Changes in HIV incidence are the “gold standard” when it comes to measuring the effectiveness of HIV intervention programs, and reducing HIV incidence is the ultimate objective of HIV intervention. Therefore, reliable estimates of HIV incidence are critical for monitoring and evaluation. In the early 1990s, back-calculation based on the number of reported AIDS cases and the incubation period distribution from HIV infection to AIDS was used to estimate HIV incidence in the United States.1,2 This method was no longer valid after highly active antiretroviral treatment became available in the mid-1990s because treatment altered progression to AIDS, and it was no longer possible to accurately estimate the natural incubation period distribution of HIV infection.
The Serologic Testing Algorithm for Recent Seroconversion is a general term for a number of laboratory techniques that distinguish recent from long-standing HIV infections.3–7 The BED HIV-1 Capture Enzyme Immunoassay is one of them and it classifies an infection as recent based on the ratio of HIV-1 IgG to total IgG. The Centers for Disease Control and Prevention (CDC) applies this technology to persons who are newly diagnosed with HIV in the United States to identity new infections and then uses a statistical model to generate the national estimate of HIV incidence.8
CDC estimated that HIV incidence in the United States had been stable at approximately 50,000 new infections per year from the early 2000 to 2010.9,10 CDC also reported that the number of HIV diagnoses declined steadily from 56,715 in 2002 to 41,720 in 2011, whereas HIV case finding had been improving.11–14 A declining trend in the number of new diagnoses accompanied by improving HIV case finding is an indication of a declining trend in HIV incidence or fewer new infections than new diagnoses, or both, but CDC reported a stable trend in HIV incidence and more new infections than new diagnoses in a number of years. This contradiction suggests that the CDC estimates of HIV incidence may not be accurate.
To develop a different approach to estimate HIV incidence, we examined HIV case reporting data. Previously, the number of newly diagnosed HIV cases in a year was used as a proxy measure for HIV incidence but has never been used as the sole data source to calculate HIV incidence because HIV infection can remain asymptomatic for many years and newly reported cases represent both recent and long-standing infections.15,16 We present a novel use of HIV case reporting data to estimate HIV incidence.
HIV case reporting data on the number of new diagnoses in the United States overall and by sex were directly obtained from a CDC report.11 The analysis period (2002–2011) was broken down into 3-year periods with overlaps (eg, 2002–2004, 2003–2005 … 2008–2011). As CDC did, we assumed that all HIV infections would eventually be diagnosed either through testing or through death.10 We also assumed that, within each 3-year period, HIV incidence and case finding were stable. The relationship between the number of new diagnoses and HIV incidence is presented in the equations shown below. Using the CDC definition, we defined HIV incidence as the number of new HIV infections in a year. By solving the equations, we obtained HIV incidence for year 2. Detailed steps for solving the equations are presented in the Supplemental Digital Content, https://links.lww.com/QAI/A927.R: case finding rate, which is the probability of an HIV-infected individual being diagnosed in a year;Di: number of new diagnoses in year i, where i = 1, 2, or 3;Ui: number of undiagnosed persons at the beginning of year i, where i = 1, 2, or 3;Ii: HIV incidence in year i, where i = 1, 2, or 3.
The assumption that HIV case finding is stable is sometimes violated. In September 2006, CDC revised its HIV testing guidelines to recommend routine HIV screening in healthcare settings for all patients aged 13–64 years.17 The release of the new guidelines led to an improvement in HIV case finding12,14,18 and made the assumption of stable HIV case finding invalid. However, once the testing expansion was fully implemented, HIV case finding would become relatively stable again, at a higher level. A new testing initiative usually takes several years to be fully implemented.19 We treated the 3 years (2006–2008) following the release of the revised guidelines as the expansion period with improving HIV case finding and excluded these 3 years from our HIV incidence calculation because of the violation of the assumption of stable HIV case finding.
