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

Modelling

Is increasing gonorrhoea resistance in MSM is a result of more treatment, rather than greater sexual activity?

20 Jul, 16 | by Leslie Goode, Blogmaster

Emerging antibiotic resistance to the last-ditch treatment of Neisseria gonorrhoeae compels health policy-makers to balance opposing concerns.  On the one hand, successfully combating spread of the infection requires targeted treatment of core-group individuals.  On the other, a focus on the core-group causes a rebound in core-group incidence, with maximal dissemination of resistance (Chan & McCabe/STIs (C&M); Chan & Fisman/STIs).

Recent public health orthodoxy has tended to favour the more intensive screening of core-group individuals (Ison & Unemo (STI); Giguere & Alary/STIs; Lewis/STIs).  However Fingerhut & Althaus (F&A), in a recent modelling study, seek to shift the balance in the opposite direction.  They claim their model demonstrates that the wide disparity in the spread of resistance spread as between populations of MSM and of HMW (heterosexual men and women) reflects differing levels of treatment rather than differences in sexual behaviour (‘more sexual partners’).

So far as concerns the first part of the claim (‘gonorrhoea spreads faster with more treatment’), F&A’s findings corroborate those of C&M.  However, in coupling this with the claim that gonorrhoea spread is not the result of sexual behaviour (‘gonorrhoea (does not) spread faster with more sexual partners’) they place the balance of responsibility for spread with the prevailing policy of treatment.  This is presumably intended to push policy makers in the direction of a more conservative attitude to targeting testing and screening.

But can F&A really justify this  change of emphasis by differentiating the respective contributions of ‘more treatment’ as against ‘greater sexual activity’ to the difference in resistance between MSM and MSW popultions ?  We are wrong, the authors argue, to assume that ‘more partners’ amongst the MSM population necessarily entails more transmissions (p. 11) – and their model apparently demonstrates this.   A common sense response, however, would be to object that ‘more partners’ presumably implies ‘more sex acts with more partners’ – and that, even if ‘more partners’ does not in itself entail more transmissions, ‘more sex acts with more partners’ might certainly be expected to do so.

Interestingly, Althaus in another paper (see Althaus & Alizon) – admittedly, in connection with heterosexual groups – corroborates our common sense expectation by showing that the number of partners displays, if not a proportional, then at least a linear, relation  to number of sex acts. So can it really be the case that there is not a greater number of transmissions amongst the MSM population, given the greater number of partners? The authors evidently believe not.

Nevertheless, it would be interesting – as well as pertinent, I suspect, to the goals of the study – to have a more satisfying explanation of why, here, as elsewhere, common sense turns out to be wrong.

 

 

 

Mathematical models say: switching to HPV nonavalent vaccine brings cost benefits.

20 Jun, 16 | by Leslie Goode, Blogmaster

STI journal issues of nearly a decade ago, when HPV vaccination was a relatively new thing, hosted a discussion on the issue of which vaccine to choose. The choice at that time, readers will remember, was between GSK’s Cervarix 2vHPV and Merck’s Gardasil 4vHPV (Morris/STIs)*.  Now, the introduction of a third alternative, Gardasil (9vHPV), seems to have fuelled a similar burst of activity amongst mathematical modellers – at least in the US, where the new vaccine was licensed in 2014.

Gardasil 9vHPV elicits immunity to nine oncogenic (i.e. associated with cancer) serotypes – i.e. five more serotypes than Gardasil 4vHPV, and seven more than Cervatrix 2vHPV.  The nonavalent vaccine (9vHPV) is expected to extend protection from >66% to 80% of cervical cancers.  It will also, it should not be forgotten, have some benefit in preventing HPV-related oropharygneal cancers (Field and Lechner/STIs).  However, Gardasil 9vHPV is approximately $13 per dose more expensive than Gardasil 4vHPV, and $18 more expensive than Cervatrix 2vHPV.  In 2015 the US Centers for Disease Control and Prevention (CDC) recommended vaccination with any of the three alternatives for females aged 9-26yrs, and with 4vHPV or 9vHPV for males aged 11-21yrs.

