Since 2004, the New York City Department of Health and Mental Hygiene (NYC DOHMH) has recommended that New York City residents get tested for HIV, as a step to leading longer and healthier lives.1 In subsequent years, NYC DOHMH has promoted HIV testing and has funded agencies to expand HIV testing activities.2 In 2006, the Centers for Disease Control and Prevention (CDC) released revised recommendations for HIV testing in health care settings.3 HIV screening should be performed routinely for all patients aged 13–64 years; testing should not be based on assessment of patients' risks. CDC also recommended using opt-out screening for HIV; HIV tests would be performed routinely unless a patient explicitly refuses to take an HIV test. At the time of CDC's recommendations, New York State law required that patients must sign an HIV consent form before testing, and thus opt-out testing was not allowed in New York.4 A review of state laws in 2006 showed that the majority of states also had HIV testing requirements that were inconsistent with CDC recommendations.5
In 2010, the New York State Legislature enacted a law that authorized significant changes in HIV testing in New York State.6 One key provision of the law was that HIV testing must be offered to all persons between the ages of 13 and 64 receiving hospital or primary care services, with limited exceptions. The law also expanded available options for obtaining consent for HIV testing, including using documented oral consent and integrating consent for HIV testing with the general consent for medical care. The law went into effect September 1, 2010. Regulations that provide guidance on the interpretation and implementation of the law were published in November 2011, and adopted in February 2012.
The NYC DOHMH's Primary Care Information Project (PCIP) presented a unique opportunity to examine the law's impact, as part of the mandated evaluation. Founded in 2005 to improve the quality of care for low-income New Yorkers through health information technology, PCIP helps more than 9000 providers and 1200 ambulatory care practices using multiple electronic health record (EHR) platforms to improve quality of care, achieve Meaningful Use certification, and adopt health care system innovations.7–10 In 2011, PCIP developed the Hub Population Health System (the Hub), a dynamic tool that allows DOHMH to send queries to collaborating practices and receive back automated counts of patients who meet the queries' criteria.11 No protected health information is exchanged. The Hub connects in real-time to partner practices using the eClinicalWorks EHR system. In 2013, the Hub was connected to 650 practices serving 1.6 million patients annually and 4 million patients since 2009.
The Hub allowed DOHMH to objectively capture institutional behavior change in the health care sector over time, rather than relying on self-report by institutions. Using the Hub, we aimed to measure changes in HIV testing in ambulatory practices during a baseline period, the first year of the law's implementation (year 1) and the second year of the law's implementation (year 2). We hypothesized that although there would be no significant change during the baseline period, the likelihood of testing would increase significantly in both year 1 and year 2. We also hypothesized that we would see larger changes in age groups targeted by the testing law (ages 13–64) compared with those not targeted by the law (ages 65 and above). This project was found to be exempt by the NYC DOHMH institutional review board.
We used the presence of test results from an HIV screening test in the EHR as a proxy indicator of HIV test offers. Offers were not documented in a standard format in the EHR and could not be captured. Test results were identified by the presence of unique Logical Observation Identifiers Names and Codes (LOINC codes). LOINC codes for initial HIV screening tests were collected from the national LOINC compendium. Confirmatory, viral detection, polymerase chain reaction, CD4, and viral load tests were not included. This approach captured initial screening test results that had been returned electronically to the EHR through a laboratory interface or entered manually by office staff. If paper test results were not entered into the EHR, they could not be captured.
The outcome of interest for this study was the quarterly HIV testing rate, defined as patients with at least 1 HIV screening result out of all patients with an office visit in the reporting period. Rates were stratified by age, defined as those targeted by the law (teenagers aged 13 to 17, adults aged 18–64) and those not targeted by the law (older adults, defined as adults aged 65–100). Because HIV testing consent processes might differ for adults and teenagers accompanied by parents or guardians, we separated these 2 age groups.
