An estimated 5%-21% of persons who are HIV infected have not been recently tested for HIV and therefore are not aware of their status.1,2 Identifying these people is central to HIV prevention efforts in the United States because those unaware of their status are believed to be over 3 times more likely to transmit HIV sexually than those aware of their status.3 A focus of HIV prevention has been to develop and support programs to diagnose those not aware of their status,4 including recruitment of at risk persons from social networks of already known positives,5 partner counseling and referral services,6 and increased HIV testing.7 Such initiatives aim to identify persons not previously known to be HIV infected rather than those already aware.
Recently, improvements in rapid HIV test technology8 and the implementation of recent Centers for Disease Control and Prevention guidelines supporting routine testing in medical settings9 have helped to facilitate increased HIV testing. New York City has increased the number of tests conducted in recent years by increasing investment by the city's Department of Health and Mental Hygiene (DOHMH). In 2007 alone, the number of HIV tests conducted or funded by DOHMH increased by over one-third from the previous year, to over 169,000 tests performed. Increased HIV testing may increase reports of positive HIV cases, but whether such reports are for truly newly diagnosed infections may be difficult to assess. Patient self-report may be useful in trying to determine this but cannot be assumed accurate, especially in certain populations.10
Surveillance registries can be used to evaluate HIV testing efforts because they contain historical clinical data on previously reported HIV infections in a given area. Using surveillance data, we quantified the contribution of truly new diagnoses to all positive HIV-1 Western blot tests in New York City to evaluate HIV testing efforts between 2004 and 2006. We determined the contribution of certain types of testing sites, focusing on those participating in a city initiative to increase HIV testing. Finally, we characterized repeat positive testers vs. new HIV diagnoses to better understand who undergoes repeat testing.
The New York City HIV Surveillance Registry contains demographic and clinical information confidentially reported to the DOHMH as required by New York State law on all persons diagnosed with HIV infection and/or AIDS in New York City.11,12 Since June 2000, all positive confidential HIV-1 Western blot results have been reportable. Additional HIV-related events reportable by laboratories via electronic submission in New York State include all HIV detectable (since June 2000) and undetectable (since June 2005) viral load test results and all CD4+ lymphocyte test results below 500 cells per microliter (since June 2000) and above 500 cells per microliter (since June 2005). Events reportable by providers via the New York State Provider Report Form include all diagnoses of HIV infection, HIV illness in a previously unreported individual, and AIDS-defining conditions.
Reportable events both identify new HIV diagnoses and update records of previously known cases. A deterministic algorithm based on first name, last name, date of birth, and Social Security number is used to generate potential matches with previously known cases. Exact matches on first name, last name, and date of birth are accepted as “true” matches. Other potential matches are verified by 2 independent reviewers, with discordant decisions resolved by a third independent reviewer. Any events matching with already known cases are linked to the Registry case record, whereas tests that do not match are confirmed through a field investigation by surveillance staff via medical record review at testing sites. The HIV surveillance case definition must be met for adult cases to be included in the Registry; a positive Western blot test for HIV or a diagnosis of HIV infection that is documented in a medical record by the physician.13
Population and Exclusion Criteria
For this analysis, we identified any positive confidential HIV-1 Western blot in the Registry performed in New York City between January 1, 2004 and December 31, 2006 with a result reported to the DOHMH by September 30, 2007. We excluded duplicate Western blot records and any Western blot in a newly reported case that could not be confirmed through medical record review. Preliminary positive results from rapid HIV antibody testing, positive anonymous Western blot test results, and Western blot test results from clinical trials or for insurance purposes only were not included because such results are not reportable in New York State.
Outcome of Interest
The outcome of interest was the categorization of each Western blot as either a diagnostic positive test or a repeat positive test. A test was considered a diagnostic positive test if the earliest evidence of HIV infection for the corresponding case in the Registry was either that Western blot or another reportable HIV-related event (ie, a physician-documented diagnosis of HIV infection, an HIV-1 viral load result, or a CD4+ lymphocyte test) occurring in the same month as the Western blot, or in the month immediately prior, to allow for some uncertainty in the actual date of first evidence of infection. In contrast, a test was considered a repeat positive test if there was any documentation in the Registry of an HIV-related event occurring more than 1 month before the Western blot, even if this event occurred outside New York City.