The method requires the number of new diagnoses in the previous year, current year, and following year to calculate HIV incidence in the current year. With 10 years of data (2002–2011), we should be able to calculate HIV incidence for 8 years (2003–2010), but because the assumption of stable HIV case finding was violated during 2006–2008, we were only able to report HIV incidence for 5 years (2003–2005 and 2009–2010).
We first calculated HIV incidence for the total, males, and females based on the number of new diagnoses and found that there was a slight mismatch between the sum of males and females and the total, with a difference between −1.08% and 2.21%. We required that the sum of estimated HIV incidence by sex equaled the estimate of the total, so we proportionally rescaled the estimated HIV incidence for males and females.20 The uncertainty intervals were estimated based on the estimate that the number of new diagnoses may be within ±10% of the published figure in the HIV surveillance reports due to duplicates and reporting delays.11,21,22 All analyses were conducted using Microsoft Excel.
The number of new HIV diagnoses in the United States decreased about 26% overall from 56,715 in 2002 to 41,720 in 2011, 20% among males from 41,010 to 32,980, and 44% among females from 15,706 to 8740 (Table 1).
Figure 1 shows HIV incidence estimates from CDC and our new method. For the revised estimates, dashed lines between 2005 and 2009 indicate that the estimates for 2006–2008 are excluded because of unstable HIV case finding during that period. Compared with the CDC estimates, the revised estimates were lower. Although CDC reported stable HIV incidence, the revised estimates showed a decline in HIV incidence from 52,721 (range: 47,449–57,993) in 2003 to 39,651 (range: 35,686–43,617) in 2010, and among males from 38,164 (range: 35,051–42,840) to 33,035 (range: 29,088–35,553), and among females from 13,557 (range: 12,133–14,830) to 6616 (range: 5825–7120).
To estimate HIV incidence in the United States, we used a simple mathematical model with HIV case reporting data as the sole data source and fewer assumptions. There have been previous studies reporting declines in HIV incidence at the local level and in some populations.23–26 We reported a steady decline since 2003 in HIV incidence in the United States at the national level.
A recently published article using CDC's estimates of HIV prevalence and all-cause mortality also reported that CDC may have overestimated HIV incidence and there was an overall slow, monotonically downward trend in HIV incidence.27 For 2009 and 2010, 2 years that all 3 methods made estimates, the 2 new methods produced similar results with 38,066 and 37,270, respectively, for 2009 and 37,253 and 39,651, respectively, for 2010, both lower than the CDC estimates, 45,000 for 2009 and 47,500 for 2010. The estimate (25,561 for 2013) from the Global Burden of Disease Study was also lower than the CDC estimate.20 The findings from these 3 studies along with the contradiction in the CDC estimates between a declining trend in new diagnoses and a stable trend in new infections suggest that CDC may have overestimated HIV incidence.
Compared with the method used in the National HIV Incidence Surveillance,10,28 our method is simple, with HIV case reporting data as the sole data source, and makes fewer assumptions. Because it does not require biomarker test results to distinguish recent from long-standing infections, the method was able to use the data reported before the implementation of the National HIV Incidence Surveillance in the United States in 2005. With a new method and more years of data (2002–2011), we were able to detect a declining trend in HIV incidence in the United States.
Our new method has limitations. First, it depends on the completeness and accuracy of HIV case reporting, without which this method could not be used to estimate HIV incidence. This limitation applies to both the back-calculation method and the National HIV Incidence Surveillance.1,28 Given the maturity of HIV case reporting in the United States,29,30 this limitation is unlikely to affect our estimates.