We defined 3 time periods of interest for our quarterly rates around the testing law, which went into effect in September of 2010. Baseline was defined as Q4 2009–Q3 2010, the year preceding the law. Year 1 was defined as Q4 2010–Q3 2011. Year 2 was defined as Q4 2011–Q3 2012. Quarterly testing rates were calculated among all patients with an office visit between Q4 2009 and Q3 2012.
Study clinics were assigned a practice type. Federally Qualified Health Centers and look-alike facilities were classified as community health centers (CHCs). One hospital outpatient clinic was eligible for the study and was grouped with the CHCs for analysis because of similar facility function and size. A small practice was defined as any practice with fewer than 10 providers that was not a CHC or hospital-affiliated clinic. Study practices were assigned a borough (county) location based on the primary clinic site. Five practices had sites across 2 or more boroughs and were assigned to the borough of their largest site.
To ensure that study practices were receiving laboratory results electronically and were actively using the EHR to see patients, we established inclusion criteria. Practices had to see at least 50 patients in the first observed quarter (Q4 2009) and have electronically stored laboratory results for at least 25 patients by the end of 2009. These levels were set to allow small practices with lower patient volume to contribute, while eliminating practices that were in the process of implementing the EHR.
Data were collected with the Hub Population Health System using an aggregate query mechanism described fully by Buck et al.11 Numerator and denominator concepts were translated into SQL code queries by a programmer, and code was independently reviewed by 2 analysts. Queries were run on practices during January 2013. No individual-level information was transmitted, and no protected health information was received by the Health Department.
Practice and patient characteristics were described at the start and end of baseline, year 1 and year 2. Total and stratified crude testing rates were calculated at the population level. Crude rates were plotted over time and mapped by borough using ArcGIS 10.0 (Environmental Systems Research Institute Inc., Redlands, CA).
We evaluated the change in likelihood of testing from the first to final quarter of baseline, year 1 and year 2 using univariate and multivariate generalized estimating equation models, clustered by unique practice identifier. In each model, our outcome was quarterly practice testing rate. Multivariate models were built iteratively by adding significant covariates to the base generalized estimating equation model for each period. The final multivariate model minimized the Quasilikelihood under the Independence Model Criterion for 2 of the 3 periods. The final model retained time (first vs. final quarter of each period), age category, interaction of age category and time, practice type (CHC vs. small practice), and borough. No other interaction terms were significant. The year 1 and year 2 models also included a baseline performance variable, indicating whether a practice had an HIV test result for any patient by the end of the baseline period. This definition of baseline performance outperformed other rank indicators, such as HIV testing rate quartiles. Racial, ethnic, and socioeconomic data about patients seen in study practices were not available for inclusion in the model. We report crude and adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) for each period. We also report the adjusted ORs for significant covariates. All statistical analysis was performed using SAS 9.2 (SAS Institute Inc., Cary, NC).
Characteristics of Study Population
Queries were assigned to 590 New York City ambulatory care practices that were enabled on the Hub and had agreed to share data with PCIP. Of those practices, 560 returned query responses, but only 359 were using the EHR in 2009 and had data. Thirty-six practices saw fewer than 25 patients in the first quarter of the study, and an additional 105 practices were not receiving laboratory results electronically. After applying laboratory and patient volume inclusion criteria, 218 practices were eligible for the study. Of the 218 retained practices, 205 were small practices, 12 were CHCs, and 1 was a small hospital outpatient clinic that was analyzed as a CHC.
Small practices included in our study had 330.3 full-time equivalent (FTE) providers, defined as physicians (MDs or DOs), nurse practitioners, or physician assistants (Table 1). On average, small practices had 1.6 FTEs (minimum 1.0, maximum 9.2). CHCs had a total of 212.9 FTEs, with a mean of 16.4 (minimum 1.8, maximum 48.3). At the start of the study (Q4 2009), small practices saw 185,708 patients, with the majority being adults (72.4%), followed by older adults (19.4%) and teenagers (8.2%). In the same quarter, CHCs saw more adults (81.5%) and fewer older adults (10.0%). By the end of the study (Q3 2012), small practice patient volume increased by 11.4%, whereas CHC volume increased 39.6%. Age distribution remained consistent (<1% point difference). CHCs were more likely to have tested at least 1 patient by the end of baseline than small practices (76.9% vs. 52.1%).
