By exploring various two-sex transmission models, we demonstrate that directing prophylactic intervention at a single sex more effectively reduces heterosexual STI transmission than any allocation that includes both sexes. In addition, we demonstrate that a strategy of protecting the sex with the highest endemic prevalence generally achieves the largest reduction in the population prevalence. The implication of our finding is that the prevaccine prevalence of infection might be a good proxy to determine which individuals should be vaccinated in order to achieve the highest impact of vaccination at the population level.
Our results provide a justification, under most circumstances, for the intuitively plausible strategy of targeting intervention at the subgroups that harbor most infections and that act as a reservoir for transmission. An alternative strategy that uses allocation rules defined on the basis of sex-specific reproduction numbers would also be intuitively plausible but performs poorly in minimizing the population prevalence of infection. Our results can be viewed as a generalization of a recently formulated argument for prioritization of vaccination to groups with the highest product of incidence and force of infection 
. Although we have already identified several exceptions (e.g., arising from different degrees of natural immunity throughout the population), it would be logical and prudent to further test the generality of the rule of targeting intervention at the subgroups with the highest endemic prevalence.
The allocation that achieves the largest reduction in the population prevalence of infection for a fixed amount of vaccine is not necessarily the most attractive from an economic point of view. The cost per vaccine dose delivered is subject to logistics, and universal vaccination could sometimes be a cost-effective alternative to single-sex vaccination. For example, the variable costs of vaccine purchase and delivery could be low compared to the total costs of running a vaccination program. In addition, the marginal cost of increasing vaccine uptake might depend on the coverage already achieved and might be different between the sexes. Males and females, or in the case of preadolescent vaccination, their parents, likely have different perceptions of the risk from HPV infection and different attitudes towards vaccination, although more research is needed to reliably measure vaccine acceptability 
. Finally, differences in cost-effectiveness between sex-specific vaccination programs are determined by the relative benefits of preventing infections in men and women. For example, HPV prevention programs started off offering vaccine to females because it is on average more beneficial to prevent HPV infection in a woman than in a man. Directing interventions at the sex most affected by disease makes sense from an equity perspective, and will also have the strongest impact on heterosexual transmission if infection is more prevalent in this sex.
Our analysis adds new arguments to the ongoing debate about whether males should also be offered HPV vaccination 
. A common rationale for including boys in existing vaccination programs is that they experience not only a direct benefit, but that vaccinating males also creates herd immunity that helps to protect women 
. The herd immunity argument can as well be used against male vaccination, for men already derive a substantial benefit from female-only vaccination 
. A recent modeling study concluded that heterosexual males would benefit almost to the same extent as females from a girls-only HPV vaccination program, due to herd immunity 
. We show that, once routine vaccination of one sex is in place, increasing the coverage in that sex is much more effective in bolstering herd immunity than switching to a policy that includes both sexes. Universal vaccination against HPV should therefore only become an option when vaccine uptake among girls cannot be further increased. Adding boys to current vaccination programs seems premature, because female coverage rates still leave ample room for improvement in most countries that have introduced HPV vaccination 
. So far, only three countries have achieved a three-dose coverage of 70% or more in females 
We have focused on a heterosexual population. Often, bisexuality acts as a bridge for transmission between heterosexual and homosexual subpopulations. This bridging phenomenon is especially important for the persistence of STIs, such as hepatitis B virus 
. Because of bisexuality, MSM can be expected to derive some benefit from a reduced transmission of HPV in the general population. Our study shows that female-only vaccination will never achieve the maximum possible reduction in HPV prevalence among MSM, but the realized reductions could constitute a considerable health benefit. The extent to which MSM may benefit from female-only vaccination should be contrasted with the effectiveness of targeted vaccination of MSM, who are at high risk for anal cancers 
. A recent publication reported that vaccination of MSM remains cost-effective up to 26 y of age 
, an age range that might render targeted HPV vaccination acceptable 
. Targeted vaccination of homosexual and bisexual men is an important topic for further investigation.
