Getting HIV-Positive People Into Care -- and Keeping Them There

Sobering Numbers on Staying in Care

After HIV diagnosis and linkage to care, the third leg in the HIV-health tripod is staying in care, or retention. Using data from the CDC and other sources, researchers estimated that 79% of the 1.1 million HIV-positive people in the United States get diagnosed, 59% enter care, and only 40% stay in care.³⁸ (See Figure 1 in the article "Late HIV diagnosis.") Looking at specific HIV populations, other investigators usually make similarly depressing estimates of retention in care.

Meta-analysis of HIV linkage and retention studies with data from May 1995 through 2009 figured that, of 75,655 people who entered care, only 59% kept multiple HIV visits averaged across assessment intervals ranging from 6 months to 3 to 5 years.³⁹ A British study of 16,595 people in HIV care determined that 44% did not have a CD4 count for a year or more, and 40% of that group fell out of care for the duration of follow-up.⁴⁰ A 12,304-person EuroSIDA analysis published in 2008 defined loss to follow-up as no clinic visit, CD4 count, or viral load assay after January 1, 2006; the researchers counted 2712 people (22%) who met those criteria.⁴¹ Recruitment of the analyzed cohorts began in May 1994 and ended in December 2005. In contrast to these findings, another prospective cohort, the French Hospital Database on HIV, found in 2006 that only 2950 of 34,835 people (8.5%) did not have a medical visit for at least 12 months after their last visit in 1999.⁴² But loss to follow-up was 16.8% among people diagnosed with HIV in the past year.

Retrospective analysis of 2619 male US veterans who started antiretroviral therapy after January 1, 1998 determined that 36% went at least 3 months without seeing their HIV clinician in the first year of therapy.⁴³ Among 1636 people who entered the University of North Carolina Center for AIDS Research prospective clinical cohort from January 2001 through January 2008, 414 (25%) dropped out of care, defined as missing appointments for 18 months.⁴⁴ A New York City study of 842 people diagnosed with HIV from July 1 to September 30, 2005 found that 650 (77%) started care within 3 months.⁴⁵ Of those, only 45% maintained regular care, defined as at least one clinic visit every 6 months.

Why People With HIV Drop Out of Care

Many variables that shorten the odds of starting HIV care also explain why people later quit. Whether the study group lives in the United States or Western Europe, factors that make poor retention more likely often include younger age, minority or immigrant status, substance use, and a very low CD4 count (Figure 4). People with AIDS, in contrast, seem more likely to keep appointments than people without AIDS.

Figure 4. An array of factors independently predicted dropping out of care in nine US and European studies. From study to study, the most consistent predictors were younger age, not having AIDS, injection drug use, and numerous factors indicating social marginalization. The appendix following the references in this article spells out results of these studies.

To analyze predictors of retention in HIV care, RITA! sifted results of 9 studies that used multivariate analysis to pinpoint retention predictors. The four biggest studies scrutinized large cohorts -- 34,835 people in the French Hospital Database on HIV,⁴² 12,304 in EuroSIDA,⁴¹ 2619 men cared for at US Veterans Affairs (VA) centers,⁴³ and 2411 or 1924 HIV-positive women (depending on the analysis) in the six-site US Women's Interagency HIV Study (WIHS).⁴⁶ (WIHS kept tabs on cohort visits, not on primary care HIV visits.) Smaller populations included 1636 people seen at the University of North Carolina HIV clinic,⁴⁴ 1007 patients in five French HIV clinics,⁴⁷ 650 people in New York City,⁴⁵ 567 people at the University of Alabama HIV clinic,⁴⁸ and 398 people in Los Angeles clinics.⁴⁹ The VA study differed from the others in the stringency of its definition of poor retention -- missing a clinic visit in any one of four quarters in the first year of antiretroviral therapy.⁴³ The appendix following the references to this article details how each of these studies defined retention and outlines key results.

