Death rates in people with HIV in resource-limited settings are likely to be more substantially reduced where people are diagnosed with HIV earlier, CD4 counts can be carried out regularly in untreated people, and treatment started at a CD4 count around 350, according to results of a mathematical modelling exercise carried out by epidemiologists at Imperial College in London.
Although there have been calls for people in Africa to receive antiretroviral treatment earlier, the authors say their model is the first to calculate the impact of earlier treatment for health systems and societies hard hit by HIV.
The findings were published in the March 2008 edition of PLoS Medicine by Timothy Hallett, Geoff Garnett and colleagues in the Department of Infectious Disease Epidemiology at Imperial College, London.
Mathematical, computerised models are used by epidemiologists to predict how events will unfold using a mixture of data taken from real-life cohorts and parameters or rules that are determined by the researchers based on current knowledge. The model is then set to run with an element of chance built into that is designed to reflect the unpredictable nature of human life.
The parameters of the model were determined using data on HIV disease progression and CD4 cell decline from several African cohort studies, while the composition of the model’s cohort of 1000 hypothetical individuals was determined based on the age and sex distribution of new HIV infections in rural Zimbabwe between 1998 and 2002, and life expectancy data from Zimbabwe.
The model showed that in the absence of treatment, infected individuals would lose 22 life-years and die at an average age of 39 years.
Survival after initiation of antiretroviral therapy was modelled using data from the ART-LINC collaboration, which documents treatment response in resource-limited settings.
In the absence of long-term follow-up data on ART patients in resource-limited settings, treatment responses were modelled in terms of best, medium or worst. The `medium` response was assumed to be 90% of people still living after four years of treatment initiated with a CD4 count between 200 and 349 cells/mm3, and 75% survival in those who initiated treatment with a CD4 count below 50 cells/mm3.
The study analysed the effect on survival of nine different rules for starting HIV treatment. Using only symptoms as an indicator of when to start treatment, in a context where uptake of voluntary counselling and testing is low, the average number of life years saved among all those infected in the population – not just those treated - is only two to four years. For those treated, life expectancy would be extended by six to 17 years.
If annual CD4 counting were introduced, but HIV diagnosis rates remained low, life expectancy for those treated would rise to between eight and 18 years, but for all those infected in the population, the number of life years saved among all those infected in the population would also be modest – between three and five years.
Greater frequency of monitoring and a higher level of HIV diagnosis would lead to much more substantial improvements in life expectancy – between 17 and 27 years after infection for those treated.
With high VCT uptake and high rates of referral from antenatal clinics, the number of people treated would grow threefold compared with the baseline scenario, in which only sick people who present for care are treated. While the number of life-years saved per person treated would not differ radically if the proportion of people diagnosed with HIV increased from 51% to 82%, the total number of person-years in the population lost to HIV/AIDS could be halved or even reduced by two-thirds in some scenarios.
Inevitably, higher rates of diagnosis and monitoring would lead to greater utilisation of the health care system. The model found that earlier diagnosis, regular monitoring and an increased number of people on ART would lead to 50% more clinic appointments, assuming that those with CD4 counts below 350 cells/mm3 were seen more frequently than those with higher CD4 counts.
However the researchers say that further cost-effectiveness analysis would be needed using costs from specific settings to determine the most cost-effective changes to current practice in initiating treatment, frequency of monitoring and referral of patients.
They say that current estimates of the need for antiretroviral therapy over the longer term are flawed because they rely on data derived from the observation of individuals in the last few years before death, and do not take into account changes in monitoring or diagnostic strategy.