Once again, then, the question of the relative cost-effectiveness of HPV vaccines raises its head – this time in the US, and in connection with a possible switch from 4vHPV and 2vHPV to 9HPV as the vaccine of choice. Mathematical modellers in the US have risen to the challenge in at least two recent studies.  Brisson & Markowitz conclude that making the switch would be cost-effective ‘under most scenarios’ (Chesson & Saraiya/STIs; Brisson & Markowitz).  Now, Durham and Galvani (D&G), in another US modelling study, have reached the same conclusion.  Not content, however, with a response for current levels of vaccination coverage, they also consider the impact on cost-effectiveness of raising national coverage with 9vHPV to higher levels.  This requires them to take into account the effect of herd-immunity, which ensures that returns on investment diminish to the extent that higher levels of coverage have already been achieved.  They also consider the relative cost-effectiveness of distributing the investment in 9vHPV vaccination in such a way as to bring up the vaccination levels in states where it is low (e.g. Arkansas, Missisipi, Missouri, Kansas) towards the levels already achieved in other states (e.g. Illinois, Montana, N. Carolina, Washington DC), as against that of an even distribution.  (At present, state vaccination rates vary between 20-57% for females, and between 9-43% for males – though inter-state migration rates are such that 29-84% of the long-term health benefits of vaccination will be realized by beyond the boundaries of the state where vaccination took place).

The findings of the study are as follows.  9vHPV is cost-effective – as compared with the alternatives – at any level of coverage.  Comprehensively switching to 9vHPV would yield the same benefit as raising levels of coverage with existing vaccines across the population of the US by 11%.  Second, assuming a comprehensive switch to 9vHPV: a national increase in coverage of 10% would show an incremental cost-effectiveness ratio (ICER) corresponding to a willingness to pay (WTP) of $40,000 per QALY, and increases of 20%, or 40%, ICERs equivalent to WTPs of $53,000 and $106,000, respectively.  Finally, the figure of $40,000 WPT per QALY given above represents only an average, since, in practice, the cost-benefit of an increase of 10% in coverage would differ widely between states with low current levels of coverage, like Arkansas, where the cost-benefit would be around $13,500, and states with high levels, like California, where it would be around $56,400. The authors therefore advocate focussing the investment needed to achieve increases in coverage on states that currently have low levels of coverage.

  • a previous version of this blog mistakenly mistakenly gave the names of the manufacturers of Gardasil and Cervarix as GSK and Merck respectively.  The mistake has recently been brought to our attention, and the manufacturers as given in the emended blog (23.6.16) are the correct ones. (Blogmaster)

Modeling the potential effectiveness of PrEP as against other preventative interventions in addressing MSM HIV

1 Mar, 16 | by Leslie Goode, Blogmaster

 

Despite the known preventative benefits of ART, the incidence of HIV among UK MSM population has remained relatively constant over the last 10 years and looks set to remain so. The UN 90:90:90 target will soon be achieved for this population, yet the goal of eliminating the infection seems no nearer.  Not surprisingly there is an appetite among health professionals for alternative measures – like PrEP.

A recent study (Punyacharoensin & White) claims to be the first to model, on the basis of detailed behavioural and surveillance data, the  impact of seven potential interventions – including PrEP – on HIV incidence in this population.  There is a considerable diversity in the nature of these interventions.  One of them consists in the roll-out of the recent WHO policy of ‘diagnose and treat’: others in the achievement of specific targets, such as yearly HIV testing for a given percentage of MSM: others again in the fulfilment of what seem little more than aspirations, given the absence of any indication of how they might be achieved – such as halving one-time partners or unprotected anal intercourse. The potential impact of each intervention is assessed independently, and then in various combinations with other interventions.

It is evident from the way that the various combinations of interventions are grouped into two sequences that the primary issue for these authors is the potential of PrEP (which happens to be the only genuinely new tool in the box).  The first sequence combines PrEP at varying levels of coverage and effectiveness with the achievement of one-year testing for different proportions of the population, ‘test and treat’, and various levels of risk compensation.  The second sequence goes through much the same process but with PrEP now replaced by a putative 0.5 drop in repeat partnerships.  The main study outcome appears to be the demonstration that introducing PrEP at certain levels of coverage and effectiveness could, in certain combinations, have an impact on incidence (-43.6%) that would be of the same order (-41%) as the achievement (by unspecified means) of a putative 0.5 drop in repeat partnerships.

