Persons Living With HIV in the United States: Fewer Than We Thought

Abstract

Objectives: 

To estimate the number of persons living with HIV (PLWH) in the United States and to describe their care status.

Methods: 

Estimates of diagnosed PLWH in New York City and other 19 jurisdictions based on HIV case reporting were compared with those based on HIV laboratory reporting. A revised HIV care continuum was constructed based on previously published data.

Results: 

The estimate of PLWH based on HIV case reporting was 25.6% higher than that based on HIV laboratory reporting data in New York City. There were 819,200 PLWH in the United States at the end of 2011 (plausible range: 809,800–828,800), of whom 86% were diagnosed, 72% were retained in care (≥1 care visit in 2011), 68% were on antiretroviral therapy, and 55% were virally suppressed (≤200 copies/mL).

Conclusions: 

The current method based on HIV case reporting may have overestimated PLWH in the United States. While we continue cleaning HIV case reporting data to improve its quality, we should take the opportunity to use comprehensive HIV laboratory reporting data to estimate PLWH at both the national and local levels.

The Centers for Disease Control and Prevention (CDC) estimated that there were 1,201,100 persons living with HIV (PLWH) in the United States at the end of 2011.¹ For many years, HIV case reporting data were the sole registry data source that could be used to estimate the number of PLWH. This changed as comprehensive HIV-related laboratory data became reportable.^2,3 HIV case reporting is the process whereby a new diagnosis of HIV infection or AIDS is reported to the surveillance registry, and HIV laboratory reporting is the process whereby an HIV-related laboratory test (eg, a viral load test) is reported to the registry.

The accuracy of the estimates based on HIV case reporting depends on the accuracy of the counts of new diagnoses and deaths among HIV-infected persons. Individuals can only be diagnosed with HIV infection once, but in the real world, people can be “diagnosed” more than once and entered in one or more HIV registries multiple times because of migration between surveillance jurisdictions, the use of aliases or name changes, or data entry errors. A recent study reported that the estimate based on HIV case reporting in Switzerland had greatly overestimated the number of PLWH in that country.⁴

In the United States, each jurisdiction reports HIV cases to the CDC, not by patient name but by soundex (a phonetic algorithm for indexing names by sound), and cases are deduplicated through a process entitled the “Routine Interstate Duplicate Review.”⁵ The CDC use the Routine Interstate Duplicate Review to identify possible duplicate cases across jurisdictions based on patient soundex, date of birth, and sex at birth,⁵ and each pair of possible duplicates is then investigated by the reporting jurisdictions to determine if they are the same individual based on patient's name, date of birth, social security number, and other available information. Duplicate cases can be missed in this process if cases have missing information, alias names, name changes, or data entry errors.

The local health departments routinely match their HIV registry and death registries to ascertain deaths among persons with HIV. However, undercounting can occur because of incomplete data, common names, inconsistencies in names and other matching variables, or data entry errors. Persons who die outside of the United States and dependent territories may not be included in the US national death registry and therefore would not be identified through death registry matching.⁶

The current method of basing PLWH counts on HIV case reporting can overestimate PLWH in the United States if duplicate cases are missed by the deduplication process and deaths are missed by death registry matches. We aimed to estimate the number of PLWH in the United States and describe their care status.

METHODS

We compared the number of diagnosed PLWH in New York City (NYC) and other 19 jurisdictions based on HIV case reporting with those based on HIV laboratory reporting and constructed a revised national HIV care continuum based on previously published data.

Diagnosed PLWH in Selected Jurisdictions

In the 2013 CDC HIV Surveillance Report, the CDC published HIV care outcomes among diagnosed PLWH in 19 jurisdictions using CD4 and viral load tests as proxy measures for the receipt of HIV care.² These 19 jurisdictions required reporting of all CD4 and viral load test results and had reported to the CDC at least 95% of the test results they had received by December 2012. We compared the estimates of diagnosed PLWH in NYC and these 19 jurisdictions between HIV case reporting and laboratory reporting. The estimate of PLWH in each of the 20 jurisdictions based on HIV case reporting was obtained from the CDC reports.^2,7 The estimate based on HIV laboratory reporting for NYC was obtained from the NYC HIV laboratory reporting data.^3,8 With no access to other jurisdictions' data, the estimates for the other 19 jurisdictions were based on the number of patients with ≥2 care visits at least 3 months apart in 2010 in the CDC reports and the following 2 assumptions: (1) assumed completeness of laboratory data reporting and (2) assumed proportion of patients with ≥2 care visits at least 3 months apart.^7,9,10

The number of diagnosed PLWH (N) equals the number of patients with ≥2 care visits at least 3 months apart divided by the product of the assumed completeness of laboratory data reporting (p₁) and the assumed proportion of patients with ≥2 care visits at least 3 months apart (p₂), ie, N = n/p₁p₂. We made the estimates based on 2 assumption scenarios.