Trends in HIV Testing in the Ambulatory Setting
Crude quarterly testing rates increased more during year 1 than in baseline, overall, and in each age group (Table 2). The overall crude testing rate increased 0.5% points in baseline (2.8%–3.3%) and 1.8% points in year 1 (3.2%–5.0%). Year 2 gains surpassed baseline but were smaller than year 1 (5.0%–5.7%). Crude testing rates among older adults not targeted by the law increased very modestly from the start of baseline to the end of year 2 (0.2%–0.9%) (Fig. 1). Baseline gains were low in teenagers (2.6%–3.0%) and adults (3.4%–4.0%). Adult rates jumped in year 1 (3.8%–6.0%). Teenage testing rates increased more slowly but rose 2.1% points from the start of year 1 to the end of year 2.
At the beginning of the study, CHCs were testing at a higher rate than small practices, and CHCs improved more from start to end. Small practice crude testing rates increased less than 1% point in each time period. CHC testing rates decreased in baseline (8.3%–7.2%) and jumped in year 1 (7.0%–11.1%), remaining steady in year 2. Practices that had not tested any patient for HIV by the end of baseline improved little across the study (0.0%–1.0%). Practices that tested in baseline increased 2.7% points in year 1 and 4.3% points across the study.
At the borough level, crude testing rates increased the most in the Bronx, rising from 2.1% to 8.4% across the study period (Fig. 2). In the same time frame, testing rates increased 3.2% points in Queens, 2.2% points in Manhattan, and 1.6% points in Brooklyn. Staten Island rates decreased slightly (0.3%–0.2%). The Bronx and Manhattan improved more than twice as much in year 1 as in baseline.
Changes in Odds of HIV Testing Before and After the HIV Testing Law
Table 3 explores the change in the odds of testing from the first to final quarter of the baseline period, year 1, and year 2, as well as the factors that were significantly associated with testing during each period. In the univariate model, there was no significant change during baseline (OR: 1.2, CI: 0.9% to 1.5%). During year 1, there was a 50% increase in odds of testing (OR: 1.5, CI: 1.0% to 2.2%). During year 2, there was a 10% increase in odds of testing (OR: 1.1, CI: 1.0% to 1.3%). Changes were significant in year 1 (P = 0.04) and year 2 (P = 0.03).
The multivariate model also showed no significant change during the baseline period (OR: 1.4, CI: 0.8% to 2.3%), after adjusting for CHC status, borough, patient age group, the interaction of age and time, and baseline performance. From first to final quarter of year 1, the adjusted odds of testing doubled (OR: 2.0, CI: 1.3% to 3.1%). After adjusting for covariates, there was no longer a significant change during year 2 (OR: 1.1, CI: 0.9% to 1.5%).
Across every time period, CHCs were more likely to test a patient for HIV, as compared with small practices (P = 0.01). Patient age was a significant predictor of a practice's likelihood of testing during baseline, year 1, and year 2. Throughout the study, the odds of testing a teenager or an adult were at least 3.5 times greater than the odds of testing an older adult (P < 0.01). During the baseline period, Manhattan was the only borough with significantly higher odds of testing, as compared with Staten Island (OR: 7.1, CI: 2.1% to 24.0%). By year 1, practices in all other boroughs were significantly more likely to test a patient for HIV than those in Staten Island, and these differences persisted in year 2. In year 1, the strongest predictor of testing was baseline performance. Practices that did not test during baseline continued not to test in year 1, whereas practices that were already testing in baseline improved (OR: 17.0, CI: 4.7% to 61.6%). In year 2, this effect was smaller but still significant (OR: 7.6, CI: 2.3% to 25.6%).
Our evaluation showed that the odds of HIV testing in New York City outpatient practices increased significantly in the 2 years after the law became effective, compared with the baseline period. The adjusted odds of testing improved more than 200% during the first post-law year, supporting our hypothesis that the implementation of the HIV testing law may have contributed to an increase in screening. Improvement was more modest in year 2 in both crude rates and model results.
