Graaf is acknowledged for providing data on the frequency of sex between MSM from the survey of Rutgers WPF. The authors thank Femke Koedijk, Amy Matser, Ingrid van der Broek, Ineke Stolte, and Wijnand van den Boom for providing data on chlamydia testing, chlamydia prevalence, and sexual behaviour of MSM. The authors are grateful to the reviewers for their suggestions that considerably improved the manuscript.
This project was partially funded by the Dutch Ministry of Health, Sport, and Welfare.
- 1.Heiligenberg M, Rijnders B, Schim van der Loeff M, de Vries H, van der Meijden W, Geerlings S, Fennema H, Prins M, Prins J: High prevalence of sexually transmitted infections in HIV-infected men during routing outpatient visits in the Netherlands. Sex Transm Dis. 2012, 39: 8-15. 10.1097/OLQ.0b013e3182354e81.CrossRefPubMedGoogle Scholar
- 2.Kent C, Chaw J, Wong W, Liska S, Gibson S, Hubbard G, Klausner J: Prevalence of rectal, urethral, and pharyngeal chlamydia and gonorrhea detected in two clinical settings among men who have sex with men: San Francisco, California, 2003. Clin Infect Dis. 2005, 41: 67-74. 10.1086/430704.CrossRefPubMedGoogle Scholar
- 4.Dang T, Jaton-Ogay K, Flepp M, Kovari H, Evison J, Fehr J, Schmid P, Boffi El Amari E, Cavassini M, Odorico M, Tarr P, Greub G: High prevalence of anorectal chlamydial infection in HIV-infected men who have sex with men in Switzerland. Clin Infect Dis. 2009, 49: 1532-1535. 10.1086/644740.CrossRefPubMedGoogle Scholar
- 9.Sadiq S, Taylor S, Kaye S, Bennett J, Johnstone R, Byrne P, Copas A, Drake S, Pillay D, Weller I: The effects of antiretroviral therapy on HIV-1 RNA loads in seminal plasma in HIV-positive patients with and without urethritis. AIDS. 2002, 16: 219-225. 10.1097/00002030-200201250-00011.CrossRefPubMedGoogle Scholar
- 12.Gras L, van Sighem A, Smit C, Zaheri S, Prins M, de Wolf F: Monitoring of Human Immunodeficiency Virus (HIV) infection in the Netherlands Report 2010. 2010, Amsterdam, The Netherlands: HIV Monitoring FoundationGoogle Scholar
- 13.McKay M, Beckman R, Conover W: A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics. 1979, 21: 239-245.Google Scholar
- 14.van Sighem A, Smit C, Gras L, Holman R, Stolte I, Prins M, de Wolf F: Monitoring of Human Immunodeficiency Virus (HIV) infection in the Netherlands Report 2011. 2011, Amsterdam, The Netherlands: HIV Monitoring FoundationGoogle Scholar
- 15.Stolte I: Sexual behaviour. Final report of the health monitor of Amsterdam, 2008. Edited by: Dijkshoorm H, van Dijk T, Janssen A. 2009, Amsterdam: GGD Amsterdam, 121-125.Google Scholar
- 16.van Valkengoed I, Morre S, van den Brule A, Meijer C, Bouter L, van Eijk J, Boeke A: Follow-up, treatment, and reinfection rates among asymptomatic Chlamdyia trachomatis cases in general practice. Brit J Gener Pract. 2002, 52: 623-627.Google Scholar
- 19.Boily M-C, Bastos F, Desai K, Mâsse B: Changes in the transmission dynamics of the HIV epidemic after the wide-scale use of antiretroviral therapy could explain increases in sexually transmitted infections. Sex Transm Dis. 2004, 31: 100-113. 10.1097/01.OLQ.0000112721.21285.A2.CrossRefPubMedGoogle Scholar
- 21.Kelley C, Haaland R, Patel P, Evans-Strickfaden T, Farshy C, Hanson D, Mayer K, Lennox J, Brooks J, Hart C: HIV-1 RNA rectal shedding is reduced in men with low plasma HIV-1 RNA viral loads and is not enhanced by sexually transmitted bacterial infections in the rectum. J Inf Dis. 2011, 204: 761-767. 10.1093/infdis/jir400.CrossRefGoogle Scholar
- 24.Korenromp EL, Van Vliet C, Grosskurth H, Gavyole A, van der Ploeg C, Fransen L, Hayes R, Habbema J: Model-based evaluation of single-round mass treatment of sexually transmitted diseases for HIV control in a rural African population. AIDS. 2000, 14: 573-593. 10.1097/00002030-200003310-00013.CrossRefPubMedGoogle Scholar
- 25.Freeman EE, Orroth KK, White RG, Glynn J, Bakker R, Boily M, Habbema D, Buve A, Hayes R: Proportion of new HIV infections attributable to herpes simplex 2 increases over time: simulations of the changing role of sexually transmitted infections in sub-Saharan African HIV epidemics. Sex Transm Infect. 2007, 83 (Suppl 1): i17-i24.CrossRefPubMedGoogle Scholar
- 26.Foss AM, Vickerman PT, Mayaud P, Weiss H, Ramesh B, Reza-Paul S, Washington R, Blanchard J, Moses S, Lowndes C, Alary M, Watts C: Modelling the interactions between herpes simplex virus type 2 and HIV: implications for the HIV epidemic in southern India. Sex Transm Infect. 2011, 87: 22-27. 10.1136/sti.2009.041699.CrossRefPubMedGoogle Scholar
- 28.Galvin SR, Cohen MS: The role of sexually transmitted diseases in HIV transmission. Nat Rev. 2004, 2: 33-42. 10.1038/nrmicro794.Google Scholar
- 29.Op de Coul ELM, Warning TD, Koedijk FDH, on behalf of the Dutch STI clinics: Sexual behaviour and sexually transmitted infections in STI clinic attendees in the Netherlands, 2007–2011. Intern J STD AIDS. 2013, In pressGoogle Scholar
- 33.Renton A, Whitaker L, Ison C, Wadsworth J, Harris JRW: Estimating the sexual mixing patterns in the general population from those in people acquiring gonorrhoea infection: theoretical foundation and empirical findings. J Epidem Commun Health. 1995, 49: 205-213. 10.1136/jech.49.2.205.CrossRefGoogle Scholar
- 34.Koopman J, Simon C, Jacquez J, Joseph J, Sattenspiel L, Park T: Sexual partner selectivenss effects on homosexual HIV transmission dynamics. JAIDS. 1988, 1: 486-504.Google Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2334/13/436/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.