Demographic and Clinical Variables
Demographic and clinical information for each case was recorded based on medical record review when it was first investigated by field surveillance staff. Variables included sex, age at initial HIV diagnosis, race/ethnicity, country of birth, and clinical status (HIV non-AIDS vs. AIDS). HIV transmission category was classified as injection drug use history, men who have sex with men, high-risk heterosexual risk, perinatal transmission, or other/unknown risk, based on previously described methods.14
Each testing site reporting at least 1 Western blot test between 2004 and 2006 was categorized into 3 main categories: health department (DOHMH) clinic, other city-run testing site (ie, hospitals or clinics operated by the New York City Health and Hospitals Corporation), and non-city-run testing site. Health department clinics and other city-run sites were of particular interest because programmatic efforts to increase and routinely offer HIV testing were made at these sites during the analysis period.15 Health department clinics were further categorized into sexually transmitted disease (STD) clinics, tuberculosis (TB) clinics, and DOHMH-run clinics at city correctional facilities. Non-city-run sites included a broad range of providers categorized into the following types: hospitals, clinics, community-based organizations (CBOs), including sites serving the homeless, private medical providers, drug treatment programs, and miscellaneous sites or those that could not be categorized. Because some large providers could be categorized in more than 1 category (eg, a clinic may also function as a drug treatment program), the category best representing the majority of reported cases from each site was used.
The percentage of all positive Western blot tests considered new diagnostic positive tests (vs. repeat positive tests) was determined overall between 2004 and 2006, as well as in each year. In addition, the percentage considered diagnostic positive tests was determined within each testing site category. To assess yearly trends, the Mantel-Haenszel χ2 test was used to examine differences in proportions, whereas linear regression was used to examine differences in counts.
To characterize individuals with positive HIV test results, we collapsed all tests occurring in 2006 to the person level, as more than 1 test could be reported per person. Data for person-level analyses were limited to tests conducted in 2006 to focus on the most recent complete year available, given anticipated differences over time. Cases with a diagnostic positive Western blot occurring in 2006 were considered “new diagnoses.” All other cases with a positive Western blot in 2006 were deemed “repeat positive testers” because there was evidence of their HIV infection diagnosed sometime before 2006. Proportions of “new diagnoses” vs. “repeat positive testers” by demographic and clinical characteristics were determined. We used χ2 or Wilcoxon tests to identify significant differences at the 0.05 level.
To evaluate the possibility that repeat positive testing was sometimes conducted to confirm the status of patients initiating or reinitiating clinical care, we examined laboratory reporting of HIV-1 viral load or CD4+ lymphocyte testing among all repeat positive testers in 2006 in the 6 months before and 6 months after the repeat positive test. This interval was chosen because current recommendations for frequency of laboratory monitoring for HIV disease progression ranges from every 2 to every 6 months.16-18 A report of at least 1 such test in each 6-month period was used as a marker of HIV-related clinical monitoring. We hypothesized that repeat positive testers would be more likely to have clinical monitoring exclusively in the 6 months after their test than in the 6 months before their test, providing evidence that some repeat positive testing was being done to confirm HIV-positive status when returning to care. We did not conduct this analysis in new diagnoses because by definition they would not have any clinical care in the 6 months before first being diagnosed. Statistical analyses were performed in SAS 9.1 (SAS Institute, Cary, NC) and Stata 8.2 (StataCorp, College Station, TX).
After excluding duplicate results, we identified 35,594 positive Western blot tests conducted in New York City between January 1, 2004 and December 31, 2006. Of these, 31,504 tests (88.5%) were linked to 23,090 cases in the HIV Surveillance Registry. The remaining tests could not be linked to Registry cases for the following reasons: not enough information was available to properly investigate or link the test to a Registry case, such as a valid name (94.8%); the result was reported in someone who was not HIV infected (eg, perinatal HIV exposure that did not progress to HIV infection, or confirmed false-positive result) (4.6%); or the case was still under investigation at the time of the analysis (0.6%).
Of positive Western blot tests linked to cases in the Registry, 11,600 (36.8%) were considered diagnostic, with the remaining 19,904 tests (63.2%) representing repeat positive tests (Table 1). The percentage of tests that were diagnostic did not differ significantly by test year (P = 0.45). Among repeat positive tests, median time between initial HIV diagnosis and repeat positive test was 5.6 years (interquartile range: 2.7-10.8 years).