The changes in ambulatory testing practices seen in this study mirrored NYC survey results. In the Community Health Survey, a telephone survey weighted to represent adult New Yorkers, the percent of respondents tested in the past year increased from 31.4% in 2010 (CI: 29.9% to 32.9%) to 34.6% in 2012 (CI: 33.1% to 36.2%).12 The small but significant increase on an individual level echoes the modest increase in our study, which captures only outpatient testing behavior. However, the final crude testing rate in our study was small (5.7%). Although encouraging, these changes might not be big enough to increase detection of HIV positive cases.
In our study, age groups that the law targeted for the routine offer of HIV testing were more likely to be tested than older adults, who were not targeted. Crude testing rates for older adults remained largely static, compared with post-law improvements in teenagers and adults. Community Health Survey estimates of testing are also consistently low in older adults.12 To further increase testing among New Yorkers, lawmakers should consider the costs and benefits of raising the upper age limit for the mandatory offer of testing. A recent study suggests that screening adults aged 55–75 years, under certain conditions, may be cost-effective.13 Further research is needed to determine the optimal age to stop routine screening. To support expansion of HIV screening among older adults, Medicare HIV testing eligibility would need to be expanded to cover routine screening in those persons older than 65 years.14
CHCs were substantially more likely to test a patient for HIV than a small practice in all time periods, after adjusting for confounders. CHCs also improved significantly more after the HIV testing law than small practices. CHCs may be able to more effectively respond to changes in HIV guidelines because of bigger institutional capacity and a mission-driven focus on underserved populations. The importance of baseline performance in predicting improvement suggests that an HIV testing law may be more effective at improving testing rates in practices already screening than in convincing doctors to add HIV screening to the care they deliver. Policymakers may need to consider other levers to reach patient populations seen in smaller health care settings and practices that are not currently testing for HIV.
Concurrent HIV testing promotion activities limited our ability to ascribe the changes seen in this study solely to the 2010 testing law. The NYC DOHMH has been encouraging and funding HIV testing in NYC since 2004. CDC provided funding for HIV testing in NYC starting in 2007 as part of the Expanded Testing Initiative.15 In 2008, NYC DOHMH launched The Bronx Knows, a borough-wide HIV testing campaign featuring 75 organizations spanning hospitals, clinics, community-based organizations, colleges/universities, faith-based institutions, local commercial establishments, and community boards.16,17 The Bronx Knows campaign likely drove the large increase seen in the Bronx across the study period. A similar initiative was launched in Brooklyn in 2011. Concurrently, DOHMH was reaching out to outpatient providers directly in an initiative patterned after pharmaceutical detailing to educate practices about HIV testing. An internal evaluation in 2011 found that 59% of 1110 providers at 470 detailed practices were aware of the testing law. NYC DOHMH activities may have directly contributed to the increase observed in this study. They may also have enhanced the impact of the law by encouraging practices to develop testing protocols and establish testing infrastructure. Our study controlled for borough and testing at baseline, but residual confounding likely persisted.
Using EHR data introduced other limitations to the study. We could not capture offer of a test or paper test results. This study almost certainly underestimates the true levels of HIV testing in the outpatient setting. Because of the query execution demand of the system, we did not exclude HIV-positive individuals from our denominator. In a separate query, we calculated that the percent of our sample patients with a diagnosis of HIV or AIDS in 2012 was 0.8%. Because we captured only electronically stored laboratory results, trends observed may have represented an improvement in the data entered into the EHR over time, rather than a true change in testing behavior. The patient volume at small practices and especially CHCs increased over time. This is likely an artifact of increasing proficiency on the EHR after adoption, reflecting both more patient capacity and increased visit documentation on the EHR. This could introduce bias if patients were differentially entered into the EHR based on health status. Finally, the 218 practices selected for this study were all “early adopters” that were using an EHR by Q4 2009 and serving at least 10% Medicaid, uninsured, or low-income patients to be able to join PCIP. Our results might not be generalizable to more affluent practices or practices that were not using an EHR.
