October 1997, Manila, Philippines

The Status and Trends of the HIV/AIDS/STD Epidemics
in Asia and the Pacific

4. Monitoring the HIV/AIDS epidemics

Methods for the monitoring of the HIV/AIDS pandemic in the Asia-Pacific region are, in general, no different from methods used in other regions. However, a diversity of HIV epidemics in this region requires adapting HIV/AIDS surveillance methods to measure Asian Pacific HIV/AIDS/STD patterns and prevalence levels. The following sections describe HIV sentinel, behavioral and STD surveillance in the Asia-Pacific region. In addition, the uses and limitations of HIV/AIDS modeling for forecasting and scenario development in this region are described.

4.1 Epidemiological Surveillance for HIV/AIDS and STDs

The evolution of surveillance methods

The HIV epidemic first appeared as a range of opportunistic infections and other diseases defining AIDS, and the counting of new AIDS diagnoses became a central component of HIV/AIDS surveillance systems. However, there were several important limitations. First, AIDS counts provided little information about current patterns of HIV transmission because they represented infections acquired years in the past. Second, AIDS diagnosis required clinical and laboratory expertise that was absent in many countries, particularly those of the developing world. Third, AIDS counting suffered from the limitation that health care workers did not accord a high priority to the passive reporting of cases. On the other hand, AIDS reporting did have the potential to graphically demonstrate the impact of HIV infection in terms of advanced illness, even if it was restricted in most countries to the first episode of illness that satisfied the AIDS case definition.

The advent of the HIV antibody test in 1985 opened up the possibility of directly measuring the prevalence or incidence of infection in populations. Countries of the Asia-Pacific region came up with a variety of approaches to monitoring HIV prevalence. Some countries relied on routine voluntary HIV testing as a means of monitoring prevalence while other countries established comprehensive systems that involved sampling a wide range of population groups at regular intervals. In a third group of countries, prevalence was measured through occasional surveys that were implemented as resources and necessity dictatated. The World Health Organization recommended that measurement of HIV prevalence be done using the anonymous unlinked approach which was designed to minimize the selection bias associated with voluntary testing and to protect the confidentiality of participants. Methods for serological surveillance were greatly refined in the era of HIV infection with the conduct of large-scale regular surveys in a wide variety of populations (see Box 6).

Meanwhile, there was a growing realization of the crucial link between control programs for HIV infection and those for other STDs. Monitoring of the other STDs was generally not well done since the diagnosis usually required a genital examination making special surveys expensive and logistically difficult. Furthermore, STD reporting was even more prone to underdiagnosis and underreporting than AIDS reporting since many people with STDs are asymptomatic or obtain treatment through pharmacies or other venues that do not report cases. Therefore, most countries have relied on some form of passive surveillance system for STDs.

Despite these major challenges, most countries now accept the need to monitor HIV and AIDS in a manner that is epidemiologically valid and respects confidentiality. In virtually every country, systems have been established for AIDS case reporting and in most countries the prevalence of HIV has been reasonably well characterized among the higher risk populations.

There nonetheless remains a number of issues surrounding the surveilance of HIV, AIDS and STDs that have not been well addressed. These are discussed below.

Continuing confusion in some countries about the role of surveillance

The role of surveillance is to provide information to support the development and assessment of health programs. In some countries of the Asia-Pacific region, there is confusion about the role of surveillance. In particular, there remains a view in some sectors that one of the roles of surveillance is to detect people with HIV infection for the purpose of subjecting them to direct intervention of some kind, sometimes even involving law enforcement activities.

Implementation of HIV sentinel surveillance in the Asia-Pacific region

HIV sentinel surveillance in the Asia-Pacific region does not require special procedures or methodologies. However, there are some specific issues that should be taken into consideration in implementing HIV sentinel surveillance in the region.

Implementation of HIV sentinel surveillance in the Asia-Pacific region will require a careful selection of sentinel groups, based on their relevance to the epidemic, their accessibility and the feasibility of the study.

Traditionally, the recommended groups for HIV sentinel surveillance were STD patients (high-risk behaviors) and antenatal clinic attenders (low risk or as a proxy of the general population). In addition, data could be available from blood donors and limited serosurveys in sex workers. This methodology was developed looking primarily at the African epidemic and the reality of the African countries, where STD patients are readily available in most health facilities and antenatal screening for syphilis is the most common test for pregnant women.