The free availability of quadrivalent HPV vaccine to young Australian women has led to a reduced morbidity of genital warts in STI clinics since 2007, among women as well as heterosexual men 
. Vaccinating boys might have brought about a similar—or even larger—decline in HPV infection rates than has been observed as a result of female HPV vaccination. Our analysis suggests this could have been the case if male-to-female transmissibility is substantially lower than female-to-male transmissibility, or if women have a lower degree of natural immunity than men. The latter is unlikely, because women generally have higher seroprevalence for HPV vaccine types than men 
. It has been shown that persistent infection is associated with a stronger immune response 
; hence, the higher seroprevalence in women likely reflects a higher degree of natural immunity and possibly an increased duration of the infectious period as compared to men. There is limited evidence for more efficient genital HPV transmission from women to men than from men to women 
, but whether the asymmetry in type-specific transmission probabilities is large enough to offset the asymmetry in the duration of the infectious period between men and women is not clear 
. Our analysis suggests that female-to-male transmission would need to be at least twice as likely in a partnership as male-to-female transmission for male vaccination to be more effective at reducing overall infection levels than female vaccination.
Rules for achieving the most effective reduction in the population prevalence of infection are relevant both for developed and for developing countries. Given that the worldwide burden of HPV-related cancer is concentrated in low-resource settings, HPV vaccines have the potential to dramatically aid global cancer control 
. While prohibitive prices of HPV vaccines are still a major hurdle to populations in greatest need, increased access to cheaper vaccines might soon become a reality following price negotiations and donor support—analogous to hepatitis B vaccine and antiretroviral treatment initiatives in recent history. Rational resource allocation is perhaps even more important in settings with limited resources, especially when the costs of purchasing vaccine are high in relation to other costs. Moreover, achieving the largest reduction in population prevalence is particularly important when a population perspective is employed, rather than the individual perspective commonly adopted with regard to HPV vaccination in developed countries. However, the population-level effectiveness of a single-sex vaccination program may be hindered by the high occurrence of cofactors (e.g., immune suppression and HIV infection) that potentially impede immune responses to vaccination. In populations with a high HIV prevalence, vaccination of both sexes might be needed to substantially reduce HPV transmission.
We have focused on HPV, but our findings are also applicable to other infections. Sex-specific differences in the transmissibility and in the course of infection are the rule rather than the exception in the epidemiology of STIs. These differences have been demonstrated to have implications for the effectiveness of control strategies directed at either sex, with regard to contact tracing to prevent secondary transmission 
, screening to prevent disease and transmission 
, or vaccination to prevent primary infection 
. Here, we have argued that prioritization of prophylactic interventions to the sex with the highest endemic prevalence should be the norm to achieve an optimal reduction in the population prevalence of infection. In this regard, prophylactic interventions need not be restricted to the use of vaccines. Recent modeling studies have evaluated the epidemiological impact on the HIV epidemic of male circumcision and the use of vaginal microbicides 
. These interventions are by definition sex-specific, but they could benefit both sexes even if preventative efficacy would be restricted to one sex only 
. Of note, reducing the female risk of HIV acquisition was found to have the most pronounced effect on population incidence because of the higher HIV prevalence in women as compared to men 
. It remains to be determined whether similar rules of thumb apply to different control modalities.
Our analysis extends previous modeling work on the topic of male HPV vaccination 
. Our analysis adds a fundamental understanding of the impact of current vaccination policies, and the potential benefits of expanding vaccine coverage, by examining vaccine allocation between males and females from a general viewpoint. We used a multi-modeling approach to stress that our findings do not depend on specific modeling assumptions. The generic predictions from a standard model of heterosexual transmission are confirmed by a more elaborate HPV transmission model, which has been developed to predict the long-term impact of HPV vaccination in the Netherlands 
. In addition, the generic predictions for heterosexual transmission are shown to be robust when the model includes a small proportion of MSM in the general population. The results from these different models, when taken together, provide a coherent argument in favor of increasing female vaccine coverage as far as possible, given the limits set by vaccine acceptance and economic constraints. Future research should delineate the extent to which vaccine uptake among girls can be encouraged, and how much benefit will be derived for homosexual and bisexual men from a reduced transmission of HPV in the general population.