Scenario 1 is that all jurisdictions had complete reporting of laboratory data (p₁ = 100%) and their proportion of patients with ≥2 care visits at least 3 months apart was as high (p₂ = 79.1%) as NYC.

Scenario 2 is that in other jurisdictions the completeness of laboratory data was 85% and the proportion of patients with ≥2 care visits at least 3 months apart was 67.2% (both 15% lower than those in NYC).

HIV Care Continuum in the United States

The HIV care continuum, sometimes referred to as the HIV treatment cascade, is a model that has been widely used to monitor HIV-infected persons at the population level along the steps from infection to viral suppression.^11–13 We constructed a revised HIV care continuum in the United States using data from the CDC surveillance report on the proportion of PLWH aware of their HIV status, from published journal articles on the proportion of PLWH retained in care, and from the Medical Monitoring Project (MMP) on the number of PLWH retained in care, prescribed antiretroviral therapy (ART), and virally suppressed. The Medical Monitoring Project is a supplementary surveillance project sponsored by the CDC to provide representative, population-based data on clinical status, care, outcomes, and behaviors of HIV-infected persons receiving care in the United States.^{1,3,9,14–16}

Retained in Care

PLWH retained in care was defined as ≥1 care visit in a calendar year.¹⁷ In the MMP, retention in care was defined as ≥1 care visit during a 4-month population definition period (PDP). In the 2011 cycle, the MMP included patients in the sample who received care during January 1–April 30, 2011, and estimated that 478,433 PLWH were retained in care in the United States in 2011.^1,15 The NYC surveillance data showed that among PLWH who were retained in care in 2011, only 80.85% had at least 1 visit during January 1–April 30, the same period as the MMP's PDP. By applying the local NYC proportion to national data, we estimated that there were 591,754 (478,433/80.85% = 591,754) PLWH retained in care in the United States in 2011. Starting with the number of PLWH retained in care, we estimated the other bars of the HIV care continuum using data from multiple local and national sources.

Diagnosed

To estimate the diagnosed PLWH in the United States, we divided the number of patients retained in care (N = 591,754) by the proportion of diagnosed PLWH retained in care. The CDC reported that 47% diagnosed PLWH were retained in care in the United States.¹ At the local level, when patient migration was not accounted for, lower proportions were reported, eg, 38% in Chicago, 68% in Los Angeles County, 64% in NYC, 49% in Philadelphia, 71% in San Francisco, and 73% in Seattle^3,18; when patient migration was accounted for, a higher proportion was reported, eg, 91% in NYC and 84% in Seattle.^3,14 The proportion of diagnosed PLWH retained in care in Seattle after migration was accounted for (84%) was used as the proportion of diagnosed PLWH retained in care in the United States to estimate diagnosed PLWH.

Persons Living With HIV

To estimate PLWH in the United States, we divided the number of diagnosed PLWH by the proportion of PLWH with a known HIV-positive status, which was estimated to be 86% by the CDC.¹ The Upper and lower margins of uncertainty of the estimated number of PLWH were calculated based on the upper (87%) and lower (85%) confidence limits of the estimated proportion.^19,20

On ART

To estimate PLWH on ART in the United States, we multiplied the number of PLWH retained in care by the proportion of PLWH who were on ART among those retained in care, which was estimated to be 92.31% from the MMP.¹

Virally Suppressed

Viral suppression was defined as ≤200 copies per milliliter. To estimate PLWH with a suppressed viral load in the United States, we multiplied the number of PLWH on ART by the proportion of PLWH with a suppressed viral load among those on ART, which was estimated to be 81.90% from the MMP.¹ For the last 2 steps (on ART and viral suppression) in the HIV care continuum, the margins of uncertainty were not calculated because the confidence intervals were not reported by the MMP and the confidence intervals were likely to be very narrow given the project's large sample size (N = ∼10,000).^15,21

RESULTS

Diagnosed PLWH in Selected Jurisdictions

The estimates of diagnosed PLWH in NYC based on HIV case reporting and HIV laboratory reporting were 97,128 and 77,334, respectively (Table 1). The estimate based on HIV case reporting was 25.6% higher than the estimate based on HIV laboratory reporting.

