Delays in linkage to care after HIV diagnosis pose the biggest barrier to treatment access

Following individuals through the cascade of care, from HIV diagnosis to treatment and undetectable viral load, rather than taking snapshots of performance, has led researchers on a major South African study to pinpoint linkage to care after testing HIV positive as the biggest weakness of treatment programmes seeking to achieve the UNAIDS 90-90-90 targets, an article in the journal The Lancet HIV reports.

The study, carried out through the Africa Health Research institute in the Hlabisa district of KwaZulu-Natal province in South Africa, found that less than half of the population with HIV in the district had linked to care within eight years, despite the fact that 82% were aware of their HIV infection.

As well as raising questions about how to improve linkage to care, the findings also highlight weaknesses in the ways that countries and treatment programmes are reporting treatment cascade data. Treatment cascades have become a popular way of reporting programme outcomes and have become enshrined in the UNAIDS 90-90-90 target, which aims to achieve 90% of people with HIV diagnosed, 90% of people diagnosed on treatment and 90% of people on treatment with undetectable viral load. Treatment cascades report the proportions of people who reach each stage and help to identify areas of health system weakness.

Problems with HIV care cascade reporting

The authors have previously identified a number of problems with studies of the cascade of care. The first issue is the lack of information about the earliest stages of the cascade: HIV status, HIV diagnosis and the intervals between infection and diagnosis, and between diagnosis and linkage to care. It is very difficult to capture this information without monitoring the whole population in a geographic area, but without such monitoring, studies may tend to overestimate performance in the early stages of the cascade – or to discount the importance of these stages.

A lack of stages in a cascade may also disguise problems in a health system, or make it difficult to identify where people are being lost from care. For example, differentiating between people linked to care, people evaluated for antiretroviral therapy (ART) eligibility and people retained in care after evaluation may be useful in a setting where rapid ART initiation after diagnosis is not the norm. Being able to determine what proportions of people are lost between linkage to care and evaluation for treatment, and what proportion are lost after evaluation for treatment, will provide useful information about any problems with the local model of treatment initiation.

But, cascades may be difficult to interpret if they are just snapshots of a particular moment in time. For example, reporting on the proportions of people diagnosed with HIV and with undetectable viral load in the same month is not describing what has happened to a single group of people. Instead, it lumps together groups of people separated in time: everyone who has ever been diagnosed with HIV, and everyone now on treatment who has undetectable viral load. The cross-sectional method cannot take into account changes in the population, such as an increase in the number of adolescents, and so may cause us to miss the reasons for losses from the cascade. Whereas 65% of the population with undetectable viral load today may be women over 30, the entire population of people with HIV may be 65% women under 30, who may have very different reasons for staying in or dropping out of care.

Another difficulty, reported by several research groups, is the challenge of measuring losses and reentries into the treatment cascade. For example, a longitudinal cohort study within a specific geographic area may underestimate the number of people who are still in care, because it does not capture information about people who have left care in one place and sought it elsewhere. On the other hand, a longitudinal cohort study of the cascade may be better suited to picking up on the proportion of people who disappear from care for long periods, but who are not entirely lost.

The KwaZulu-Natal study

Taking a more rigorous approach – by following every individual from the point at which they were identified as HIV positive in a population serosurvey – allowed the researchers to track what proportion of those individuals passed through each of the following stages:

• learnt of their HIV positive status
• linked to care
• became eligible for ART
• began ART
• achieved undetectable viral load.

The researchers drew on data from a longitudinal population cohort study in northern KwaZulu-Natal, in which people were tested for HIV through household visits from 2006 (this surveillance test result was not disclosed to the study participant).

The study population comprised everyone who tested positive between 2006 and 2011. The progress of each individual was tracked through a district HIV clinical care and treatment database for each year up to the end of 2013, in order to measure what proportion of people progressed from one stage to another, and how long it took them to do so.