Although evaluating the testing law using EHR data introduced challenges, it also conferred benefits. Our measures were objective and sidestepped recall entirely. We received query answers from 560 of the 590 assigned practices, giving us a 94% “response rate,” although only 359 of those practices had 2009 EHR data. Because the study leveraged the existing Hub Population Health System infrastructure, this evaluation was performed at no additional cost beyond analyst time. Data like these can be available with unprecedented speed, providing near real-time snapshots of diverse health behaviors and conditions. This kind of rapid evaluation of change in the health care environment may become more feasible in the future, as the federal Meaningful Use incentive program pushes EHR adoption in inpatient and outpatient settings and health information exchange expands.18
Our evaluation showed that after the implementation of an HIV testing law, there was an increase in HIV testing among NYC ambulatory practices. Quarterly testing rates remained modest, but significant improvement was seen in CHCs, in practices screening patients for HIV at baseline, and in age ranges targeted by the law. This study suggests that HIV testing legislation is one of many tools that should be used simultaneously in a comprehensive strategy to increase HIV testing.
The authors thank Sam Amirfar, Laura Jacobson, and Yinjie Hu for their work on this project.
1. NYC DOHMH. Take Care New York. 2014. Available at: http://www.nyc.gov/html/doh/html/about/tcny.shtml
. Accessed February 5, 2014.
2. Report of the NYC Commission on HIV/AIDS. New York (NY): New York City; 2005.
3. 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; quiz CE11-14.
4. NYS Public Health Law Article 27F 2000 (NYS).
5. Wolf LE, Donoghoe A, Lane T. Implementing routine HIV testing
: the role of state law. PLoS One. 2007;2:e1005.
6. NYS. Chapter 308 of the Laws of 2010: HIV Testing
Law Mandated Report, August 2012. Albany (NY): New York State Department of Health; 2012.
7. Mostashari F, Tripathi M, Kendall M. A tale of two large community electronic health record extension projects. Health Aff (Millwood). 2009;28:345–356.
8. Ryan AM, Bishop TF, Shih S, et al.. Small physician practices in New York needed sustained help to realize gains in quality from use of electronic health records
. Health Aff (Millwood). 2013;32:53–62.
9. Bardach N, Wang J, De Leon S, et al.. Effect of pay-for-performance incentives on quality of care in small practices with electronic health records
: a randomized trial. JAMA. 2013;310:1051–1059.
10. Wang JJ, Sebek KM, McCullough CM, et al.. Sustained improvement in clinical preventive service delivery among independent primary care practices after implementing electronic health record systems. Prev Chronic Dis. 2013;10:120341.
11. Buck MD, Anane S, Taverna J, et al.. The Hub Population Health System: distributed ad hoc queries and alerts. J Am Med Inform Assoc. 2012;19:e46–e50.
12. NYC DOHMH. Epiquery: NYC Interactive Health Data System. 2014. Available at: http://nyc.gov/health/epiquery
. Accessed February 4, 2014.
13. Sanders GD, Bayoumi AM, Holodniy M, et al.. Cost-effectiveness of HIV screening in patients older than 55 years of age. Ann Intern Med. 2008;148:889–903.
14. Medicare. 2014. Available at: http://www.medicare.gov/coverage/hiv-screening.html
. Accessed February 5, 2014.
15. Expanded and Integrated Human Immunodeficiency Virus (HIV) Testing for Populations Disproportionately Affected by HIV, Primarily African Americans. New York (NY): Centers for Disease Control and Prevention; 2007:58.
16. 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.
17. NYC DOHMH. The Bronx Knows HIV Testing
Initiative: Final Report. 2011. Available at: http://www.nyc.gov/html/doh/downloads/pdf/ah/bronx-knows-summary-report.pdf
. Accessed January 10, 2014.
18. Hsiao CJ, Hing E. Use and characteristics of electronic health record systems among office-based physician practices: United States, 2001-2012. NCHS Data Brief. 2012:1–8.