CROI: Ugandan study confirms low CD4 count, low body weight, TB and anaemia major risk factors for death after starting ART

This article is more than 15 years old. Click here for more recent articles on this topic

A Ugandan study exploring why mortality is so high during the first few months of starting antiretroviral therapy (ART) among people with HIV in resource-limited settings has confirmed that advanced immune suppression, active tuberculosis, severe anaemia and low body mass are major risk factors for mortality after treatment is started.

The study, presented on Monday at the Fourteenth Conference on Retroviruses and Opportunistic Infections in Los Angeles, provided more evidence that mortality drops dramatically after the first few months on treatment, in this particular case, in people with HIV receiving ART through a home-based programme in rural eastern Uganda.

But — with the exception of wasting — the relative contribution of different illnesses to the mortality observed in people on ART appeared to remain fairly constant over a median of two years of follow-up.



A type of fungal infection usually affecting the membrane around the brain, causing meningitis. It can also affect the lungs and chest.

body mass index (BMI)

Body mass index, or BMI, is a measure of body size. It combines a person's weight with their height. The BMI gives an idea of whether a person has the correct weight for their height. Below 18.5 is considered underweight; between 18.5 and 25 is normal; between 25 and 30 is overweight; and over 30 is obese. Many BMI calculators can be found on the internet.


A mental health problem causing long-lasting low mood that interferes with everyday life.


A shortage or change in the size or function of red blood cells. These cells carry oxygen to organs of the body. Symptoms can include shortness of breath, fatigue and lack of concentration.

haemoglobin (HB)

Red-coloured, oxygen-carrying chemical in red blood cells.

What stood out, however, were several baseline risk factors for mortality (whether early or late) in people on ART including a low CD4 cell count, high viral loads, a history of or current tuberculosis (TB), anaemia, a low body mass index (BMI) and, notably, depression. Dr David Moore, of the British Columbia Centre for Excellence in HIV/AIDS suggested that screening for and managing these risk factors could have a survival benefit.

Background on the high early mortality on ART in resource-limited settings

Mortality among people on ART in sub-Saharan Africa has been reported to range between 5-19% in the first year on therapy, which is much higher than the rates observed in industrialised countries. But there has been limited information on specific clinical conditions that might be related to early mortality on ART in African settings.

Several studies have also shown that the high rate of mortality decreases after a few months on ART — which has led some to surmise that the conditions associated with early deaths while on ART might be different than those associated with later deaths.

So the investigators in this study tried to collect mortality data in a cohort of people on ART to try to differentiate between early and later mortality, and to identify clinical conditions and risk factors associated with mortality during each time period.

Study methods and population

The study examined data from adults with HIV participating in the Home-Based AIDS Care (HBAC) project in Tororo and Busia districts in eastern Uganda. To be eligible for ART in the project, people need to have CD4 counts at or below 250 cells/mm3 or WHO stage III/IV disease (excluding pulmonary TB).

The data used for this analysis were taken from hospitalisation records and outpatient clinic visits. Field workers identified deaths that occurred outside of the hospital during home visits. Viral loads, CD4 cell counts and complete blood counts were measured quarterly in all participants.

The investigators also used the Center for Epidemiologic Studies Depression (CES-D) Scale to measure depressive symptoms at baseline and regularly thereafter.

The study enrolled 1,120 subjects who have initiated ART (73% women). The median CD4 cell count at initiation of ART was 127 cells; median hemoglobin was 11.3 g/dL; and median body mass index was 19.7 kg/m2. Approximately 39% had WHO stage III or IV disease (and 8% had WHO stage IV disease) at baseline. Follow-up is now out to a median of two years.

Timing of and conditions associated with death

Kaplan Meier curves showed that most of the deaths occurred within the first three months of starting therapy, and then fell substantially again after about ten months on treatment. Over time, the decreasing death rate in patients on ART becomes profound.

“Mortality decreased substantially from 16.4 deaths per 100 person years of observation in the first three months of ART to 1.3 per 100 patient years by months 18-24,” said Dr. Moore.

Clinical conditions that were present at the time of death, or diagnosed within one month of death in the first three-month period were then compared to those associated with all the deaths that occurred thereafter. More than one diagnosis could be associated with each death and no clinical diagnosis was identified for 36% of the deaths within the first three months, and 46% of those thereafter.