Table 1 shows substantial differences in the percentage of positive Western blots that were diagnostic over the 3-year period by testing site type. For example, 64.4% of positive Western blots conducted at health department clinics were diagnostic, whereas only 41.7% of those at other city-run sites and 33.8% of those at non-city-run sites were diagnostic. For each of the health department clinic types (STD, TB, and correctional facility), the percentage of tests that was diagnostic (ranging from 55.3% to 70.0%) was far greater than at non-health department sites as a whole. The percentage of tests that were diagnostic at health department clinics increased significantly over time, from 59.8% in 2004 to 69.0% in 2006 (P = 0.001) (Fig. 1A). STD clinics showed a significant increase over time, from 66.3% in 2004 to 75.2% in 2006 (P = 0.01), whereas similar increases in TB clinics and correctional facility clinics were not statistically significant (P = 0.10 and 0.65, respectively). Trends over time in counts of diagnostic tests were similar but did not reach statistical significance at these sites, although the observed increase in counts at STD clinics was marginally significant (P = 0.06).
Non-health department testing sites showed substantial variation in the overall percentage of positive tests that were diagnostic. Private physicians' offices had the highest percentage of positive tests that were diagnostic, at 58.3%. Whereas diagnostic tests comprised 42.8% of positive tests conducted at city-run hospitals, only 33.0% of those at non-city-run hospitals were diagnostic. Similarly, 31.9% of those conducted at city-run clinics were diagnostic compared with 28.6% at other clinics. CBOs and other sites reported 30.2% and 21.5% of positive tests that were diagnostic, respectively. The lowest percentage was observed in drug treatment sites, with 11.8% of positive tests that were diagnostic. Yet within most of these categories, there was significant variability by individual testing sites. When we examined individual sites that reported at least 10 positive confidential tests in the analysis period, the percentage of tests that were diagnostic varied from a low of 0.0%, that is, all positive tests were repeat positive tests (7 sites, including drug treatment sites, general clinics, and an extended care facility), to a high of 94.7% (a college health clinic). The percentage of tests that were diagnostic increased significantly over time in non-health department city-run sites as a whole (from 41.0% in 2004 to 44.2% in 2006, P = 0.049), non-city-run clinics (from 26.8% to 32.7%, P < 0.001), and CBOs (from 26.6% to 37.1%, P < 0.001) (Figs. 1A, B). Counts of tests that were diagnostic at non-health department sites did not significantly increase over time, although a marginally significant increase was observed in city-run hospitals (P = 0.07).
Positive Western blot tests conducted in 2006 corresponded to 8539 Registry cases. Of these, 3192 (37.4%) were defined as new diagnoses, whereas 5347 (62.6%) were repeat positive testers. Compared with new diagnoses, repeat positive testers were significantly older (median age 43 vs. 38 years), less likely to be male (64.0% vs. 72.1%) or white (9.8% vs. 16.0%) and more likely to be US born (62.9% vs. 49.7%) (all P's < 0.0001) (Table 2). Repeat positive testers were also more likely to have a history of injection drug use (29.5% vs. 6.2%, P < 0.0001) and less likely to be men who have sex with men (23.0% vs. 37.9%). Repeat positive testers were almost twice as likely to be diagnosed with AIDS compared with new cases (61.8% vs. 32.2%, P < 0.0001).
The percentage of repeat positive testers who showed evidence of HIV-related clinical monitoring exclusively in the 6 months after the repeat positive test was substantially higher than the percentage with monitoring exclusively in the 6 months before the repeat positive test (30.6% vs. 2.5%) (Table 3). This suggests that many repeat positive testers were being tested to confirm their HIV-positive status as part of initiation or reinitiation into clinical care. 62.4% of repeat positive testers had care both before and after their repeat positive test, whereas only 4.5% had no evidence of clinical monitoring at all in the 6 months before and after testing.
We found that most positive Western blot tests conducted in New York City do not diagnose new cases. Certain sites showed consistently higher proportions of truly diagnostic testing, including health department clinics and private physician offices. An increase in positive diagnostic testing conducted at health department clinics and other city-run facilities over time is likely the result of conscious efforts to increase HIV testing, especially in settings not previously targeted (ie, emergency departments and ambulatory care clinics) using “routine” HIV screening independent of risk factors.7,15,19 Linking positive Western blot testing results to surveillance data allowed us to document improved case finding by these city-run testing programs during a time of intervention.
Differences in repeat positive testing may reflect differences in the populations served or targeted by individual sites. For example, a youth clinic may report more positive tests that are truly diagnostic because its clients are younger and may be less likely to already know that they are HIV infected than older testers. In contrast, sites that provide HIV clinical care, such as hospitals or general clinics, may be consciously repeating testing in some patients already known to be HIV positive to confirm infection before treatment initiation. The fact that clinical monitoring was observed in almost one-third of repeat positive testers only after Western blot testing supports this interpretation; otherwise, we would expect the proportion of repeat positive testers exclusively receiving care before and after testing to be equal. Other explanations for repeat positive testing are that patients themselves may choose to be retested to confirm their HIV-positive status before receiving care or to see whether they are still infected.
