Many undiagnosed HIV infections picked up when testing people who have other medical conditions, but implementation challenges remain

A Europe-wide project offering HIV tests to hospital patients with glandular fever symptoms, swollen lymph nodes, a low white blood cell count, a low platelet count or pneumonia has found that over 3% of tested patients had previously undiagnosed HIV. This significantly exceeds the level of 0.1% HIV prevalence at which routine HIV testing interventions are considered to be cost-effective. 

The results, presented to the British HIV Association (BHIVA) conference in Brighton last week, show the importance of getting non-HIV specialist clinicians to consider HIV testing in patients with a wide range of illnesses that might be indicative of HIV infection. However this remains challenging.

Since 2007, specialist HIV clinicians in Europe have been attempting to encourage other doctors to offer HIV tests to patients with specific ‘indicator conditions’. An indicator condition is one which is known or believed to be associated with an excess risk of being HIV positive. For example, some are sometimes caused by having a weakened immune system, while viral hepatitis is transmitted in similar ways to HIV. Guidance on testing people with indicator conditions was published in 2012.

The new research builds on earlier studies in order to clarify which medical conditions are associated with higher than average HIV rates. In 42 clinics in 20 European countries, doctors were asked to routinely offer an HIV test to all patients (aged 18 to 65) who presented with one of 14 indicator conditions.

In total, 9741 people took an HIV test, most of them for the first time in their life.

In those testing, the overall prevalence of undiagnosed HIV was very high at 2.5%. Seven in ten of those testing positive were diagnosed late (CD4 count below 350 cells/mm³) and half were diagnosed very late (below 200 cells/mm³).

Routine HIV testing is thought to be cost-effective in any population where the prevalence of HIV is 0.1% or above. In this study, overall prevalence in people with one of the indicator conditions was twenty-five times greater than that. Moreover, it was over 0.1% in 11 of the 14 conditions tested:

• Hepatitis B and hepatitis C co-infection, 9.6%.
• Ongoing symptoms similar to glandular fever (mononucleosis), 5.3%.
• Unexplained swollen lymph nodes (lymphadenopathy), 4.4%.
• Unexplained, ongoing low white blood cell counts (leukocytopenia) or low platelet counts (thrombocytopenia), 4.0%.
• Hospitalised with pneumonia, 3.2%.
• Unexplained peripheral neuropathy, 2.4%.
• Hepatitis C, 2.3%.
• Rash on the face, scalp or chest (seborrheic dermatitis), 2.0%.
• Hepatitis B, 1.2%.
• Abnormal changes in cervical cells (CIN II) or cervical cancer, 1.0%.
• Cancer of the lymph nodes (lymphoma), 0.7%.

Perhaps surprisingly, no cases of undiagnosed HIV were picked up in people with anal cancer or lung cancer. But this may also reflect limited engagement with HIV testing by clinicians in these specialities – only 53 and 144 people who had these conditions were tested.

The large variations in the number of people testing between medical conditions and between European regions do suggest that implementation of the “routine offer” may have been patchy, even if the researchers did not present data on this. Previous research has highlighted numerous challenges to the implementation of HIV testing by non-HIV specialists. While patients generally agree to be tested, clinicians may have other priorities or feel ill-equipped to deal with the topic of HIV.

In a related project (presented in 2013), the same group of researchers collected audit data from across Europe to assess day-to-day clinical practice regarding HIV testing in relation to six indicator conditions in which the link with HIV is widely accepted. These were tuberculosis, non-Hodgkin's lymphoma, anal cancer, cervical cancer, viral hepatitis and oral thrush. Overall, 72% of patients who had one of these conditions actually received an HIV test, but this varied considerably by medical condition. Whereas 31% of people with a cancer received a test, 97% of those with hepatitis did. Performance appeared to be better in participating clinics in eastern Europe than in western Europe.

Unsupportive guidelines

One of the problems in getting people with indicator conditions tested for HIV is that guidelines for the health condition often make no mention of HIV testing. The scale of the problem in the UK was assessed in another study presented at the BHIVA conference.

The UK’s HIV testing guidelines – prepared by HIV specialists – were published in 2008 and recommended HIV testing in all patients with 11 AIDS-defining conditions and 37 clinical indicator diseases. But as clinicians managing these diseases may not be aware of HIV guidelines, it’s important that guidance issued by the National Institute for Health and Care Excellence (NICE) and by specialist societies such as the British Society of Gastroenterology also recommend HIV testing.

Sixty clinical guidelines, all published since 2008, were identified. Less than half (26) made any mention of HIV testing and only a third (20) actually recommended it. While only six recent guidelines on AIDS-defining conditions were identified, only three of them recommended HIV testing. The authors recommend that HIV clinicians and specialist societies should engage with guideline development groups, including NICE, to make the case for HIV testing to be included in future guidelines. The new data from the Europe-wide study may help them do so.

Making testing the default option

One of the highest rates of undiagnosed HIV (5.3%) in the European study was for individuals with symptoms that could be glandular fever (mononucleosis). These include tiredness, fever, sore throat, skin rash and swollen lymph nodes. The same symptoms frequently occur during acute HIV infection as people seroconvert. (The person may not have glandular fever at all.)

The BHIVA conference heard that a technically straightforward automated system in primary care clinics can dramatically improve HIV testing rates in people with suspected glandular fever. The programme was implemented through the ViaPath laboratory at St. Thomas’ Hospital, which serves many general practices in an area of London with a very high HIV prevalence.

GPs can order glandular fever screens through a paper form or electronically. Previously, samples were tested for Epstein-Barr virus (the cause of glandular fever), cytomegalovirus and toxoplasmosis. Test results came back with a printed suggestion of recommending an HIV test but only a third of patients were tested.

The system for electronically requesting a glandular fever screen was improved so that an HIV test would be added, unless the doctor or patient specifically refused it. The result was that 68% of patients were tested for HIV and 1% of those tested were diagnosed with HIV.

Nonetheless, there remains great variation between different primary care clinics in terms of the proportion who turn down the test, suggesting that some doctors remain uncomfortable with the topic.