EACS: Hospital admissions and appointments predict subsequent HIV diagnosis and death

A study from Denmark presented at the eleventh European AIDS Conference (EACS) adds support to the idea that pre-HIV-diagnosis morbidity from other diseases is significantly correlated with HIV diagnosis and subsequent mortality, and that therefore a list of ‘indicator diseases’ might be a good way of targeting people likely to have or acquire HIV – see separate story.

Denmark has a unified register of hospital admissions, discharges and out-patient attendances, so epidemiologist Nicolai Lohse was able to trawl through the hospital records of 2,099 patients with HIV and over 200,000 HIV-negative controls matched for gender, age and locality. He calculated the Charlson Co-Morbidity Index (CMI) for each person during their hospital stay or visit.

The CMI is a list of common illnesses in which each disease diagnosis contributes a point to the person’s CMI score. The CMI predicts the one-year mortality for a patient who may have a range of co-morbid conditions such as heart disease, AIDS (though not HIV infection per se), or cancer (a total of 22 conditions). Each condition is assigned with a score of 1,2,3 or 6 depending on the risk of dying associated with this condition. Then the scores are summed up and given a total score that predicts mortality risk.

Lohse found that death rates in people with HIV were related to their pre-diagnosis CMI score, and more strongly so than in members of the general public.

Lohse calculated a quality called the ‘population attributable risk’ (PAR) of having a particular CMI score and the risk of subsequent HIV diagnosis and death.

PAR is an indirect measurement of the amount a particular condition or behaviour contributes to another health risk. It is a measure of correlation, not causation, so we can say that something contributes a particular PAR to another health problem without knowing exactly how it does so.

PAR can also be larger in some populations than others. If, for instance it was noted that being a smoker seemed to increase the risk of death in people with HIV more than it increased the risk of death in other people, we could say that the PAR of smoking to mortality in people with HIV is larger than it is in the general population – as may well be the case.

Lohse found that the PAR of having had a CMI-listed condition was 10.3% in people later diagnosed with HIV and 7.5% in HIV-negative controls. In other words, if you were found to have HIV you were also more likely than the general public to have turned up at hospital in the past with one or more of a list of common diseases.

The PAR of having a high CMI score for eventual death in people with HIV was also greater than it was for the general public. Lohse was able to calculate that 21% of the PAR of eventual death in people with HIV was due to CMI conditions reported pre-diagnosis, and another 16% of PAR was added by hepatitis C.

The PAR of CMI conditions for eventual death was consistently greater in people with HIV and especially in people also coinfected with hepatitis C than in the general public. For instance the PAR of having had three or more CMI conditions for later death was 6.6% in members of the general public, 16% in people with HIV, and 24% in people with HIV and hepatitis C. In other words nearly a quarter of mortality in people with HIV and HCV could be linked to conditions acquired and reported pre-diagnosis.

This PAR does not say that 21% of people with HIV actually died from conditions they caught before diagnosis, but that there was a 21% correlation between having these conditions and later death. But it does support the idea that morbidity with diseases apparently unrelated to HIV is more common in people with HIV, even before they are diagnosed.