The biggest problem for the modelers, of course, is the difficulty of predicting the effectiveness of PrEP in a real-life setting.  When empirical evidence of this finally emerges, then the value of this intervention as against traditional interventions will presumably be determined by its relative cost-effectiveness.  In other words, the question to confront health policy makers will be how much of what they would otherwise have spent on traditional interventions should be diverted into PrEP.

In the meantime, if there is key message to emerge from this study, it is that PrEP is worth a try.

Increased HIV infectivity in the acute phase of infection may be a less important factor in HIV transmission than we thought

12 Jun, 15 | by Leslie Goode, Blogmaster

Assessing, as far as we can, the preventative impact of ART on HIV transmission dynamics is evidently very important – both to inform judgments about ART initiation (Wayal & Hart (STI); Cohen (STI)), and also, at the policy level, to be able to evaluate the possible preventative gains of ART scale-up (Shafer & White (STI); Boily & Mishra (STI)).   One important piece of the jigsaw is the impact of ART on sexual behaviour.  This has been discussed by a number of recent studies (Wayal & Hart (STI); Hogben & Ford (STI); Shafer & White (STI)).  Another piece of the jigsaw is the impact of ART on HIV infectivity.  Of particular concern here are the relatively high levels of infectivity that occur in the period immediately after infection.  In view of this, investigators have stressed the importance of the earliest possible initiation of therapy, if the full preventative benefits of ART are to be enjoyed (Cohen (STI)).

The recent study, Bellan & Meyers (B&M), addresses itself to this second, important but potentially less easily investigable piece of the jigsaw. They observe that investigators have tended to proceed on the basis of the known relationship between viral load and infectivity. Empirical evidence of relative infectivity of acute versus chronic phases of the infection is practically unobtainable, for various reasons.  For a start, newly-infected individuals are rarely diagnosed in the acute phase and, if infected by stable partners may provide no evidence on onward tradition; if susceptible non-infected partners are at risk, then, clearly, ethical guidelines dictate that further transmission be stopped – not investigated.  According to B&M, most subsequent studies have relied for direct epidemiological measurement of acute phase infectivity and duration on a retrospective cohort in Rakai, Uganda (Wawer & Quinn; Hollingsworth & Fraser). B&M reassess previous analyses of this evidence.  They find significant bias – especially in two areas.  The first has to do with the neglect of the contribution to total risk of couples who were censored from the cohort owing to couple dissolution, loss to follow-up or study termination.  The second concerns the extent to which some of the estimated difference in risk between the acute and chronic phases may reflect heterogeneity in the risk behaviour of those couples entering the study sero-discordant, as against those entering it sero-concordant negative.

The findings of B&M are intriguing. They argue that combined effect of these sources of bias in earlier analysis of the Rakai evidence has been enormously to inflate estimates of relative acute phase – relative to chronic phase – HIV infectivity. B&M estimate the relative hazard of transmission during acute phase at 5.3, the acute phase duration at 1.7 months, and the “extra-hazard months” contributed by the acute phase (a measure adopted by the authors in order to ensure comparability of study results) at 8.4. Previous estimates give levels of increased infectivity due to acute phase which are equivalent to between 31 and 141 hazard months. If the results of B&M are confirmed in subsequent studies, the preventative gains of ART scale-up could be greater than hitherto supposed.

Achieving HPV herd immunity cost-effectively. When does it make sense to allocate resources preferentially to boys?

