rss
Sex Transm Infect 89:336-340 doi:10.1136/sextrans-2012-050707
  • Programme science

Coverage, context and targeted prevention: optimising our impact

  1. Willard Cates Jr2
  1. 1Division of STD Prevention, Centers for Disease Control and Prevention, Atlanta, Georgia, USA
  2. 2FHI 360, Research Triangle Park, North Carolina, USA
  1. Correspondence to Dr Sevgi O Aral, Division of STD Prevention, The National Center for HIV/AIDS, Viral Hepatitis, STD and TB Prevention, Centers for Disease Control and Prevention (CDC), 1600 Clifton Road, Mailstop E-02, Atlanta, GA 30333, USA; SAral{at}cdc.gov, soa1{at}cdc.gov
  • Received 15 June 2012
  • Revised 7 November 2012
  • Accepted 18 November 2012
  • Published Online First 27 December 2012

Abstract

Development of efficacious interventions is only the first step in achieving population level impact. Efficacious interventions impact infection levels in the population only if they are implemented at the right scale. Coverage must be prioritised across subpopulations based on the diversity and clustering of infections and risk in society, and expanded rapidly without delay. It is important to prioritise those who are most likely to transmit infection first.

Over the past few years the HIV prevention agenda has changed drastically. Results of clinical trials demonstrated that, under ideal conditions, acquisition and transmission of HIV can be reduced remarkably. Three clinical trials conclusively showed that male circumcision effectively prevented HIV acquisition in men.1 Three trials suggested that vaginal tenofovir gel and oral tenofovir/emtricitabine if used correctly, could prevent HIV acquisition in women and men who have sex with men (MSM), respectively.2–4 More recently, findings from the HIV Prevention Trials Network (HPTN) 052 trial showed that treating HIV infected partners in discordant couples with antiretroviral (ARV) therapy reduced HIV acquisition among uninfected sex partners by 96%.5

However, these sequential successes, achieved under artificial clinical trial conditions, inevitably call into question the potential for their real world effectiveness and population level impact.6 Preventing the spread of HIV in populations necessitates that the right interventions are delivered to the right people at the right scale; the delivery of interventions is sustained for the right duration; and, where applicable, the adherence of individuals is ensured.6 In fact, as early as 2007, the global HIV Prevention Working Group suggested that, if even those HIV prevention tools available then were brought to scale, half of the infections projected to occur by 2015 could be averted.7

The history of sexually transmitted disease prevention in general provides interesting insights. One review of trials of interventions to prevent transmission of all sexually transmitted infections (STI) including HIV concluded that, although many approaches have been found to be effective against STI including HIV, few have been replicated, widely implemented or carefully evaluated for effectiveness in other settings.8 Despite the availability of efficacious interventions, classical sexually transmitted bacterial infections such as syphilis, gonorrhoea, and chlamydia have survived and even thrived. The failure to adequately scale-up the implementation of available interventions in a timely manner is what allows these STIs to continue to spread.

Determination of the appropriate scale of coverage and the appropriate subpopulation focus involve important strategic programme science decisions. The considerable heterogeneity of context suggests that targeted HIV programming based on the local situation and epidemiology may be the most effective approach to reducing HIV incidence.9 The socioeconomic setting and existing public service capacity determine whether an intervention can have a sizeable outcome in terms of a reduction in a defined risk. The epidemiological context determines whether such risk reduction translates into a measurable impact on HIV incidence.10

In this article we discuss issues related to the scale-up of coverage and the ‘prevention cascade’; the role of context in determining scale-up strategies, and the importance of prioritisation (or targeting) within context in the scale-up of coverage (figure 1). Efficacious interventions impact infection levels in the population only if they are implemented at the right scale. Coverage must be prioritised across subpopulations based on the diversity and clustering of infections and risk in society, and expanded rapidly without delay. It is important to prioritise those who are most likely to transmit infection first.

Figure 1

Scale-up of coverage in context with prioritisation and targeting.