Second, we assumed stable HIV incidence within each of the 3-year periods when calculating HIV incidence but later reported an overall declining trend in HIV incidence during the analysis period. The seemingly inconsistency between our assumption and conclusion can be explained by that we assumed relatively, not absolutely, stable HIV incidence within each of the 3-year periods. Based on I2002 ≈ I2003 ≈ I2004 and I2003 ≈ I2004 ≈ I2005, we cannot conclude I2002 ≈ I2005 because each 3-year analysis period was independent and HIV incidence was only assumed to be relatively stable within each 3-year period. An easy solution to avoid this inconsistency can be done by removing overlapping years, but it would give us fewer data points. For example, with the 10-year data (2002–2011), we could calculate HIV incidence in 3 years, HIV incidence in 2003 based on the number of new diagnoses in 2002–2004, HIV incidence in 2006 based on the number of new diagnoses in 2005–2007, and HIV incidence in 2009 based on the number of new diagnoses in 2008–2010. The current method provided us identical estimates for 2003, 2006, and 2009, plus estimates for 5 more years (2004, 2005, 2007, 2008, and 2010). The additional data points using the same method provide us more information on the trend in HIV incidence and give us more flexibility dealing with the data in years when the assumption of stable HIV case finding was violated. Therefore, we do not think the seemingly inconsistency would affect our conclusions.
Third, the assumption of stable HIV case finding is sometimes violated. However, the assumption does not have to hold during the entire analysis period. In our calculation, HIV case finding was assumed to be stable within each 3-year period during 2002–2006 and 2008–2011 and unstable (improving) during 2006–2008. The method was able to estimate HIV incidence for the period of 2003–2010, with HIV incidence in 2006, 2007, and 2008 excluded. We excluded these 3 years without formal testing of the assumption but based on the knowledge that CDC's revised HIV testing guidelines were released in September 2006. It could be expected to take several years for the recommendations to be implemented,17,19 during which time case finding would have been unstable. Further research is needed to develop a method for researchers to test the validity of the assumption before deciding which years of data should be excluded.14,31,32
Fourth, we assumed stable HIV case finding within each 3-year period during 2002–2006 and 2008–2011 and used the exactly equal HIV case finding rates in the equations to estimate HIV incidence. HIV case finding was very likely to be slightly improving within each 3-year period during 2002–2006 and 2008–2011.14 Using the exactly equal HIV case finding rates enables us to provide the simplest equations to estimate HIV incidence with reasonably accuracy. In addition, estimating an accurate trend in HIV incidence may be even more important than an accurate number. We have already found a declining trend in HIV incidence in the United States by assuming a stable HIV case finding. If indeed HIV case finding was improving, the decrease in HIV incidence would be more significant than we reported.
The findings from our analysis suggested that CDC may have overestimated HIV incidence and HIV incidence in the United States may have been declining since 2003.20,27 Our method is very simple using existing HIV case reporting data as the sole data source. The method may need continued evaluation using data from other countries. The gold standard in measuring HIV incidence is prospective cohort studies. When it is impractical, if not impossible, to conduct such a study among a nationally representative sample in the United States, we should consider all available methods, rather than relying on one, to provide more accurate estimates of HIV incidence to guide our intervention programs.
Three major factors may have contributed to the declining trend in HIV incidence in the United States: (1) expanded HIV testing led to fewer undiagnosed HIV infections,33,34 (2) expanded antiretroviral treatment reduced the number of people living with HIV with an unsuppressed viral load,35–38 and (3) increased availability of HIV behavioral interventions reduced risk behaviors among both HIV-positive and HIV-negative individuals.39
It is promising to see a declining trend in HIV incidence, but there are still about 40,000 persons newly infected with HIV every year in the United States. We should continue our effective interventions, including HIV case finding, HIV treatment, and behavioral interventions, and identify new interventions, such as pre-exposure prophylaxis, to prevent new infections.40–42
The authors would like to thank Kent Sepkowitz, Demetre Daskalakis, Jay Varma, and James Hadler for their review and comments on this article.
1. Biggar RJ, Rosenberg PS. HIV infection/AIDS in the United States
during the 1990s. Clin Infect Dis. 1993;17(suppl 1):S219–S223.