23 Apr, 15 | by Leslie Goode, Blogmaster

Recent empirical studies of HPV vaccination have provided evidence that marginal vaccination costs increase with coverage.  Let us take into account – they argue – not just the vaccine price, but the cost of education and outreach programmes that would be needed so as to reach the yet unvaccinated population.  If we do so, we are likely to find that raising the vaccination rate for pre-adolescent girls, let us say, from 40-41%, proves considerably more expensive than raising it from 20-21%.  This, in turn, raises the question whether – given the achievement of herd immunity is the ultimate goal – resources are necessarily best allocated when directed to the female rather than the male pre-adolescent population. Could it even be that – at certain levels of coverage and a certain rates of increase in marginal vaccination cost (for girls and for boys) – resources might be more effectively allocated to the vaccination of pre-adolescent boys? Ryser & Myers in a recent study seek to model the impact of marginal cost increase in order to answer this question.  In the case of US, at least – with rates of vaccination standing currently, for girls and boys, at, respectively 37% and 13.9% – they argue in cost benefit terms for re-directing resources to boys.  However, the question is no doubt relevant to other countries in which HPV has been introduced, but levels sufficient for herd immunity, have not yet been achieved.

Optimal allocation of new resources as between girls and boys for a given level of vaccination is indicated in the diagrams on p.40.  Marginal vaccination cost increase is estimated at a higher, a lower, and zero level.  What is striking is the radically different patterns of optimal allocation between girls and boys in those three scenarios.  At the very least, the results challenge the orthodoxy of the superior cost benefit of female vaccination. They also indicate the importance of further empirical research into marginal vaccination cost increase.

The limitations of the study are largely due to the assumption of a closed sexual network of 14-18 yr old heterosexual adolescents.  The attempt is made to factor in the impact of various complicating factors, such as the assortativity of vaccine uptake with sexual activity, the likely presence of additional relationships between females inside, and older males outside, the network, and asymmetric vaccination cost curves as between girls and boys.  But the most serious limitation was beyond the power of the study to address.  This is the restriction of the modelled network to heterosexual relations, and the exclusion from consideration of a large number of HPV-related conditions such as anal and oro-pharyngeal cancers which have higher incidence among MSM and HIV-infected individuals (Lawton & Asboe (STIs); English & Pourbohloul (STIs)).  One wonders, therefore, whether the case for male HPV vaccination is not a great deal stronger than might appear from this paper.  At all events, a case is made that, even when these conditions are excluded, a greater allocation of HPV vaccination resources to males may be justifiable – e.g. currently in the US.  For the impact of the recent extension of the HPV vaccination programme to males in Australia (the first country to have taken this step (2013)), see Korostil & Donovan (STIs).

Are African HIV epidemics sustained by exogenous introduction of infection?

24 Apr, 14 | by Leslie Goode, Blogmaster

What is the relative importance of exogenous and endogenous transmission in sustaining HIV epidemics?  In a study of HIV sub-type distribution in the Middle East, Mumtaz & Abu Raddad (STIs) stress the role of multiple exogenous introductions, as evidenced in the wide diversity of genetic sub-types present in most countries.  At a more local level, the answer to our question will, of course, depend on how “exogenous” and “endogenous” are defined – and may have little meaning where we are concerned with the  HIV transmission networks in gay communities that are the object of a number of studies featured in STIs (Potterat & Muth (STIs); Drumright & Frost (STIs)).  Yet, the situation is surely very different when it comes to the kind of geographical communities that are constituted by the studies designed to evaluate the local epidemiological effect of ART deployment – such as HPTN 071 study in Zambian and South Africa.  Here, the location of the communities targeted by the trial establishes a clear boundary, and gives meaning and importance to our question.  How far are the preventative effects of ART coverage within the community likely to be neutralized by introductions that are exogenous to it?

Grabowski & Ray, in recent analysis of data deriving from the Rakai Community Cohort Study, Round 13 (2008-9), has sought to give some insight into the spatial dynamics of HIV transmission, through investigating a cohort of 14,594 individuals in 46/50 communities in the Rakai region of Uganda.  Its goal: to determine the relative epidemiological importance, in this particular context, of transmissions within the household, across the boundary of the household but within the community, and across boundary of the community.  A difficulty foreseen by the study is that transmissions from outside tend to be less easily traced.  The researchers have therefore adopted a multi-faceted approach:  an analysis of the spatial clustering (1) and a phylogenetic analysis (2) are complemented by an analysis of individual partnerships on the basis of data supplied at interview (3).  The phylogenetic analysis investigated the relationship of phylogenetic clustering (in terms of genetic closeness in the gag and env genes) to geographical location.