The prevention cascade—the importance of coverage

Development of efficacious interventions is only the first step in achieving population level impact. Implementation of the prevention tool with adequate coverage is the next necessary step. Prevention of mother-to-child transmission (PMTCT) provides a prototypical example. The PMTCT ‘cascade’ identifies the sequences of steps needed to deliver ARV interventions to HIV infected women and their infants: counselling, HIV testing, CD4 testing, dispensing of ARV medications and testing of infants at 6 weeks.11 To optimally reduce the number of infants who become infected with HIV and ensure that mothers receive HIV care and treatment, each step of the PMTCT pathway needs to be delivered with greater than 90% reliability. Introduction of more efficacious ARV interventions will have limited effect unless coverage is addressed adequately.11 To reach the goal of minimal infant transmissions, the overwhelming majority of pregnant women in the community need to complete the full cascade. As services are scaled-up and their quality improved, demand in the population may increase, requiring more services. In cases where demand does not increase on its own, it may be necessary to intervene further to create the demand for services.

Complete (100%) coverage may not be necessary for all interventions in all population contexts. The relationship between intervention efficacy and population impact (defined as change in HIV/STI incidence), is complex, nonlinear and dependent on epidemiological context.10 ,12 Population level reduction in HIV/STI incidence necessitates achievement of required thresholds in coverage; mathematical modelling can help identify these tipping points, which, if reached, can lead to substantial decreases in HIV/STI prevalence and incidence.13 Consideration of required coverage thresholds and their feasibility in specific contexts is an important component of strategic prevention programme planning.14

Scale-up of coverage: beyond the numbers

Scaling up coverage of health interventions involves complex processes beyond simply the number of people who need to be reached. Strategic planning for HIV/STI programme scale-up needs to consider the policy context, delivery mechanisms and resource requirements, as well as the pace of change, sequencing of activities, areas for prioritisation and monitoring and evaluation.15 Multiple other constraints can affect the ability to scale up health service delivery, including a lack of infrastructure and equipment; inadequate drugs and medical supplies; shortage and distribution of qualified staff; weak management and technical knowledge; inadequate demand for services by the populations; and inadequate supervision.16

Cost considerations are especially salient; costing emphasises the importance of determining the existing availability and use of resources as well as the marginal cost of additional infrastructure, equipment and human resources required. Generation of cost estimates also helps identify economies and diseconomies of scale; separate the fixed and variable cost components and include transitional and administrative costs which can constitute a sizeable proportion of total costs in the short run.15

Policies and management at the health sector level and leadership at the national level also affect the ability to scale-up delivery of health interventions. The success of increased coverage of medical male circumcision in Kenya is attributed to the influence of effective national leadership, particularly in contrast to the relatively slow uptake of medical male circumcision in Uganda.17

Whether resources should be used to maximise coverage across all population groups, or directed to target poor and vulnerable groups, poses a trade-off between efficiency and equity.18–20 It is often more efficient to achieve higher levels of coverage by expanding access to easy-to-reach groups (typically upper socioeconomic groups and urban dwellers). The poorest and vulnerable populations are often not reached unless special measures are implemented to cover them. Calculations of the mean coverage index and the coverage gap across socioeconomic strata demonstrate this pattern clearly.20 Thus, scale-up may actually widen inequalities in health outcomes. In the case of STIs, including HIV, this trade-off has further implications for the spread of infection in populations since, more often than not, the most stigmatised groups (such as sex workers, MSM and injection drug users) are also the most likely to acquire and transmit infection.

Scale-up of coverage: the temporal components

There are three temporal components to expanding coverage: duration, timeliness and rapidity. First, issues faced by scale-up efforts change with duration of such effort. Second, timeliness of coverage expansion is critical to reducing population level infectiousness. Third, the rapidity with which scale-up is achieved needs to match and exceed the rate at which infection spreads to achieve a favourable ‘prevalence dynamic’.

The passage of time brings with it evolution of epidemics. Planning for HIV/STI prevention strategic scale-up must take into account the non-linearity of epidemic evolution and the effects of interventions on the parameters driving the epidemic.21 Resource allocation models employ production functionsi to capture the estimated efficacy of interventions in given settings. Production functions allow for non-linearity of intervention effectiveness (eg, level of behaviour change or number of people reached) as a function of investment. For example, as people who are easier to reach or those whose behaviour is easier to change are reached, it may become increasingly costly to continue expanding coverage.21 Thus, challenges of expanding coverage vary over time and need to be taken into consideration in strategic prevention programme planning and resource allocation. Not doing so may result in unintended and unexpected failure to reduce the incidence and prevalence of infections.