An additional potential reason for repeat positive testing is that these sites or others like drug treatment programs may also be performing Western blot testing to certify status for social service benefits, which could also lead to a higher number of repeat positive tests performed. We could not directly confirm this interpretation from the data; however, some of the 62% of repeat positive testers whom we found to have evidence of HIV-related clinical monitoring both before and after testing may fit this category. In particular, persons with a history of injection drug use were more likely to be repeat positive testers, which may be a consequence of being less likely to produce documentation of a previous positive test result for certification purposes. Testing sites' contractual obligations to meet units of service might also account for some repeat testing by providers; however, we could not evaluate this possibility with our data.
Finally, sites that use financial incentives to encourage HIV testing, such as cash or transportation vouchers, may be unintentionally encouraging repeat testing of known positives. Financial incentives have been used to varying degrees of success in both TB and HIV screening programs,20,21 but their influence on true case-finding rates has not been described in the literature to our knowledge. Injection drug users' greater frequency of repeat testing may be related to their increased motivation to retest for monetary incentives because of their drug use interfering with their ability to work and their need for funds to support their drug use. Also, health department clinics, which do not offer patient incentives, reported higher proportions of new diagnoses than other facilities. However, health department STD clinics do not provide ongoing HIV-related clinical care, which could also explain their relatively high case-finding levels because patients at these sites have less valid reasons for repeat testing. Further, although we do not have city-wide information on which other providers offer HIV testing incentives, we are aware of at least 1 CBO temporarily reporting a high percentage of repeat positive tests as a consequence of a holiday-related food giveaway.
Although positive Western blot tests performed at non-city-run hospitals represented newly diagnosed cases less often than at city-run hospitals, they nonetheless represent one-quarter of all positive Western blots performed in the city (2968 of 11,600 tests). Given their substantial impact on overall city testing levels, future evaluation efforts aimed at improving case finding rates should include these sites.
We documented for the first time that many positive tests conducted in New York City do not diagnose new cases of HIV infection. This information suggests that testing programs may inadvertently misclassify new positive tests as diagnostic and overstate their case-finding effort yield. Because New York State's HIV surveillance does not include reporting of negative test results, we could not calculate actual case-finding rates for individual testing sites, which would be of further value for evaluation, for example, helping to identify potential structural barriers to testing or calculating cost-benefit ratios. Accordingly, we have begun to collect individual-level testing data from DOHMH-funded testing programs. These data are confidentially matched with surveillance history as in this analysis to determine which cases are truly newly diagnosed and provide aggregate case-finding rates back to individual programs. Preliminary results have demonstrated a 5-fold cost advantage to routine screening vs. targeted testing strategies in finding a newly diagnosed HIV case in New York City (median cost $9021 vs. $47,189).22
There are some limitations to our findings. First, the information on which tests were classified as diagnostic is based on the availability of HIV-related clinical data in the Registry, and these data in turn are based primarily on laboratory reporting and chart review. If previous HIV-related diagnostic or care-related medical history is not available from these sources, or if a test is incorrectly linked to the wrong case, then some misclassification may occur. We tried to minimize this possibility by instituting consistent algorithms and protocols when matching tests to previously known cases. Also, we linked observed increases in case finding in city-run facilities to known increases in investment in HIV testing. Similar initiatives to expand testing unknown to us may have been occurred at other sites, and therefore, we may not have captured the full impact of such initiatives. Finally, incomplete provider information from surveillance data reported from large hospitals with multiple testing sites limited how specifically we could categorize sites. Individual settings could have influenced repeat positive testing (eg, outpatient hospital clinics might perform more diagnostic testing, whereas specialized HIV clinics might perform more repeat positive testing to confirm clinical status). Such information would have allowed for a more nuanced evaluation of settings in which guidelines have recommended increased testing, such as emergency departments.9
Despite these limitations, this is the first attempt to use testing histories of positive cases from HIV surveillance data to evaluate testing programs to our knowledge. Repeat positive testing of known HIV-positive persons is common, and therefore, confirmatory Western blot testing alone is not enough for HIV testing programs to conclude that someone is a truly new case. For more efficient use of HIV prevention resources, testing programs can and should be evaluated using local surveillance information to ensure that cases identified are truly newly diagnosed and not repeat positive testers.