2. Bacchetti P, Segal MR, Jewell NP. Backcalculation of HIV infection rates. Stat Sci. 1993;8:82–101.
3. Janssen RS, Satten GA, Stramer SL, et al. New testing strategy to detect early HIV-1 infection for use in incidence estimates and for clinical and prevention purposes. JAMA. 1998;280:42–48.
4. Parekh BS, Kennedy MS, Dobbs T, et al. Quantitative detection of increasing HIV type 1 antibodies after seroconversion: a simple assay for detecting recent HIV infection and estimating incidence. AIDS Res Hum Retroviruses. 2002;18:295–307.
5. Murphy G, Parry JV. Assays for the detection of recent infections with human immunodeficiency virus type 1. Euro Surveill. 2008;13. pii: 18966.
6. Mastro TD, Kim AA, Hallett T, et al. Estimating HIV incidence in populations using tests for recent infection: issues, challenges and the way forward. J HIV AIDS Surveill Epidemiol. 2010;2:1–14.
7. UNAIDS/WHO Working Group on Global HIVA/IDS and STI Surveillance. When and How to Use Assays for Recent Infection to Estimate HIV Incidence at a Population Level. Geneva, Switzerland: UNAIDS/WHO; 2011.
8. Lee LM, McKenna MT. Monitoring the incidence of HIV infection in the United States
. Public Health Rep. 2007;122(suppl 1):72–79.
9. Hall HI, Song R, Rhodes P, et al. Estimation of HIV incidence in the United States
. JAMA. 2008;300:520–529.
10. Centers for Disease Control and Prevention. Estimated HIV incidence in the United States
, 2007–2010. HIV Surveill Supplemental Rep. 2012;17:1–26.
11. Johnson AS, Hall HI, Hu X, et al. Trends in diagnoses of HIV infection in the United States
, 2002–2011. JAMA. 2014;312:432–434.
12. Woodring JV, Kruszon-Moran D, Oster AM, et al. Did CDC's 2006 revised HIV testing recommendations make a difference? Evaluation of HIV testing in the US household population, 2003-2010. J Acquir Immune Defic Syndr. 2014;67:331–340.
13. Ransome Y, Terzian A, Addison D, et al. Expanded HIV testing coverage is associated with decreases in late HIV diagnoses. AIDS. 2015;29:1369–1378.
14. An Q, Kang J, Song R, et al. Bayesian hierarchical model with novel prior specifications for estimating HIV testing rates. Stat Med. 2016;35:1471–1487.
15. Kuo JM, Taylor JM, Detels R. Estimating the AIDS incubation period from a prevalent cohort. Am J Epidemiol. 1991;133:1050–1057.
16. Fauci AS, Pantaleo G, Stanley S, et al. Immunopathogenic mechanisms of HIV infection. Ann Intern Med. 1996;124:654–663.
17. Branson BM, Handsfield HH, Lampe MA, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep. 2006;55:1–17.
18. Cooley LA, Wejnert C, Rose CE, et al. Increases in recent HIV testing among men who have sex with men coincide with the Centers for Disease Control and Prevention's expanded testing initiative. Clin Infect Dis. 2015;60:483–485.
19. Myers JE, Braunstein SL, Shepard CW, et al. Assessing the impact of a community-wide HIV testing scale-up initiative in a major urban epidemic. J Acquir Immune Defic Syndr. 2012;61:23–31.
20. Murray CJ, Ortblad KF, Guinovart C, et al. Global, regional, and national incidence and mortality for HIV, tuberculosis, and malaria during 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 2014;384:1005–1070.
21. Centers for Disease Control and Prevention. Diagnoses of HIV infection in the United States
and dependent areas, 2012. HIV Surveill Rep. 2014;24:1–83.
22. Centers for Disease Control and Prevention. Diagnoses of HIV infection in the United States
and dependent areas, 2013. HIV Surveill Rep. 2015;25:1–82.
