Questions and Answers: Advances in Methods of Measuring Incidence
September 3, 2008
- Why does there appear to be an annual increase in new HIV infections?
It is important to note that the new estimates do not show an increase in new HIV infections. The new estimates are from the first national surveillance system of its kind in the world that is based on direct measurement of new HIV infections using technology that can distinguish recent from long-standing, population-based infections. Using the new surveillance system, CDC estimates that 56,300 new HIV infections occurred in the United States in 2006. CDC previously estimated that approximately 40,000 new HIV infections occurred annually since the 1990s. The 2006 incidence estimate is about 40% higher than the previous estimates. The new system reveals that the epidemic is -- and has been -- worse than previously estimated and underscores the need to expand HIV prevention to reach those at greatest risk.
- What is STARHS?
AdvertisementThe CDC-developed, breakthrough technology known as STARHS stands for Serological Testing Algorithm for Recent HIV Seroconversion. This technology allows us to know, for the first time, which HIV infections are new. CDC applied this technology in the first national surveillance system of its kind in the world to directly measure new HIV infections. CDC scientists worked for almost a decade to realize the promise of STARHS technology by combining it with an expanded HIV reporting system in the U.S. This combined approach will allow CDC to better monitor the current course of the epidemic, target prevention efforts where they are needed most, and assess the impact of prevention efforts.
- Why is the new national HIV incidence estimate more precise and more reliable than prior estimates of HIV incidence?
CDC's new HIV surveillance system is based on an approach known as STARHS (Serological Testing Algorithm for Recent HIV Seroconversion), which uses innovative testing technology to determine at the population level which positive HIV tests represent new HIV infections (those that occurred within approximately the past five months). Before the widespread availability of this technology, HIV diagnosis data provided the best indication of recent trends in key populations. However, diagnosis data only indicate when a person is diagnosed with HIV, not when an individual was actually infected, which can occur many years before a diagnosis.
By applying this technology to new HIV diagnoses in 22 states with mature name-based HIV reporting systems, CDC was -- for the first time -- able to identify those diagnoses in a given year that represented new infections. Using a complex statistical model, these data were extrapolated to the general population to provide the first national estimates of HIV incidence based on direct measurement.
CDC researchers also used a separate method called extended back calculation to confirm the official 2006 incidence estimates and to examine historical trends in HIV infections in the U.S. from 1977 to 2006. The method uses a statistical model that considers all HIV and AIDS cases diagnosed through 2006 and reported to CDC, as well as HIV testing patterns. This particular extended back calculation approach has become possible in the United States because of an expanded name-based HIV reporting system, which provides a population-based system for identifying new diagnoses. However, the method is an indirect measure of incidence and is most reliable for earlier years; data for the most recent years (2003-2006) must be interpreted with caution. Additionally, extended back-calculation does not generate single-year estimates, instead providing averages over multiple-year periods.
The statistical methods used to develop the 2006 incidence estimates, as well as the extended back calculation of historical trends, were developed in consultation with outside experts, and both the methods and their application underwent rigorous external scientific reviews.
Moving forward, the STARHS-based surveillance system will provide the most reliable way to monitor incidence trends. Over time, the picture will become even more clear as analyses for specific populations are completed (e.g., black women, young men who have sex with men). Trend information from this system will allow for improved targeting and evaluation of prevention efforts for the populations at greatest risk and improved monitoring of the impact of HIV prevention efforts.
- How was incidence extrapolated from 22 states to the entire United States and Washington DC?
- Incidence in 22 states was estimated in almost 70 strata (or subgroups) based on sex, race/ethnicity, age, and transmission category. Using the ratio of the number of new infections to the number of AIDS cases within these strata based on the 22 states, incidence was extrapolated using the number of AIDS cases within these strata for the remaining states and the District of Columbia (all grouped together). This method assumes that the strata-specific ratio of HIV infection to AIDS incidence in the 22 states (all grouped together) is similar to the ratio in other states and DC (all grouped together), an assumption generally supported by historical trends in HIV and AIDS diagnoses in the United States
- How can states calculate their own area-specific incidence estimates?
- For states that are not conducting HIV incidence surveillance, local data on HIV diagnoses are best to describe the status of the local epidemic. These states do not have the data that are needed to compute their own incidence estimates.
- Areas that are part of the national HIV incidence estimate can calculate their own estimates. Some areas that participated in HIV incidence surveillance and are not part of the national estimate can also calculate their own incidence estimate. CDC provided training on how to do this and will continue to provide technical assistance.
- To calculate a local incidence estimate, an area must have participated in HIV incidence surveillance and have a sufficient number of BED results and HIV testing histories for their cases. (The numbers required are specified in the CDC technical guidance).
- How does CDC know that the higher incidence estimate is not due to an actual rise in new HIV infections?
It is important to note that the new estimate for national HIV infections does not represent an actual increase in the annual number of new HIV infections. A separate CDC historical trend analysis suggests that the number of new infections was never as low as 40,000 and has been roughly stable since the late 1990s. Although CDC's analysis suggests that overall the epidemic has been stable in recent years, steady increases in new HIV infections are estimated to have occurred among men who have sex with men (MSM) since the early 1990s, and the overall number of new HIV infections remains unacceptably high. These findings underscore the ongoing challenges in confronting this disease and the urgent need to expand access to effective HIV prevention programs.