TABLE 1.: Comparison of the Estimated Number of Persons Aged 13 Years or Older With HIV Infection Diagnosed by the End of 2009 and Alive at the End of 2010 in Selected Jurisdictions, by Data Source

Based on HIV case reporting, the estimates of PLWH in the other 19 jurisdictions ranged from 174 in North Dakota to 126,377 in New York State. Based on HIV laboratory reporting, the estimates ranged from 92 in North Dakota to 93,105 in New York State under assumption scenario 1, and ranged from 128 in North Dakota to 128,864 in New York State under assumption scenario 2. Compared with the PLWH estimates from HIV laboratory reporting under assumption scenario 1, the estimates from HIV case reporting were higher in all 19 jurisdictions, ranging from 34.5% higher in Iowa to 220.9% higher in Illinois; under assumption scenario 2, the estimates were higher in 15 of the 19 jurisdictions, ranging from 1.0% higher in South Carolina to 131.8% higher in Illinois (Table 1).

HIV Care Continuum in the United States

Figure 1 shows the CDC HIV care continuum, with 1,201,100 PLWH in the United States at the end of 2011, of whom 86% were diagnosed, 40% were retained in care, 37% were on ART, and 30% were virally suppressed.

FIGURE 1.: CDC estimates of HIV-infected persons engaged in selected stages of the continuum of HIV care in the United States, 2011. Estimates were rounded to the nearest 100. Adapted from Bradley H, et al. MMWR 2014; 63(47):1113–1117.

Figure 2 shows the revised HIV care continuum. There were 819,200 PLWH in the United States at the end of 2011 (plausible range: 809,800–828,800), of whom 86% were diagnosed, 72% were retained in care, 68% were on ART, and 55% were virally suppressed.

FIGURE 2.: Revised estimates of HIV-infected persons engaged in selected stages of the continuum of HIV care in the United States, 2011. Estimates were rounded to the nearest 100.

DISCUSSION

By comparing the estimates of PLWH based on HIV case reporting with those based on HIV laboratory reporting, we found that HIV case reporting may have overestimated PLWH in the United States. The estimate of diagnosed PLWH in NYC (N = 97,128) based on HIV case reporting was 25.6% higher than the estimate (N = 77,334) based on HIV laboratory reporting. The estimate based on HIV case reporting was higher in all 19 jurisdictions under assumption scenario 1 and in 15 of 19 jurisdictions under assumption scenario 2. It is possible that the estimate based on HIV laboratory reporting was an underestimate because of incomplete data reporting. But it is more likely that HIV case reporting overestimated the number of PLWH in many, if not all, of these 19 jurisdictions for the following reasons: (1) these 19 jurisdictions had relatively mature HIV laboratory reporting and were selected by the CDC to be included in HIV surveillance reports,² (2) HIV case reporting produced a higher estimate in all 19 jurisdictions under assumption scenario 1 and in 15 of 19 jurisdictions under assumption scenario 2, and (3) the estimate based on HIV case reporting was 34.5%–220.9% higher than the estimate based on HIV laboratory reporting under assumption scenario 1 and 1.0%–131.8% higher under assumption scenario 2.

While the method based on HIV case reporting can overestimate the number of PLWH if duplication, migration, and missed death matches are not accounted for the method based on HIV laboratory reporting does not have such limitations, because only in-care patients, who received at least 1 CD4 or viral load test in the year, are included in the analysis and out-of-care patients are estimated from in-care patients who were previously out of care and recently returned using a statistical weighting method.³ A patient with duplicates in the registry will have his CD4 and viral load tests linked to one patient identification number and will be counted as one in-care patient. Patients who have migrated out of the jurisdiction or died will have no CD4 and viral load tests linked to them and will not be counted or included in the analysis.

Our analyses are subject to a number of limitations. First, our estimate of PLWH in NYC based on HIV laboratory reporting depends on the completeness of reporting. Without complete laboratory data, our laboratory data–based method will definitely underestimate the number of PLWH in a jurisdiction. An NYC Department of Health and Mental Hygiene internal evaluation showed that the completeness of the NYC HIV laboratory reporting was greater than 97% (NYC Department of Health and Mental Hygiene, unpublished data), which means that the underestimation caused by incomplete laboratory data reporting in NYC may be minimal compared with the overestimation from using HIV case reporting data.