In several Asian countries, STD patients are rarely seen in the public health facilities. For confidentiality or other reasons, they seek treatment at private practices or prefer to try self-medication. The small numbers and the implicit selection bias make them a difficult group to monitor for HIV surveillance. In addition, antenatal clinic care does not usually include a syphilis test at the first visit (probably due to the low prevalence of syphilis). When blood is collected from pregnant women, it is usually only a small drop for hemoglobin evaluation. This makes unlinked anonymous testing in this group extremely difficult.

On the other hand, sex workers, who are mostly floating and difficult to access in Africa, are often organized in well established brothels or massage parlors in Asia and the Pacific. In this setting, HIV sentinel surveillance can be easily implemented if blood is regularly taken for screening of other diseases, e.g., syphilis. Military recruits have also been used successfully in Thailand to monitor the HIV epidemic. Migrant workers, sailors and fishermen can also be considered for HIV surveillance in specific situations. Blood donors remain a useful sentinel group in most populations that have universal HIV screening programs.

Appropriate analysis of HIV sentinel surveillance data requires the calculation of confidence intervals (90 percent or 95 percent) around the prevalence estimate. Confidence intervals are essential statistical measures of error and should be included in the analysis and reports of HIV sentinel surveillance. Finding them in surveillance reports is, however, the exception rather than the rule. Attempts should be made to be more scientific and in presenting results of HIV sentinel surveillance indicate both the relevance and the limitations of the data.

Can HIV sentinel surveillance monitor trends in a "slow epidemic"?

Data available from several Asian countries, including the Philippines, Indonesia, Bangladesh and China, show HIV prevalence rates increasing very slowly, even in high-risk behavior groups such as sex workers. This contrasts with the exponential increase seen in other populations. Several reasons have been proposed for this pattern, including the relatively low intensity of risk behaviors, particularly commercial sex.

Under these circumstances, analyzing these trends poses a special challenge to the surveillance system because HIV sentinel surveillance can only detect relatively marked increases in prevalence in a short period of time. If the curve is flat, the confidence intervals will overlap for several years and it will be almost impossible to tell whether HIV prevalence is actually progressing, stabilizing or even decreasing.

More samples do not always mean better surveillance

It is commonly accepted that, for meaningful interpretation, results of sentinel surveillance cannot be aggregated between different sites, except under the very unusual circumstance where it can be assumed with confidence that the sites draw from populations uniformly affected. Since this is rare, it is not recommended that data from separate sentinel sites be aggregated. Prevalence rates should be calculated separately per site and sentinel group. If a summary statistic is required, the most appropriate would be a median value with range of rates, or a confidence interval, for any given type of sentinel site.

Incidence monitoring

Of central importance in targetting or assessing prevention activity is the rate at which new HIV infection is occurring, or the incidence. Apart from direct measurement through longitudinal surveys, which are expensive and difficult to sustain, HIV incidence can be estimated from back-projection of AIDS cases, or from repeat cross-sectional surveys. At present, few countries are attempting to regularly and systematically estimate HIV incidence using one or more of the available methods. Although some expertise in modeling and statisics is available in many countries, it is often underutilized in the estimation of HIV incidence. Another measure of HIV incidence can be obtained through the monitoring of HIV prevalence in young age groups.

Monitoring of advanced manifestations of HIV infection

A good understanding of the spectrum of advanced HIV disease is essential for the planning of support and care services, and for assessing the benefit over time of new therapeutic and prophylactic agents. Monitoring of disease spectrum need not take place on any wide scale; it is more practical to select a small number of experienced clinical sites, to regularly report on advanced treatment manifestations, and patterns or therapy.

AIDS case definition

Countries in the Asian Region have so far used the WHO case definitons or modifications of the CDC case definitions. The situation in Asia is complicated by the presence, among countries and within countries, of facilities and groups that can access highly sophisticated diagnostic tools with facilities and groups that can only use basic laboratory or health facilities. In addition, clinical manifestation of HIV opportunistic infections in Asia may be different from the clinical pictures found both in Western and in African countries. As described above, specific AIDS case definitions have been developed for use in Africa, Europe, Latin America and the U.S. By now, we have accumulated sufficient experience and expertise in this region to be able to look critically at the existing definitions and begin to develop AIDS case definition that may be more appropriate for the Asia-Pacific region.

Monitoring of the prevalence and incidence of other STDs

Considerable thought and effort must continue to go into the design of surveillance systems for STDs other than HIV. For too long, methods that were clearly developed for other infectious diseases have been uncritically used for STD surveillance, with the consequence that there is little in the way of reliable estimates of the incidence and prevalence of most STDs, and the extent of morbidity that they actually cause. In addition, innovative ways must be developed to include cases that are treated in venues that often do not report to public health authorities, such as private physicians and pharmacies.