The findings of the study suggest the relative importance of repeated introductions of HIV across the community, and indeed, regional boundary, as against the importance of onward transmission through intra-community networks (other than those within the household).  The spatial clustering analysis shows very strong household clustering (RR of HIV+ person, as opposed to non-HIV+ person, being in the same household as another HIV+ person 3.2, and RR 10.8 for incident cases), but practically no clustering outside the household.  The phylogenetic analysis identified 95 clusters, of which 53 (55.8%) spanned households; of these 53, 38 (71.7%) crossed community boundaries; of the 38, 18 (47.4%) spanned geographic regions.  The individual transmission analysis shows 39.5% of new cases from extra-household partners; of these, 62.1% were from partners outside the community; of these (where location of partner was known) 50% outside Rakai district, and geographically dispersed throughout Uganda.

Taken together, the three analyses seem to offer a consistent picture. The surprise is the importance of more exogenous, as against more endogenous, transmissions, with intra-community transmission (excepting within the household) not playing the role that might have been expected.  Of course, these findings may not be generalizable to other sub-Saharan, still less non-African, settings.  But they do raise pertinent questions  to any attempts to evaluate the preventative possibilities of localized ART interventions.

 

What can cost-effectiveness modelling tell us about the feasibility of eliminating congenital syphilis in sub-Saharan Africa?

20 Nov, 13 | by Leslie Goode, Blogmaster

The WHO global initiative for the elimination of congenital syphilis (2007) set the goal of expanding antenatal testing to >90% by 2015.  In sub-Saharan Africa (SSA) recent estimates place the number of mothers infected with active syphilis at 535,000 p.a..  Adverse outcomes – stillbirths, neo-natal morbidity and congenital disease – affect 53%-82% of these pregnancies, as compared with 10%-21% of women without syphilis.  Yet, perhaps surprisingly,  74% of pregnant women in SSA are reported to attend an antenatal clinic at least once.  Administration at the routine visit of a simple point-of-care test (POCT), plus, in the case of detection, an intra-muscular injection of benthazine penicillin, is all that, in most cases, would be required to deal with the problem.

A recent modelling study (Kuznik & Manabe ) seeks to make the economic case for universal POCT syphilis screening in 43 countries in SSA.  It estimates the cost of increasing syphilis testing at antenatal clinics from whatever it is at present to 100% through universal adoption of the immunochromatographic strip test (ICT), and the benefits that would result in terms of saved lives and reduced morbidity.  It then calculates the cost-benefit in US dollars per Disability Adjusted Life Year (DALY).

The findings are given for each of the 43 countries considered. Here are the aggregate findings for SSA as a whole.  In order to ensure screening coverage for the 23.5 million (74%) pregnant mothers attending ante-natal clinics, the cost is estimated at a comparatively modest US$20.8 million per year, and would, it is claimed, reduce incidence of still-birth, neo-natal death and congenital syphilis by 64,000, 25,000 and 32,000, respectively.  Of the 43 countries in the model the cost of such an intervention per DALY averted would be less than US$20 in 37 cases, and less than US$10 in 23 cases.

The study claims to be the first to have evaluated the cost-effectiveness of ICT across SSA.  What is the value of this exercise?  It seems to consist primarily in the extra rhetorical “punch” it can lend to the argument in favour of doing something – and something very cheap – in order to alleviate the deplorable ravages of perinatal mortality and morbidity due to syphilis.  One can only applaud the intentions of its authors.  At the same time, one is struck by the inadequacy of the measure employed (cost per DALY ) to capture any real sense of the economic (let alone human) cost of MCST.  The impressive figures they arrive at for cost-effectiveness of the proposed interventions owes not a little to the fact that the lost life-years averted happen to be those of the stillborn or neonatal deceased.  This does not belittle the cost, but feels a rather odd way of looking at perinatal outcomes.  Consequently the whole exercise, while it may be well-intentioned and ticks all the official boxes, strikes this reader at least as rather specious.  One wonders if it can offer any real ground for the prioritization of MCST as against other equally laudable interventions: if it does, one feels somehow that it ought not to!