23. Xia Q, Nonoyama A, Molitor F, et al. Recent decline in the incidence of human immunodeficiency virus infection among California men who have sex with men. Am J Epidemiol. 2011;174:203–210.
24. Des Jarlais DC, Perlis T, Arasteh K, et al. HIV incidence among injection drug users in New York City, 1990 to 2002: use of serologic test algorithm to assess expansion of HIV prevention services. Am J Public Health. 2005;95:1439–1444.
25. Scheer S, Nakelsky S, Bingham T, et al. Estimated HIV incidence in California, 2006–2009. PLoS One. 2013;8:e55002.
26. Neaigus A, Jenness SM, Hagan H, et al. Estimating HIV incidence and the correlates of recent infection in venue-sampled men who have sex with men in New York City. AIDS Behav. 2012;16:516–524.
27. Bonacci RA, Holtgrave DR. Simplified estimates of HIV incidence and transmission rates for the USA, 2008-2012. AIDS. 2016;30:332–333.
28. Karon JM, Song R, Brookmeyer R, et al. Estimating HIV incidence in the United States
from HIV/AIDS surveillance data and biomarker HIV test results. Stat Med. 2008;27:4617–4633.
29. Glynn MK, Lee LM, McKenna MT. The status of national HIV case surveillance, United States
2006. Public Health Rep. 2007;122(suppl 1):63–71.
30. Glynn MK, Ling Q, Phelps R, et al. Accurate monitoring of the HIV epidemic in the United States
: case duplication in the national HIV/AIDS surveillance system. J Acquir Immune Defic Syndr. 2008;47:391–396.
31. Xia Q, Torian LV, Shepard CW. Limitations of indicators of HIV case finding. Epidemiology. 2015;26:e6–e8.
32. Xia Q, Kobrak P, Wiewel EW, et al. The high proportion of late HIV diagnoses in the United States
is likely to stay: findings from a mathematical model. AIDS Care. 2015;27:206–212.
33. Hall HI, Holtgrave DR, Maulsby C. HIV transmission rates from persons living with HIV who are aware and unaware of their infection. AIDS. 2012;26:893–896.
34. Hall HI, An Q, Tang T, et al. Prevalence of diagnosed and undiagnosed HIV infection–United States
, 2008–2012. MMWR Morb Mortal Wkly Rep. 2015;64:657–662.
35. Schwarcz S, Hsu LC, Scheer S. Disparities and trends in viral suppression during a transition to a “test and treat” approach to the HIV epidemic, San Francisco, 2008–2012. J Acquir Immune Defic Syndr. 2015;70:529–537.
36. Torian LV, Xia Q, Wiewel EW. Retention in care and viral suppression among persons living with HIV/AIDS in New York City, 2006–2010. Am J Public Health. 2014;104:e24–e9.
37. Xia Q, Braunstein SL, Wiewel EW, et al. Persons living with HIV in the United States
: fewer than we thought. J Acquir Immune Defic Syndr. 2016;72:552–557.
38. Xia Q, Kersanske LS, Wiewel EW, et al. Proportions of patients with HIV retained in care and virally suppressed in New York City and the United States
: higher than we thought. J Acquir Immune Defic Syndr. 2015;68:351–358.
39. Shea MA, Callis BP, Cassidy-Stewart H, et al. Diffusion of effective HIV prevention interventions–lessons from Maryland and Massachusetts. AIDS Educ Prev. 2006;18(4 suppl A):96–107.
40. Xia Q, Ning Z, Torian LV. A run-in period is needed in randomized controlled trials of directly observed antiretroviral therapy for HIV infection. J Acquir Immune Defic Syndr. 2015;68:e20–e23.
41. Grant RM, Lama JR, Anderson PL, et al. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med. 2010;363:2587–2599.
42. Kessler J, Myers JE, Nucifora KA, et al. Evaluating the impact of prioritization of antiretroviral pre-exposure prophylaxis in New York. AIDS. 2014;28:2683–2691.