Surveillance of HIV in children

Given the number of women affected by HIV in the Asia and Pacific region, the potential for vertical transmission to newborn children is significant. Although the number of children in the region infected by this route is unknown, the numbers are likely small at present compared to adult infections. However, HIV infections in newborns have an importance that far outweighs their numerical value because they are potentially preventable.

Programs that offer voluntary HIV counseling and testing to pregnant women and appropriate therapy to those infected are needed to prevent vertical transmission of HIV to newborns. In this regard, the surveillance of HIV in children is important since it can help to define the need for such programs, guide their implementation, and evaluate their effectiveness.

Surveillance of HIV infection in early childhood and adolescence in the region is also deficient. The first opportunity surveillance systems have to observe the emergence of HIV infection in this population is when girls become pregnant and attend antenatal clinics, or when boys are enrolled in the military. By the age of 15¾the earliest age included in routine surveillance in most countries¾some boys and girls may have already become infected with HIV, as found in Manipur state in India, Cambodia and Thailand. Surveillance systems should become more sensitive to age and sex differentials in young people.

4.2 Understanding HIV Spread: the Role of Behavioral Surveillance

Perhaps more so than any other continent, Asian countries are experiencing highly varied HIV epidemics, which are due to a number of factors including differential levels of risk behaviors such as multiple sexual partnerships and injecting drug use. Additionally, because Asia consists of both developing and developed countries with a variety of religious and cultural backgrounds, these risk behaviors are situated in highly varied socioeconomic and cultural environments that can either help or hinder the spread of HIV.

For example, the widespread epidemics in Thailand and Cambodia are based on the high patronage of commercial sex workers in those countries by males. Vietnam, on the other hand, has to date experienced a predominantly IDU-associated HIV epidemic, with concern growing that significant heterosexual spread will occur as well. Although HIV and AIDS case reports are still moderate for Papua New Guinea, that country is expected to face the prospect of a rapidly expanding HIV epidemic because of high rates of multiple sexual partnerships that are not commercially based. One hypothesis of differential rates of HIV spread in Asia appears in Box 7.

A growing number of Asian countries such as Thailand, the Philippines, Indonesia, India, and Cambodia have initiated behavioral surveillance surveys (BSS) to understand the role of and track behavioral risks for HIV. BSS has its roots in HIV and STD surveillance and uses these methodologic concepts for monitoring HIV risk behaviors. HIV prevalence taken from HIV surveillance is indicative of sexual behaviors from several years back and is slow to change even when risk behaviors may be changing rapidly. BSS thus bring surveillance data closer to what prevention programs require by providing systematic measurements of risk behaviors over time (see Box 8).

Given the emerging epidemics in the region, other countries should be seriously considering adding behavioral surveillance to their epidemic monitoring efforts. Specifically, while existing BSS projects have concentrated on sex workers and various occupational sub-groups of males that tend to visit sex workers, efforts are needed to conduct behavioral surveillance in areas where injecting drug users and men who have sex with men are located in sizable numbers.

Researchers have struggled with determining the important risk behaviors to track and how to best measure them. Consensus is beginning to form around six types of behaviors and characteristics that help to predict the course of the epidemic:

• Age at first intercourse
• Type of sex partners (e.g., casual, commercial)
• Number of sex partners by type
• Condom use
• Needle-sharing (for injection drug user surveys)
• Other factors affecting risk taking, such as alcohol and substance use

Such behaviors are sensitive and respondents may be reluctant to discuss them openly. Surveys thus require well-trained interviewers and appropriate interview settings. Behavioral surveillance, like other surveys reliant on self-reports, must be carefully implemented and validated through other data since self-reported behavioral data can be biased.

Behavioral surveillance is not a panacea for understanding all facets of HIV risk behaviors. In fact, it should be limited to only a few key target groups with a survey instrument containing rapport-building questions and a small but carefully chosen set of risk behavioral questions. In-depth information about target groups, the evaluation of specific interventions, and relationships between several behavioral variables are better obtained through quantitative and qualitative behavioral research specifically designed to answer these questions. They are necessary adjuncts to behavioral surveillance that together form a comprehensive package of monitoring and evaluation.

For example, sexual networking patterns play an important role in the rapidity with which HIV is spread within special populations (e.g., sex workers, clients of sex workers or highly mobile populations, etc.) and within broader segments of the society (e.g., women without sexual risk behavior). These patterns can be monitored to some extent through the application of behavioral surveillance, but also require other types of socio-behavioral investigations (e.g., qualitative/ ethnographic studies).