Second, we do not have access to other jurisdictions' HIV registries and made estimates based on 2 arbitrarily selected assumption scenarios. In assumption scenario 1, we assumed that all jurisdictions had complete laboratory data (p₁ = 100%) and their proportion of patients with ≥2 care visits at least 3 months apart was as high (p₂ = 79.1%) as that of NYC. These assumed proportions may be too high for some jurisdictions given the wider availability and accessibility of HIV care in NYC. However, we also provided estimates based on a conservative assumption scenario (scenario 2) that in other jurisdictions the completeness of laboratory data was 85% and the proportion of patients with ≥2 care visits at least 3 months apart was 67.2% (both 15% lower than those in NYC). It is unlikely that in these 19 jurisdictions both the completeness of laboratory data and the proportion of patients with ≥2 care visits at least 3 months apart were 15% lower than those in NYC. With such a conservative assumption scenario, we still found that HIV case reporting overestimated the number of PLWH in 15 of 19 jurisdictions.

Third, we constructed a revised HIV continuum based on the number of PLWH retained in care from the national MMP data and the proportion of diagnosed PLWH retained in care from Seattle (84%). Although the MMP is a large and rich source of information of persons in HIV care, it has its own limitations of incomplete reporting and low response rate, which may lead to underestimate the number of PLWH retained in care.^8,15,22 The proportion of retention in care in Seattle may be higher than the national average, and we may underestimate PLWH in the United States by applying a higher proportion of retention in care. However, 84% was likely to be an underestimate for Seattle, because the authors reported that after simply removing patients without laboratory results for 5 or more years, the proportion increased to 88%.¹⁴ For comparison, NYC had an even higher proportion (91%).³ Eighty-four percent was the lowest among the 3 (84%, 88%, and 91%) and may be a good estimate for the United States. The CDC estimated that in 2011 only 35% of diagnosed PLWH in the United States were virally suppressed.¹ The low national estimate may be caused by (1) overestimation of the denominator, PLWH, and (2) misclassification of in-care patients who had no care visits during the MMP PDP, January–April 2011, as out-of-care patients. Local jurisdictions applying a refined method reported a higher viral suppression proportion: 70% in NYC based on a likelihood method that estimates the probability of an HIV-infected person being in NYC, 72% in NYC based on a statistical weighting method that allows patients who were previously out of care and recently returned to represent out-of-care patients, 78% in San Francisco based on a method that removes patients from the denominator who were known to have moved out and not receiving care in San Francisco and who were likely to have moved out because of no laboratory test results reported to their HIV registry in the previous 5 years, and 67% in Seattle based on a method that uses multiple data sources and individual case investigation findings.^3,23,24 Our revised 2011 national estimate (55%) was comparable with those from other high-income countries, eg, 68% in Switzerland (2012), 62% in Australia (2013), 61% in the United Kingdom (2013), 59% in Denmark (2010), 58% in the Netherlands (2013), and 52% in France (2010), and higher than those from some developing countries, eg, 40% in Brazil (2013) and 29% in sub-Saharan Africa (2013).²⁵

This study is not an attempt to precisely quantify the number of HIV-infected persons in the United States but rather to examine the extent to which the current estimates change when derived from evidence available at the state and local levels. We believe that we have provided enough evidence to strongly suggest overestimation of PLWH based on HIV case reporting data. The CDC has begun to address this issue because they have received more accurate surveillance data from states and territories, revising their estimates downward, from 1,178,250 PLWH in 2008¹⁹; 1,148,220 in 2009²⁶; and 1,144,500 in 2010,² despite that, we would expect the number of PLWH in the United States to increase because of improved survival among HIV-infected persons and stable incidence of new HIV infections.^27–30 The overestimation may be the result of overcounting of new diagnoses across and within jurisdictions because of duplicates and undercounting deaths among PLWH. While we continue cleaning HIV case reporting data to improve its quality, we should take the opportunity to use comprehensive HIV laboratory reporting data to estimate PLWH at both the national and local levels. Additional research is needed to identify the best way to operationalize the use of HIV laboratory data to estimate PLWH.