The only clinical condition or opportunistic infection that was significantly more likely to be associated with death in the first three months was wasting disease (9% of the early deaths, but none of the later deaths).

Tuberculosis was by far the most common infection associated with death (21% of deaths), however there was a trend towards TB being a more common cause of later deaths (27%) rather than early (11%) but this trend did not reach statistical significance. This would tend to suggest that TB-related immune reconstitution disease (which tends to occur soon after ART initiation) was not a major cause of death.

“We haven’t really looked at immune reconstitution per se, but we didn’t actually find that developing TB within the first three months of ART gave you a higher hazard of death,” said Dr Moore. “It was more or less the same risk as what you have later on… Anecdotally we didn’t see very much.”

The trends were reversed for deaths from cryptococcal disease (18% early vs. 7% late) and Pneumocystis jiroveci pneumonia (PCP) (11% early vs. 5% late) but again were not statistically significant.

Clinical implications

Many of the conditions or factors strongly associated with mortality while receiving ART can be remedied or prevented.

“Screening for and or treating anaemia and malaria prior to ART initiation may improve clinical outcomes, and the consistent use of cotrimoxazole prophylaxis could improve pre-ART haemoglobin levels in malaria endemic area,” said Dr Moore.

Likewise, food and nutritional supplementation might mitigate the effects of low body mass index. Beginning isoniazid prophylaxis before or perhaps when going on ART could also reduce the risk of deaths associated with tuberculosis.

Providing more counselling support or treatments to manage depression could also have a survival benefit.

Factors associated with death

In a Cox proportional hazards analysis starting therapy with a CD4 cell count below 50 cells/mm3 was associated with a higher risk of mortality (HR 1.006, 95% confidence interval (CI) 1.003 to 1.008). Other factors associated with an increased hazard of mortality included haemoglobin

Baseline viral load, a past history of TB treatment or being on treatment at the initiating of therapy, and the CES-D (depression) index score were also significantly associated with mortality.

“We believe this is the first study to highlight the importance of mental health issues, in terms with their association with mortality on ART,” said Dr Moore.

Study limitations

However, Dr Moore acknowledged that the high number of deaths without a clinical diagnosis — which may partly have been due to the limited diagnostic tools available to healthcare staff in this setting, could limit this studies ability to detect differences.

With this in mind, the trend towards a higher number of early deaths for the second most common cause of death in the study, cryptococcal disease, should perhaps be examined more carefully.

Another Ugandan study

An ongoing randomised double-blind, placebo-controlled study of fluconazole (200mg three times a week) as primary prophylaxis of cryptococcal disease in southwest Uganda could produce an answer to this question.

A poster on mortality data from the first three months before and after beginning ART from 805 people in this study was also presented at CROI. Because of the fluconazole prophylaxis and the blinded nature of the study, these data would tend to underestimate the role of cryptococcal disease.

Interestingly, however, the causes of death in this study before beginning ART were more or less the same after ART, (though the rate of death is reduced, especially, again, after the first three months on treatment) — with one exception.

Deaths associated with central nervous system disorders became much more common during the first three months on ART than the three months before: two cases before, for a case rate of 0.98 per 100 person years, vs. nine cases after, for a case rate of 5.47 per 100 person years (p=0.013). The exact cause of these syndromes, for the moment, is currently being investigated.

However, it should be noted that immune reconstitution inflammatory syndrome related to cryptococcal disease, or possibly tuberculosis, can occur during this period — and cryptococcal meningitis-IRIS can be very difficult to diagnose and rapidly fatal.

Once again, there is an intervention that could have an impact on this — which about half of the patients in this study are receiving. Whether or not a brief course of fluconazole is enough to produce a survival benefit in this population will be seen once the study is unblinded.


Moore D et al. Determinants of mortality among HIV-infected individuals receiving home-based ART in rural Uganda. Fourteenth Conference on Retroviruses and Opportunistic Infections, Los Angeles, abstract 34, February 2007.

Parkes R et al. Early morbidity and mortality post-ART in a rural cohort of HIV-infected Ugandan adults. Fourteenth Conference on Retroviruses and Opportunistic Infections, Los Angeles, abstract 542, February 2007.