Younger age consistently boosted the risk of poor retention in the VA study, WIHS, and the clinic-based studies,^43-48 while older age favored good retention in EuroSIDA.⁴¹ EuroSIDA linked female gender to better retention,⁴¹ while the Los Angeles study found that Latina and African-American women were more likely to keep clinic appointments than Latino or African-American gay men.⁴⁹ The all-men VA study determined that black veterans ran a one-third higher risk of poor retention than white veterans.⁴³

How HIV disease status affects retention seems a little trickier to reckon, but in the end none of these disease status results defy logic. First, people with an AIDS diagnosis were more likely to see their HIV physician regularly in the French Hospital Database,⁴² EuroSIDA,⁴¹ and the University of North Carolina clinic;⁴⁴ this consistent result reflects the likelihood that people who have an AIDS disease or had one earlier want their AIDS treated right away or want to avoid repeating the experience. Conversely, the New York City study identified early (non-AIDS) HIV infection as a predictor of poor retention.⁴⁵

People with higher viral loads proved more likely to drop out of care at the University of North Carolina,⁴⁴ in WIHS,⁴⁶ and in the French clinic study.⁴⁷ Although those findings at first seem at odds with the AIDS results, it makes sense that people with higher viral loads are dropout risks because high loads often reflect poor antiretroviral adherence; and people with poor antiretroviral adherence are candidates for poor appointment adherence. High loads may also indicate lack of antiretroviral therapy, and taking antiretrovirals correlates with good retention, as discussed below.

WIHS⁴⁶ and the five French clinics⁴⁷ linked a lower CD4 count to poor clinic attendance, while EuroSIDA⁴¹ and the Los Angles study⁴⁹ figured a higher CD4 count favored good clinic attendance. The VA and University of Alabama studies found that a higher initial CD4 count predicted poor retention,^43,48 while the French clinic study found that a lower initial CD4 count predicted good retention.⁴⁷ What do these mixed results mean? There's no way to know for sure because studies that demonstrate associations do not establish the direction of causality. But there's plenty of room for speculation:

A very low CD4 count may signal advanced HIV disease and imminent death, which would tend to keep people out of the clinic. Indeed, some people listed as "lost to follow-up" may have died. A higher CD4 count can reflect good adherence to antiretrovirals and to care in general and thus explain good clinic attendance. On the other hand, a high CD4 count when first entering care probably reflects asymptomatic disease, and people with no symptoms have less motivation to keep clinic appointments. The timing of the CD4 measurement is critical, as the five-clinic French study showed by evaluating CD4 count two ways.⁴⁷ In this 1007-person study, a low initial CD4 count favored good retention,⁴⁷ probably for the same reason that an AIDS diagnosis favors good retention: sick people who just learned they have HIV want to get care. The same French study found that a lower CD4 count during care made poor retention more likely,⁴⁷ perhaps because people whose CD4 count stays low despite being in care are not taking antiretrovirals and are missing appointments, or they are getting too sick to come to clinic.

The studies that isolated antiretroviral therapy as a retention predictor revealed the two sides of the treatment coin. EuroSIDA members who had begun antiretroviral therapy were more likely to stay in care,⁴¹ while people not taking antiretrovirals in the five French clinics were more likely to drop out.⁴⁷ One could theorize endlessly on what is cause and what is effect in these associations, but few would disagree with one interpretation: getting patients to the point where they can start antiretrovirals helps keep them from turning truant.

Minority status, figured different ways, heightens the risk of poor clinic attendance. In both French studies,^42,47 being born outside of France raised the risk of poor retention in care, and black race made poor retention more likely in the VA survey⁴³ and the New York City study.⁴⁵ But in the WIHS study of US women, white women were more likely than black women to miss WIHS study visits 7 through 10.⁴⁶ Why white race correlated with poor study attendance in WIHS is not clear; many women in this six-center cohort are socially marginalized regardless of race or ethnicity.

People without a primary care provider ran a higher risk of missing cohort appointments 7 through 10 in WIHS⁴⁶ and of dropping out of care in the French clinic study.⁴⁷ Women with no health insurance or with temporary housing were more likely to miss appointments in WIHS,⁴⁶ while having no phone number (a surrogate for poverty or transience) in the French study predicted dropping out of care.⁴⁷ Having insurance favored good retention at the University of North Carolina HIV clinic.⁴⁴ Substance abuse upped the odds of poor retention in WIHS⁴⁶ and the University of Alabama clinic,⁴⁸ while moderate (but not heavy) alcohol drinking made poor retention more likely in WIHS.⁴⁶ The Los Angeles analysis of Hispanic or black women or gay men was the only study that weighed the impact of HIV disclosure status.⁴⁹ People who told more "network members" they had HIV were more likely to stay in care. In the same study, Latino gays who felt more gay stigma made fewer clinic visits. In the VA study, men with a chronic nonviral disease such as diabetes, hypertension, or ischemic heart disease had about a 20% lower risk of poor retention (OR 0.81, 95% CI 0.66 to 0.99).⁴³