Relevant to the real economic cost confronting any country deciding to switch to the ICT would presumably be the relative cost of the ICT compared to an established alternative (e.g. Rapid Plasma Reagin (RPR)).  After all, using ICT to make up the shortfall in current testing levels presupposes the country has made the switch from RPR to ICT – which may already have cost implications (Vickerman and Watts).  Ultimately, however, the main the challenge of making this switch, may not be economic cost at all, but a problem of institutional organization and training (STI blog/Peeling & Mabey).   Peeling & Mabey observe in relation to one area in which POCT for syphilis was being introduced as part of a trial “a 65% drop over the first six months in the percentage of antenatal clinics passing the quality assurance controls due to high staff turn-over: but, subsequently, with the HWC training mechanism kicking in, there was a return to 100% levels of proficiency”.  Here and elsewhere (Peeling & Mabey) make the case for the importance of “programme science” to “address the gap” between test performance and successful deployment.

Modelling ART impact on HIV prevention within discordant couples

15 Dec, 11 | by Leslie Goode, Blogmaster

The HPTN 052 study, discussed in an earlier blog, appears to offer a godsend:  a demonstrated 96% reduction in HIV transmission in discordant couples using antiretroviral therapy (ART).   Apparently, the extension of ART conveys an additional, and unexpected, benefit in its potential impact on prevention.  But how do we quantify the additional benefit?

A recent concise communication in the journal AIDS claims that the benefit depends very much on the country you live in.  They claim that mathematical modelling can help us pick out the countries where the benefit is greatest – and consequently most likely to outweigh the various costs and dis-benefits of ART extension.  Feed into their model the three factors of population size, HIV prevalence and the percentage of discordant couples, and it will throw up results that common sense would not have predicted – as, for example, that Malawi, with its potential combination of a reduction in HIV incidence and large number of prevented infections, would benefit more than Ghana, Rwanda or Lesotho (the other countries featuring in the study).

The most striking finding of this modelling exercise is the importance of the stability of discordant couples. When the model estimates stability at 40% or 70% respectively there is a dramatic change in effectiveness of ART extension as a prevention strategy.    Even at the higher estimate, however, the prevention of HIV within discordant couples does not result in noticeable decline in national incidence overall, where prevalence is at the levels seen in Rwanda and Ghana.  This only occurs where incidence reaches higher levels seen in Malawi and Lesotho. Evidently the extension of ART, even to a large proportion of discordant couples, offers no panacea.

It is possible that a modelling exercise of this kind might enable policy makers to pick out the countries where the benefits of ART extension would be greatest.  But so far as the countries considered by the paper are concerned (Malawi, Lesotho, Rwanda, Ghana), the presentation of data in this paper is blatantly misleading.  Why do the authors give equal weight, for each country, to the absolute number of infections prevented alongside decline in incidence?  What purpose is served by treating the former as an independent factor in determining the benefit derived by each country, when there are such differences in the size of their respective populations?  We soon discover.  The authors’ conclusion that Malawi would derive greater benefit than Lesotho appears to be entirely based on the fact that in the former case the decline in incidence will be accompanied by a much higher number of infections prevented.  This leaves out of account the relative size of the two populations – and the population of Malawi is almost ten times that of Lesotho!

Have the authors rushed this paper out in response to the results from  HPTN 052?

On HPTN 052:

http://www.hptn.org/web%20documents/PressReleases/HPTN052PressReleaseFINAL5_12_118am.pdf

Wafaa M. El-Sadr, Brian J. Coburn and Sally M. Blower, “Modeling the impact on the HIV epidemic of treating discordant couples with antiretrovirals to prevent transmission”:, AIDS Volume 25, issue 18, 28th November 2011

http://journals.lww.com/aidsonline/Abstract/2011/11280/Modeling_the_impact_on_the_HIV_epidemic_of.12.aspx

Sexually Transmitted Infections blog

Sexually Transmitted Infections

Discussion and suggestion space for readers of STIs.
Visit site



Creative Comms logo

Latest from Sexually Transmitted Infections

Latest from Sexually Transmitted Infections