Most descriptions of scale-up processes depict coverage as an all-or-none phenomenon. This is appropriate in some cases; for example, in the case of the PMTCT cascade. When interventions are aimed at preventing sexual transmission, the timeliness of coverage may be important as well. The number of days an infectious person is in sexual circulation determines his/her potential to expose sex partners to infection. The ‘person-time-of-infection’ model is a prevention cascade model which depicts the sequence of steps in which delays occur in identifying infected individuals and bringing them to treatment.22 ,23 The sequence includes five steps: (1) lost to detection/resolution of infectiousness; (2) healthcare seeking delays; (3) diagnostic delays; (4) treatment delays; and (5) prevention delays (figure 2). Timeliness of coverage and delays in diagnosis and treatment may be particularly important in the case of HIV prevention in light of the high infectiousness early in HIV infection.24–26

Figure 2

The person-time-of-infection model.

The temporal dimension of rapidity of coverage scale-up is also important in determining whether the rate at which potential transmissions are averted exceeds the rate at which new transmissions occur; and whether the rate at which infectious individuals are removed from the infected pool exceeds the rate at which newly infected individuals are added to the pool. Rapid scale-up of interventions may be needed particularly in contexts where sexual mixing between infected and susceptible individuals is at high levels due to geographical mobility.

The effects of context

The socioeconomic setting and existing public service capacity determine whether an intervention can have a major reduction in a defined risk. The epidemiological context determines whether such risk reduction translates into measurable impact on HIV/STI incidence.10 Epidemiological context is defined as the current state and trends in the behavioural and biological factors that determine the transmission dynamics of a given disease and the impact of a specific intervention. The same intervention carried out in different contexts can have different impacts. For example, the reduction of HIV incidence resulting from improved syndromic treatment of cofactor STIs in Mwanza was not replicated by a community-wide mass STI treatment programme in Rakai.27 ,28 An important cause of these different impacts is likely to have been the different epidemiological contexts.29 Even an intervention with a large intermediary outcome may have minimal impact on HIV/STI incidence if implemented late in the epidemic or targeted at the wrong people.12 Thus, the epidemiological context must be correctly assessed to predict the likely impact of an intervention and determine whether it is potentially transferable to the context in question.

The socioeconomic and public service/health system context may impact the shape of the prevention cascade whether it is measured in terms of individuals ‘lost’ as in the PMTCT cascade11 or ‘person-days-of-infectiousness extended’ as in the person-time-of-infection model.22 ,23 For example, in Kiev, Ukraine, members of most at risk populations are frequently tested for STI including HIV through prevention services free of charge. However, following diagnosis they are referred to specialised clinics where they have to pay for treatments which leads to delays at best and losses at worst.

Heterogeneity of context

Heterogeneity in individual risk of HIV infection has been described as an underestimated problem in the design and interpretation of randomised control trials that test prevention methods of HIV heterosexual acquisition.30 In the context of high HIV incidence, high heterogeneity can lead to a substantial underestimation of the impact of an intervention and reduced statistical power. Heterogeneity of treatment effects in general has been recognised as a problem in the literature.w1 w2 Applying global evidence (average effects measured as population means) to local problems (individual patients or groups who may depart from the population average) is problematic. Heterogeneity in intervention effects or context heterogeneity is considered to be analogous to patient level heterogeneity in treatment effects and refers to differences in intervention effects across sites with different contexts.w3 Heterogeneity in intervention effects is specified in quantitative terms as interactions between intervention and contextual factors.w3 w4

Recent HIV prevention research has increasingly focused on context heterogeneity and its implications for an intervention's effectiveness.9 One model examined the impact on the HIV epidemic of treating discordant couples with ARVs to prevent transmission in four countries.w5 The results showed that the higher the HIV prevalence and/or the greater the percentage of couples in discordant partnerships the more incidence would be reduced. Country-specific differences in the number of infections prevented result from a complex interaction among three factors: population size, HIV prevalence and the percentage of couples that are discordant. In this comparison of four countries the model predicted that Malawi would benefit the most from the intervention even though it does not have the greatest population size, the highest HIV prevalence or the highest degree of discordancy. The complex interactions that define the demographic and epidemiological context play a major role in determining population impact of interventions.