- Do the new HIV incidence estimates indicate that the number of new HIV infections among women is increasing?
No. CDC's historical analysis indicates that the number of infections among women has remained relatively stable since the early 1990s.
- Where can I find the data for the new HIV incidence estimates
The new incidence estimates were released by The Journal of the American Medical Association (JAMA) on August 3, 2008. The new estimates are also available on this Web site.
- How do the 2006 HIV incidence estimates affect future HIV prevention efforts?
The new technology provides a critical missing piece in tracking the U.S. HIV epidemic and, over time, will allow us to better direct and measure progress in prevention efforts among populations at highest risk for HIV. The current HIV incidence estimates, as well as future annual estimates, will become one of CDC's most sensitive measures for monitoring progress toward reducing HIV transmission in the U.S. Although the 2006 national incidence estimate may appear different than suggested by previous HIV incidence studies, the overall data are consistent with CDC's HIV prevention priorities. CDC will strive to expand the reach of prevention services for at-risk populations and remain committed to its core prevention priorities: reducing new infections, increasing knowledge of HIV status, linking persons infected with HIV to care and prevention services, and strengthening the ability to monitor the epidemic.
- Will the 2006 HIV incidence estimates cause CDC's priority populations to shift?
Not at this time. Gay and bisexual men of all races and African Americans continue to be hardest hit by the epidemic; and about 25% of people living with HIV remain unaware of their infections. The new estimates also show that more HIV infections occur among young people under age 30 than any other age group. More detailed information about trends in HIV incidence among youth and other at-risk populations will become available over time. As these data become available, CDC priorities may change over time in accordance with these trends.
- Which populations experienced the greatest burden of HIV infections in 2006?
The 2006 HIV incidence estimates show:
- Gay and bisexual men of all races remain the group most heavily affected by HIV, accounting for 53% of all new infections.
- The impact of HIV is greater among blacks than any other racial or ethnic group, with an HIV incidence rate that is 7 times higher than whites (83.7/100,000 for blacks compared to 11.5/100,000 for whites) and almost 3 times higher than Latinos (29.3/100,000).
- What does the new HIV incidence estimate imply about HIV prevention efforts?
We have significant evidence that prevention works. Infection rates for people who inject drugs have declined dramatically over time. Infection rates in women of all races have been relatively stable since the early 1990s. Infection rates in heterosexuals who engage in high-risk sexual
behaviors have been relatively stable since the early 1990s. While stability and declining infection rates in certain populations is an important sign of progress, the new estimates underscore the need to expand access to HIV prevention. Many populations at risk for HIV are not being reached by our collective prevention efforts, and much more must be done. For example, recent data show that over the past year, 80% of MSM are not being reached by the intensive interventions we know to be most effective. This underscores the challenge of reaching new generations while adapting to the evolving epidemic. Additionally, far too many undiagnosed individuals remain unaware of their infections. We must ensure that resources and prevention efforts are commensurate with the scale of the epidemic.
- Will CDC be looking at new interventions and communication strategies to prevent HIV among populations hardest hit by the epidemic?
Yes, CDC is currently working to expand the use of new scientifically proven interventions for a range of populations at risk and testing a number of innovative behavioral interventions for key risk groups, particularly African American men who have sex with men (MSM). CDC remains committed to expanding HIV prevention and care services to all at-risk populations and those hardest hit by the epidemic, such as African Americans and MSM. It is clear that there is an urgent need to accelerate progress, and as a nation we must continue to invest in prevention and extend the reach of HIV prevention services.
- Why is the new method viewed as a breakthrough in HIV incidence monitoring?
This new method is viewed as a breakthrough in HIV incidence monitoring because it can distinguish recent from long-standing HIV infections. The new method provides the clearest picture of the U.S. epidemic to date, and over time, will enable CDC to better target and evaluate prevention efforts for the populations at greatest risk. Accurately tracking the HIV epidemic is essential to CDC's HIV prevention efforts.
- How will STARHS influence CDC's HIV/AIDS surveillance and prevention activities?
The 2006 HIV incidence estimates provide the clearest picture of the U.S. epidemic to date, and over time will enable CDC to better target and evaluate prevention efforts for the populations at greatest risk. By applying the Serological Testing Algorithm for Recent Seroconversion (STARHS) to new HIV diagnoses in 22 states with name-based HIV reporting systems, CDC is able to identify new diagnoses that represent new infections. These data are extrapolated to the general population to provide the first national estimate of HIV incidence based on direct measurement. The STARHS-based surveillance system will provide the most reliable way to monitor incidence trends and result in more accurate and timely HIV incidence estimates to better direct prevention programs, mobilize communities and inform resources decisions.
- Will the 2006 estimate of annual new HIV infections cause a corresponding increase in the often-cited figure of more than 1 million people living with HIV/AIDS in the United States? What about the 25% estimate of those who don't know they are infected?
No. The 2006 HIV incidence estimate does not affect the previous HIV prevalence estimate of more than 1 million people living with HIV/AIDS. New estimates of national HIV prevalence and the proportion of persons who know their HIV status in 2006 will be generated later in 2008.