Adults in resource-limited settings are most likely to experience an HIV-associated Illness in the first three months from initiating therapy

Polly Clayden, HIV i-Base

Programmes in resource-limited settings have reported relatively high mortality rates following initiation of therapy, particularly in the first few months.

In an oral presentation, Paula Braitstein from the ART-LINC Collaboration presented data from a study to evaluate HIV illnesses during the first year of receiving antiretroviral therapy.

ART-LINC is a multinational network of HIV treatment programmes in Africa, Brazil, and Asia. ART-LINC is modelled on ART-CC, which is a pooled network of databases from Western Europe and North America established in 2001. Findings from ART-LINC are compared to those from ART-CC.

The first data merger took place in December 2004 and included 23 centres across 16 countries and included data from 8734 patients receiving ART. A second merger is underway and now includes over 40 centres, 20 countries and over 31,000 people on ART.

An earlier finding from ART-LINC was that patients in this cohort had a 4 times greater mortality risk than that observed in ART-CC in the first month after initiating therapy. Causes of death are currently unknown.

Eligible for analysis were all previously treatment-naive men and women above 16 years who initiated >/=3 ART drugs, with known baseline CD4. Illnesses included all HIV-related and AIDS defining infections including TB. To avoid counting prevalent illnesses, the investigators defined events as “newly diagnosed and/or recorded” if at least 90 days had past since the last episode (or 180 days in the case of TB).

Follow up time was censored at time of event, death, last follow up visit or 12 months since initiation of therapy. The investigators evaluated the following risk factors for the occurrence of HIV-associated illness and TB: age, sex, baseline CD4, initial treatment regimen (non-nucleoside reverse transcriptase inhibitor [NNRTI]-based vs protease inhibitor [PI]-based vs other =3 drugs), and having a HIV-associated illness or TB at or pre-HAART. Incidence rates were calculated per 1000 patient years.

In the first year 4540 people were included in the analysis, with 3255 person years of follow up, in this period 689 HIV-associated events were recorded; 212 per 1000 person years. This was compared to ART-CC data with 12,574 people, 8709 person years of follow up and 710 events; giving a much lower incidence rate of 82 per 1000 person years.

Differences were observed across the two cohorts in most frequently reported HIV associated events. In ART-LINC the most frequent were: tuberculosis, Kaposi’s Sarcoma, candidiasis, bacterial pneumonia, cryptococcus, toxoplasmosis, wasting syndrome and mycobacterium avium disease. In ART–CC, mycobacterium avium disease, Kaposi’s Sarcoma, cytomegalovirus, PCP, tuberculosis, oesophageal candidiasis, encephalopathy and toxoplasmosis were the most frequently reported.

Dr Braitstein cautioned that diagnostic techniques vary significantly across regions and settings, for example one clinic may offer systematic screening for TB at 1st visit while another may only investigate on presentation of symptoms. She emphasised that the focus in this analysis is on relative changes in rates of HIV-associated events.

The investigators found: age <30 years (AHR 1.00, p=0.081), female sex (ARH 0.76 [95% CI; 0.85-0.90], p=0.001) and baseline CD4 (AHR: 1.0, 0.75, 0.85. 0.61, 0.65 for <25, 25-49, 50-99, 100-199 and 200-350 cells/mm3 respectively, p=0.001), were associated with HIV-associated events in the first year from initiation of therapy in the ART-LINC cohort.

Additionally: initial triple combination, NNRTI (AHR 1.00), PI (AHR 1.36 [95% CI: 0.97-1.91], other triple combinations (AHR 2.00 [95% CI: 1.45-2.77) p<0.001; history of OIs (vs none), pre-HAART (AHR 2.93 [95% CI: 2.07-4.16]), at HAART (AHR 2.59 [95% CI: 2.13-3.16]), p<0.001 and calendar time (AHR 1.16 [95% CI: 1.08-1.26]), p<0.001, were associated.

Up to 3 months from initiation of therapy, age and baseline CD4 were associated but female sex was not (AHR 0.88 (0.71-1.10), p=0.26.

Decline in the relative risk of an HIV associated event was steady over time in both cohorts and across infections (except that the investigators found an unexplained increase in the incidence of bacterial pneumonia at 4-6 moths in the ART-CC cohort).

Dr Braitstein noted that the study had limitations including use of diagnostic criteria and methods, distinguishing between occult and de novo infections, and that association with age, sex and calendar time could at least partially be explained by survivor bias.

The investigators concluded that HIV associated illnesses are common in resource limited settings in the first year from initiation of HAART (twice as high in ART-LINC as in high income settings). Consistent with high income settings they found the highest risk to be in the first months. Although patterns of HIV associated illnesses differ, relative declines are similar in resource limited and high income settings. They wrote: “Frequent monitoring and the use of prophylaxis may improve outcomes in the first few months following HAART initiation.”


Braitstein P, Brinkhof M, Schechter M et al. When are adults in resource-constrained settings most likely to experience an HIV-associated illness following HAART initiation and what is it related to? 13th CROI, Denver, 2006. Abstract 67.

Links to other websites are current at date of posting but not maintained.