There is a need to develop minimum common approaches and methodologies for STD surveillance as well. Given the well-documented link between STD and HIV transmission, it is critical to assess how STD rates are affecting the course of AIDS epidemics in the region. With a few notable exceptions, there is a poor understanding of STD prevalence in countries of the region. Analyzed together, HIV, STD, and behavioral surveillance can provide complementary data to understand and track different facets of sexually transmitted diseases and their risk behaviors.

4.3 HIV Testing Policies and Programs

HIV testing may serve any of the following objectives:

1. Screening: To ensure safe blood transfusion and organ transplant through the screening of blood and blood products, and of tissues, sperm or ova from donors.
2. Surveillance: Unlinked and/or anonymous testing of serum (or other body fluids) for the purpose of monitoring HIV prevalence over time in a given population.
3. Diagnosis of HIV infection: Voluntary testing in order to inform individuals about their serostatus.

HIV testing policies and programs are relevant to the monitoring of HIV/AIDS epidemics regardless of the original objective for testing, and the test results unlinked from identifying information may be used additionally for surveillance. For example, test results from blood donors and those seeking diagnosis are often used in surveillance reports.

Testing strategies vary widely and the appropriate test or combination of tests depend on three criteria:

1. the objective of the test (for screening, surveillance, or diagnosis);
2. the sensitivity and specificity of the test(s) being used; and
3. the prevalence of HIV infection in the population being tested.

Most HIV tests detect HIV antibody in serum/plasma or other body fluids. Although HIV antibody testing has been available in the region since 1985, tests have recently become available to detect viral antigen. These are much more expensive than HIV antibody tests and are used by only a few Asian and Pacific countries, generally only in specialized medical settings. Tests that detect antibody in fluids such as saliva or urine, although available, are currently used only for research purposes. These collection methods and tests have great potential for use in surveillance testing as they will help reduce some logistical difficulties and costs. Tests commonly used in developing countries are based on the presence of HIV antibody in serum/ plasma, i.e., ELISA, Western Blot, rapid tests. In most countries in the region, testing strategies consist of an initial screening test, usually an ELISA-based test, followed (for reactive specimens only) with a second (supplemented) test, generally either a second ELISA and/or Western Blot test.

Key issues

Testing strategies vary tremendously across the region and within each broad objective (screening, surveillance, and diagnosis). Complex technical, programmatic and ethical/human rights issues abound. The issues presented below are crosscutting and are encountered in most countries in the region:

Financial and qualified human resources are insufficient in most countries to provide adequate coverage for quality blood screening, surveillance, and diagnostic testing. Policymakers must carefully evaluate and provide guidelines for HIV testing options that are appropriate to the country, the purpose of testing and are affordable. Resources must be allocated for surveillance, blood screening and diagnosis, while carefully balancing these needs within the larger public health needs. Innovative approaches must be considered for tapping additional resources, such as referring testing clients to the community for counseling services; also, there is a need to increase and improve training at all levels.

There is an urgent need to assure quality throughout the testing process to ensure accurate results and confidence in data and results. This can be achieved by developing standard procedures, increasing and improving training for field workers (data collectors) and lab technicians, and monitoring the entire process (see Box 9).

HIV testing has the potential for human rights abuses, including marginalization, coercion, discrimination and violence. Policymakers should develop and enforce clear policies on confidentiality and the use of test results and include persons living with HIV/AIDS in policy formulation. Legal provisions must be established to protect testing clients from abuse. Once these policies are developed, they must be disseminated and included in training for all people involved in the testing process from the point of first contact in health care settings, to the lab technicians to the counselors.

There is an increasing demand for early diagnosis of HIV infection, both for personal knowledge and decision making and for medical management. Testing should always be accompanied by counseling, support services and treatment when available. Before aggressive case finding for medical management can be recommended, policymakers must consider the availability of treatment options and balance resource allocations of the overall public health program. Clear policies regarding use of test results and confidentiality must be established and enforced.

A number of countries in the region are considering the use of HIV Rapid Tests for home-based testing. Policymakers should carefully consider the complexity of the administration and interpretation of the tests and the wide ranging implications and effects home testing can have, including the psychosocial support requirements and opportunities for human rights abuses. These must be weighed against the potential benefits and be guided by clear regulation of the use of HIV tests outside the laboratory setting.

