Even in well-treated populations, loss to follow-up limits community viral suppression

Several papers presented at the Eleventh International Congress on Drug Therapy in HIV Infection last week showed that even in populations with consistent access to care and antiretroviral therapy (ART), high rates of loss to follow-up mean that it is difficult to push the proportion of people with HIV who are virally suppressed, and therefore potentially non-infectious, beyond about 50% of the total HIV-positive population.

The cascade problem

One new subject of discussion in both HIV prevention and treatment in the last year has been the so-called HIV “treatment cascade”, the idea that the efficacy of treatment as a prevention tool in HIV is inevitably compromised by the fact that not everyone, and perhaps not even a majority, of people with HIV, are on treatment and have an undetectable viral load.

This is because not everyone with HIV is diagnosed; not all the diagnosed are retained in care; not all those in care are taking ART; and not everyone on ART is virally suppressed (has a viral load which is so low as to be undetectable using current viral load tests).

Estimates of the proportion of people with HIV who are on ART and undetectable range from 54% in the UK (Brown) to as low as 25% in the case of black gay men in the US (Cohen, Hall).

This attrition is due to a different mix of factors in different countries. For instance, while in the UK a higher proportion of people with HIV are undiagnosed than in the US, a much higher proportion of the diagnosed are on ART.

Several presentations in Glasgow suggested that in the context of western Europe, where most people have access to free treatment, people dropping out of care and being lost to follow-up (LTFU) in the main cause of cascade attrition.

Italian attrition

In one study presented at Glasgow from the city of Modena in Italy, for instance (Lazzaretti), even though it was estimated that there was a very high rate of diagnosis amongst the HIV-positive population, a relatively high rate of loss to follow-up meant that only 45-49% of the HIV-positive population was likely to be on ART and virally suppressed.

This study looked at data from the Modena HIV Surveillance Cohort over a long period, looking at care patterns in 962 people with HIV diagnosed between the beginning of 1996 and the end of 2011. Two-thirds of them were men and 58% had acquired HIV through heterosexual sex (only 17% were gay).

During the study period, 92.6% were linked to care, meaning that they returned for at least one appointment after their diagnosis.

During the study period, 96 people (10.8%) died. If these were excluded, then 78.6% remained in care over the study period or, to put it another way, more than 20% were lost to follow-up. Of those who remained in care, only 55 people were not taking ART – in other words 91.2% were – and of the 570 who were on ART, then 516 (90.5%) had a viral load below 50 copies/ml.

Thus 54% of the entire diagnosed group was recorded as being on ART and with an undetectable viral load over the study period. But 'the entire group' includes people who died – and people who disappeared from care.

Given that those who died could not be on treatment, and that whether people lost to care had died could not be ascertained, then a maximum of 65% (of patients definitely known to be alive and in care), and a minimum of 60% (of patients not definitely known to have died), was on ART and undetectable.

However the researchers also estimated that while testing and diagnosis rates in Modena are high, at least 13% of the local HIV-positive population is undiagnosed. This means that between 45 and 49% of the local HIV-positive population is in fact on ART and virally suppressed, depending on whether the death rate in undiagnosed people is higher or lower than it was in those retained in care.

The Achilles heel of the treatment cascade, then, is the proportion of patients who are lost to follow-up. In the Modena study, these were on average younger (mean age 32 versus 36), more likely to be injecting drug users (IDUs: 17% lost to follow-up were IDUs versus 7% who stayed in care) and less likely to be gay (14% LTFU were gay men versus 22% in care). But the biggest difference was that 50% of those who were LTFU had not been born in Italy, versus 31% of those in care.

This suggests that it is the high mobility of people with HIV – who may be immigrants, have unstable housing, or switch care for confidentiality reasons – that may get in the way of the efficacy of treatment as prevention.

Lost to follow-up in the UK in Leicester...

This in turn brings up an interesting question: how many of those listed as ‘lost to follow-up’ have really dropped out of HIV care, and how many have simply moved or are getting their HIV care elsewhere and have not told their old clinic? A study from Leicester, England (Wood) suggested that a high proportion may simply have moved, some becoming untraceable through the normal surveillance system in the process.

In this study of 1274 people with HIV known to the Leicester Royal Infirmary, the initial definition of ‘not being in care’ was that the patient had not had a CD4 or viral load test result in the last six months.

In fact, 788 patients (61%) had not been seen in the last six months. On further investigation however, 455 people (35%) had in fact transferred care to another clinic, and this was recorded. Another 61 (4.8%) had died, and 51 (4.2%) were known to have moved out of the UK. That left 218 (17%) who had definitely dropped out of care. However 88 of these could, at least potentially, be traced, because they were still registered with local GPs and had addresses on record.

After a new electronic patient record system was introduced in September 2011, which alerted staff every time someone had not attended for six months, these patients were sent recall letters. A “small number” of patients did re-engage with care, and via partner notification there were several new diagnoses of HIV as a result.

However this still left 130 patients (10.2%) whose whereabouts were unknown. Of these, 54% were women, 76% were black African, and 26% were lost to follow-up in the first year after diagnosis. Nearly half (48%) had been on ART at the time they disappeared.

This study shows that by proactive contacting of patients the number truly lost to care can be reduced.

...and in the UK in general

The pattern seen in Leicester is repeated in many HIV clinics, another study showed. A review of data from the UK Collaborative HIV Cohort (UK CHIC), a database of 43,000 patient records from 20 of the largest HIV clinics in the UK, shows that, while the vast majority of patients diagnosed with HIV attend at least one follow-up appointment at an HIV clinic, less than half are involved in ‘consistent care’ – whose definition is the same as the Leicester study, namely that there are no gaps longer than six months between CD4 and/or viral load test results.

The review looked at data from 33,210 male patients only. Their average age at diagnosis was 36, two-thirds were white and two-thirds were gay; one in five was heterosexual and one in eight of African origin.

Initial engagement in care was excellent: only 3.5% of people diagnosed with HIV at a UK CHIC clinic failed to turn up for a follow-up appointment, although there were some variations: heterosexual men were 23% less likely to turn up for a second appointment and men under 30 were 14% less likely.

However only 12,644 patients (44%) met the criterion for ‘consistent engagement’, having no gaps in their test results. Again, heterosexual men were 24% less likely than average to be consistently engaged, the under-30s were 30% less likely, and the over-45s 44% more likely. This could be because younger people were more likely to be recently infected and thus not taking ART and, conversely, older people would mainly be on ART, receiving ART being strongly associated with consistent engagement.

There was no association between ethnicity and consistent engagement and, promisingly, recently diagnosed patients were more likely to be consistently engaged. Although this could be because they had had less time to develop gaps in their care record or go missing, the difference was large: patients diagnosed between 2008 and 2011 were nearly twice as likely (188%) to consistently attend their clinic as patients diagnosed between 2000 and 2004.