We would like to thank Matt Henn and Adam Karpati for the initial concept of this evaluation approach; Melissa Pfeiffer, Roy Shum, and Sonny Ly for data management and cleaning; and Wanda Davis for assistance with testing site categorization. We would also like to thank the field surveillance staff of the HIV Epidemiology and Field Services Program, New York City DOHMH, for their work in case investigation and chart abstraction and Bernard Branson and Lucia Torian for helpful feedback. The authors had full access to all the data and take full responsibility for the accuracy of the data analysis.
1. Nguyen TQ, Gwynn RC, Kellerman SE, et al. Population prevalence of reported and unreported HIV and related behaviors among the household adult population in New York City, 2004. AIDS. 2008;22:281-287.
2. HIV prevalence estimates-United States, 2006. MMWR Morb Mortal Wkly Rep. 2008;57:1073-1076.
3. Marks G, Crepaz N, Janssen RS. Estimating sexual transmission of HIV from persons aware and unaware that they are infected with the virus in the USA. AIDS. 2006;20:1447-1450.
4. Advancing HIV prevention: new strategies for a changing epidemic-United States, 2003. MMWR Morb Mortal Wkly Rep. 2003;52:329-332.
5. Use of social networks to identify persons with undiagnosed HIV infection-seven U.S. cities, October 2003-September 2004. MMWR Morb Mortal Wkly Rep. 2005;54:601-605.
6. Brewer DD. Case-finding effectiveness of partner notification and cluster investigation for sexually transmitted diseases/HIV. Sex Transm Dis. 2005;32:78-83.
7. Koo DJ, Begier EM, Henn MH, et al. HIV counseling and testing: less targeting, more testing. Am J Public Health. 2006;96:962-964.
8. Branson BM. State of the art for diagnosis of HIV infection. Clin Infect Dis. 2007;45:S221-S225.
9. 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(RR-14):1-17; quiz CE11-14.
10. Lindan CP, Avins AL, Woods WJ, et al. Levels of HIV testing and low validity of self-reported test results among alcoholics and drug users. AIDS. 1994;8:1149-1155.
11. Implementation of named HIV reporting-New York City, 2001. MMWR Morb Mortal Wkly Rep. 2004;52:1248-1252.
13. Guidelines for national human immunodeficiency virus case surveillance, including monitoring for human immunodeficiency virus infection and acquired immunodeficiency syndrome. Centers for Disease Control and Prevention. MMWR Recomm Rep. 1999;48(RR-13):1-27; 29-31.
14. Hanna DB, Pfeiffer MR, Torian LV, et al. Concurrent HIV/AIDS diagnosis increases the risk of short-term HIV-related death among persons newly diagnosed with AIDS, 2002-2005. AIDS Patient Care STDs. 2008;22:17-28.
15. Janssen RS. Implementing HIV screening. Clin Infect Dis. 2007;45:S226-S231.
16. Dybul M, Fauci AS, Bartlett JG, et al. Guidelines for using antiretroviral agents among HIV-infected adults and adolescents. Ann Intern Med. 2002;137(5 pt 2):381-433.
17. Yeni PG, Hammer SM, Carpenter CC, et al. Antiretroviral treatment for adult HIV infection in 2002: updated recommendations of the International AIDS Society-USA Panel. JAMA. 2002;288:222-235.
18. Yeni PG, Hammer SM, Hirsch MS, et al. Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel. JAMA. 2004;292:251-265.
19. Frieden TR, Das-Douglas M, Kellerman SE, et al. Applying public health principles to the HIV epidemic. N Engl J Med. 2005;353:2397-2402.
20. FitzGerald JM, Patrick DM, Strathdee S, et al. Use of incentives to increase compliance for TB screening in a population of intravenous drug users. Vancouver Injection Drug Use Study Group. Int J Tuberc Lung Dis. 1999;3:153-155.
21. Haukoos JS, Witt MD, Coil CJ, et al. The effect of financial incentives on adherence with outpatient human immunodeficiency virus testing referrals from the emergency department. Acad Emerg Med. 2005;12:617-621.
22. Tsoi BW, Woog V, Ramaswamy C, et al. Cost of finding one newly-diagnosed HIV case in New York City . Paper presented at: American Public Health Association Annual Meeting, October 27, 2008; San Diego, CA.
© 2009 Lippincott Williams & Wilkins, Inc.