4.4 Modeling HIV/AIDS/STDs in the Asia-Pacific Region: Approaches and Limitations

Models for the spread of HIV vary in sophistication from straightforward curve fits to extremely complex process models reproducing the dynamics of multiple modes of transmission, requiring extensive behavioral, epidemiological, and demographic inputs. Both types of models have been applied in the Asia-Pacific region by a number of different groups (see below). A strong demand has been seen for the output of these modeling efforts. This demand is driven by national AIDS programs and AIDS organizations, which wish to use them to advocate for expanded responses and to anticipate the magnitude of the tasks ahead of them, and by policymakers who want to understand the longer term implications of the epidemic for their own work. The value of models comes not only from providing people with concrete estimates of HIV infections and deaths for advocacy and planning purposes, but because they force people to closely examine and try to coherently interpret available epidemiological and behavioral data (see Box 10).

STD modeling approaches have rarely been applied in the Asia-Pacific region, except in the context of models such as iwgAIDS, which simultaneously tracks STDs and HIV. To a great extent, this is the result of the lack of quality STD trend data from any of the countries in the region. Thus, STD models will not be discussed further except as a key issue.

A Brief History of Past HIV/AIDS Efforts in the Asia-Pacific Region

While Epimodel, a simple curve fitting model, has been applied extensively throughout the region (in just about every country), for the most part modeling efforts using process models have been concentrated in Thailand, with some additional work done in Indonesia and Papua New Guinea. All of this work currently exists as "gray literature," and not in published journals.

The earliest and most extensive modeling efforts in the Asian region have looked at Thailand, perhaps encouraged by the ready availability of both epidemiological and behavioral data since 1990. Thailand also offers the best opportunity to check the validity of models, since eight years of epidemiological trend data along with numerous behavioral studies are available. A number of groups have examined the Thai epidemic (see Box 11).

The Thai experience is one of relatively abundant data, and an epidemiologically interesting and relatively complex (and in some respects an atypical) epidemic. Consequently, considerable modeling attention has been paid to the Thai case over the past decade (see Box 11).

These efforts, briefly described in the box, lead to a number of observations:

• Strong collaboration can help to ensure maximum use of available empirical data, while enhancing the chances for the results of modeling to be adopted by consensus and used in policy decisions and program planning.
• But even in the same country using the same sources, considerable variation exists in the translation of epidemiological data into estimated current or cumulative prevalence figures. Further study is needed of how best to convert sentinel data sets to estimates of general population prevalence.
• Trends in behavioral data are not easy to incorporate into existing models. The Thai situation in the 1990s has been a very dynamic one, which requires the ability to input behavioral parameters such as condom use and risk behavior over time if the epidemiological trends are to be reproduced.
• Despite the complexity of some of the models, they have had limited success in fitting actual trends in the data, raising concerns about whether assumptions in the models correctly describe the complexity of the Thai situation.
• Perhaps due to strong collaboration, examined even in hindsight, none of the above efforts produced results that were truly outlandish, although a number of such projections were to be found in the popular press. The higher results of the earliest efforts in part show the fact that much of the Thai behavior change had not yet been reflected in the data available.

Thailand has benefited substantially from these modeling efforts. Programs and policies have been justified by the results of these models, and the Thai AIDS prevention effort has become stronger as a consequence. But the Thai experience also highlights the complexities of modeling AIDS epidemics and the potential pitfalls along the way. Other countries, perhaps with less empirical data and fewer resources devoted to modeling, have had differential success both in developing plausible models of the local epidemic as well as in incorporating these results into their prevention efforts (see Box 12).

There is a need to improve policymakers' understanding of uses and limitations of HIV/AIDS models. However, before that can be done, modelers themselves need to reach a consensus on the uses and limitations of the models that are in use in the Asia-Pacific region.

One approach to capacity building for modeling is to establish national technical working groups to assemble the relevant data and supervise the HIV estimation and projection process. These working groups can be linked through the Internet and convened through sub-regional meetings. Current modeling approaches perform best when an epidemic is in progress; tools for estimating levels and trends of HIV in the absence of an epidemic are not yet well-developed. Alternative models need to be developed that include a few key variables and can describe the potential for an epidemic where HIV is still low.

There is still a need in Asia for more capacity building in HIV and behavioral surveillance in order to collect the data that would be required by these models. Multi-country studies, with standardized methodology for collecting HIV/STD and behavioral data, should be applied along the lines of the current UNAIDS project in several countries in Africa. Application of complex process models to produce projections should be limited to countries in which at least three or four years of reliable epidemiological and behavioral trend data are available.

Models that estimate and project levels of STDs have not been widely used in the Asia-Pacific region. There is a need for translating data on STD incidence and prevalence (by type of STD) in the region into measures of HIV epidemic potential. Countries need to be encouraged to conduct estimates and very short-term projections of HIV, revise these estimates on an annual basis, use the lower of multiple scenarios for planning and public dissemination, use these data to project the impact of HIV and of the interventions undertaken to bring the HIV epidemics under control.