A total of 5205 people with HIV were followed for a total of 24,301 person-years. Seventy-three per cent were female, the mean age of participants was 33 and the mean CD4 cell count at first measurement was 299 cells/mm³. Mean CD4 cell counts at first measurement rose somewhat over time, from 270 cells/ mm³ in the period 2006-07 to 309 cells/mm³ in the period 2010-11.

Of those who tested HIV positive during surveillance, 82% subsequently learnt of their HIV status. Sixty-seven per cent were already aware of their HIV-positive status at the time they were tested in population surveillance. Yet, despite knowing their status, the majority of participants had not linked to care. Despite the fact that 77% of people knew their status within four years, only 25% of people with HIV had linked to care, and of those not linked to care at the time they were tested in population surveillance, less than half had linked to care within eight years.

This large gap between diagnosis and linkage to care could not be explained by deaths. A total of 598 of the study population died during the follow-up period

Overall, 45% of people with HIV eventually linked to care and 39% of people with HIV were eligible for ART under South African guidelines. Prior to April 2010, treatment was provided to people with CD4 cell counts below 200; after this time treatment was also provided to pregnant women with CD4 cell counts between 200 and 350, and to people with tuberculosis. Anyone with a CD4 cell count below 350 was eligible for treatment after July 2011.

Once people became eligible for treatment, uptake was high and initiation was relatively quick. Ninety per cent started treatment, within a median of three months, and 94% were virally suppressed.

How quickly did people move from one stage in the cascade to the next?

Perhaps the most sobering insight provided by the longitudinal-linked approach to cascade measurement is the estimation of the time it took for people to move from one stage in the cascade to the next.

The first stage of the cascade measured how long it took for people who had tested positive on the surveillance test to learn their HIV status through voluntary counselling and testing at health facilities in the district or through household testing. It took a median of 52 months for people to learn their HIV status and a median of 52 months after learning their HIV status to link to care. Linkage to care was defined as the first clinic visit or first recorded CD4 cell count or viral load

The proportion who linked to care did not improve over time. Despite the fact that knowledge of HIV status improved over time, when the researchers compared the proportion of people with HIV who had been linked to care four years after their first positive surveillance test between the three periods 2006-07, 2008-09 and 2010-11, they found almost no change in the proportion who had linked to care (50%, 53%, 52%).

Once linked to care it took a median of 19 months for people to become eligible for ART and 3 months to start ART after becoming eligible. ART initiation slowed down significantly in 2010-11 compared to the preceding periods; it took a median of 5.5 months to start treatment after becoming eligible in this period, compared with 2.1 months in the period 2008-09. However, this delay was balanced by a significant decline in the 2010-11 period in the proportion of people who died in the interval between becoming eligible for treatment and starting treatment. The proportion who died fell from 5.4% to 3.7%.

The researchers say that people diagnosed more recently may have taken longer to link to care because they were less often symptomatic, and may not have perceived HIV treatment as an urgent necessity.

“These new data make clear [..] that, no matter how effectively ART reduces transmission, therapeutic goals will be difficult to reach,” George Rutherford of UCSF and Andrew Anglemyer of Naval Postgraduate School remark in an accompanying Comment article in The Lancet HIV.

They note that in the ANRS 12249 trial 47% of people were linked to care within one year of diagnosis. In comparison, 28% of people in the KwaZulu-Natal study were linked to care within two years of learning their HIV status in the period 2010-11. Neither study population received special interventions to encourage linkage to care, nor was there any significant difference in linkage to care in the cluster-randomised ANRS 12249 study between the intervention group and the control group communities, even though the intervention group would be offered immediate treatment upon linkage.

Learning more about interventions that can successfully link people to care will be critical for achieving the UNAIDS 90-90-90 targets. Speeding up linkage to care will also be critical for reducing onward transmission of HIV. Noah Haber and colleagues suggest that improving linkage to care will require interventions that address structural barriers to care, such as subsidised transport to clinics and building more clinics, as well as measures to increase demand, such as financial incentives, mobile phone reminders to link to care and motivational counselling at the time of diagnosis.