Uneven distribution of risk and outcome in populations

Many parameters relevant to HIV/STI transmission dynamics are unevenly distributed in populations. All parameters display some degree of concentration, including prevalent and incident HIV/STI cases and rates, recent seroconversions, numbers of partners and numbers of sex acts. In a widely disseminated rural South African epidemic, a recent study found the existence of several localised HIV epidemics of varying intensity, partly contained within geographically defined communities.w6 In the Bagalkot district of Karnataka in South India, 15% of the villages accounted for 54% of all rural female sex workers.w7 In the USA, 20% of women and 24% of men accounted for 60% and 61% of vaginal sex acts in the past 4 weeks, respectively.w8 Similarly, 20% of women and men accounted for 47% and 57% of opposite sex partners in the past year, respectively.w8 Finally, in the USA, based on county level analyses, 20% of the population accounts for 39% of chlamydia, 52% of gonorrhoea and 64% of primary and secondary syphilis cases.w9

Prioritisation and targeting

The uneven distribution and resulting patterns of concentration of HIV/STI incidence, prevalence and risk presents prevention opportunities. Expansion of intervention coverage should be prioritised across subpopulations by key population vulnerability, risk behaviour or geography. In the context of concentrated epidemics, during early epidemic phases, targeted interventions focused on those most at risk of acquisition improve efficiency, population impact and cost effectiveness.12 In the context of generalised epidemics, the value of targeted prevention interventions is often not considered.w6 Identification of the localised clustering of incident infections or ‘hot spots’ may have immediate implications for where to situate new, or intensify existing prevention programmes, even in the context of generalised epidemics.w10

Results of mathematical modelling demonstrate that, irrespective of strategy, prioritisation of subpopulations helps to increase population impact and cost-effectiveness of interventions while decreasing their cost.w11 While prioritisation is more effective in concentrated epidemics and when coverage levels are low, it also improves population impact in hyperendemic epidemics, though not substantially.w11

HIV prevalence is the most commonly used measure to prioritise communities for HIV prevention. However, HIV incidence (for prevention of acquisition) and HIV transmission probability (for prevention of transmission) may be better measures of prevention need and should be considered in the prioritisation of communities for prevention.w12

Concurrent with the patterns of concentration of prevalence, incidence and risk in subpopulations, considerable diversity exists even within these subpopulations. For example, MSM vary by number of partners per year, number of sex acts per partnership, proportion of sex acts protected by a condom and sexual positioning (insertive, receptive, versatile) per sex act. Sex workers vary by number of clients, type of clients, proportion of sex acts protected by a condom and sexual practices (vaginal, oral, anal). Such variability suggests that even more refined prioritisation of subpopulations may be called for in planning scale-up of prevention interventions. Application of prevention criteria in targeting interventions may necessitate that public health proactively reaches out to those HIV infected people most likely to transmit the infection to others.w13

A focus on potential future transmissions suggests that prioritisation be determined by proportion of expected transmissions that will be caused by current cases rather than the proportion of current cases in key populations. While information on the epidemiological context reveals where new infections happen at the present time, information on the sexual contact structure suggests where the new infections will happen in the future. A sexual network based approach to prioritisation and targeting would imply that population strategies for prevention vary depending on network structure.w14 Subpopulations, individual members of which are highly connected to each other through dense sexual contact links, may be targeted prioritising interventions to these groups. In addition, interventions aimed at breaking the sexual ties (eg, ‘zero grazing’, ‘delayed sexual debut’) may be delivered to these subgroups.

Conclusions

HIV/STI prevention science has only recently started focusing on issues of coverage, scale-up strategies, prioritisation and targeting of interventions and the importance of contextual characteristics such as concentration and diversity. The field poses many challenges, the learning curve is bound to be steep and the nature of the public health demands highlights the timeliness of the programme science agenda.

Acknowledgments

The authors thank Fred Bloom and Patricia Jackson for their outstanding support in the preparation of this article.

Footnotes

  • Contributors SOA and WC planned the outline and wrote the article.

  • Disclaimer The findings and conclusions in this paper are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.

  • Funding WC is funded by: Preventive Technologies Agreement, GHO-A-00-09-00016-00; HIV Prevention Trials Network Coordinating and Operations Center, 1 U01 A1068619-01; Microbicides Trials Network Operations Center, 1 U01 A1068633-01.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i In economics a production function is a function that specifies the output of a firm, industry or a system for all combinations of inputs.

References

Free sample

This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of STI.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.