Large analysis shows benefits of starting treatment with efavirenz

This article is more than 18 years old. Click here for more recent articles on this topic

An analysis of twelve European and North American cohort studies has found that HIV-positive patients starting treatment containing efavirenz (Sustiva) may be less likely to die, progress to AIDS or fail to suppress viral loads than those starting with other anti-HIV drugs. The study’s findings were published in the 1st September edition of The Journal of Infectious Diseases.

Most clinical trials of HIV drugs rely on measuring ‘surrogate’ endpoints to assess the drugs’ effectiveness, such as changes in viral loads and CD4 cell counts. However, few randomised studies have assessed ‘clinical’ endpoints, such as disease progression and death, since this would make the studies too long and would involve putting patients at risk of ill health and death.

In contrast, observational studies, which keep track of a set of patients over time, can detect levels of disease progression and death, but they are often too small to allow different drug regimens to be compared.


observational study

A study design in which patients receive routine clinical care and researchers record the outcome. Observational studies can provide useful information but are considered less reliable than experimental studies such as randomised controlled trials. Some examples of observational studies are cohort studies and case-control studies.

disease progression

The worsening of a disease.

boosting agent

Booster drugs are used to ‘boost’ the effects of protease inhibitors and some other antiretrovirals. Adding a small dose of a booster drug to an antiretroviral makes the liver break down the primary drug more slowly, which means that it stays in the body for longer times or at higher levels. Without the boosting agent, the prescribed dose of the primary drug would be ineffective.

virological suppression

Halting of the function or replication of a virus. In HIV, optimal viral suppression is measured as the reduction of viral load (HIV RNA) to undetectable levels and is the goal of antiretroviral therapy.


Confounding exists if the true association between one factor (Factor A) and an outcome is obscured because there is a second factor (Factor B) which is associated with both Factor A and the outcome. Confounding is often a problem in observational studies when the characteristics of people in one group differ from the characteristics of people in another group. When confounding factors are known they can be measured and controlled for (see ‘multivariable analysis’), but some confounding factors are likely to be unknown or unmeasured. This can lead to biased results. Confounding is not usually a problem in randomised controlled trials. 

To get around this problem, researchers from the Antiretroviral Therapy Cohort Collaboration combined the results of twelve separate observational studies carried out between 1996 and 2003 to compare the outcomes of patients starting different treatment regimens. In total, they gathered data from 17,700 patients, with 55,600 person-years of follow-up.

After comparing outcomes to those of patients starting treatment with efavirenz and two nucleoside reverse transcriptase inhibitors (NRTIs), the investigators found that patients starting with nevirapine (Viramune) were more likely to develop AIDS or to die over the course of the study (adjusted hazard ratio [HR]: 1.28, 95% confidence interval [CI]: 1.03 – 1.60). This was calculated after controlling for age, injecting drug use, stage of HIV disease, CD4 cell count and viral load

Patients starting with full-dose ritonavir (Norvir; adjusted HR: 1.31; 95% CI: 1.01 – 1.71) or ritonavir-boosted protease inhibitors (adjusted HR: 1.45, 95% CI: 1.15 – 1.81) were also more likely to develop AIDS or die. The boosted protease inhibitors included were amprenavir (Agenerase), lopinavir (Kaletra), saquinavir (Invirase / Fortovase) and indinavir (Crixivan): newer drugs could not be analysed as they have not been available long enough for long-term data to be gathered.

In contrast, patients starting with nelfinavir (Viracept), unboosted saquinavir, unboosted indinavir or abacavir (Ziagen) had a similar risk of AIDS or death to the patients starting treatment with efavirenz.

The investigators also examined death rates from any cause, finding that patients starting treatment with nevirapine were more likely to die than those starting with efavirenz (adjusted HR: 1.65, 95% CI: 1.16 – 2.36). All other drugs led to a similar risk of death to efavirenz.

Further analysis revealed that this was due to an increased risk of death in the first six months of treatment with nevirapine. This risk was elevated in women, in patients starting treatment with higher CD4 cell counts and in those with hepatitis C co-infection. However, it was not related to a poorer response to treatment in the patients taking nevirapine. This is consistent with the known risks for nevirapine-associated liver toxicity.

Patients starting with treatment containing efavirenz were more likely to have suppressed HIV viral loads to below 50 copies/ml than all other drugs by six months of treatment. They were also more likely to have remained on their first-line regimen for at least six months than all other drugs except for nelfinavir and abacavir.

“The choice of drugs used in the initial regimen was associated with the probability of viral suppression and with the sustained use of this regimen six months after starting highly active antiretroviral therapy (HAART),” conclude the researchers. “We observed few differences in the rates of mortality between regimens, and these rates were much lower than were those in the pre-HAART era.

“Efavirenz performed the best of the … drugs, but several other drugs had similar effectiveness,” they add. “Compared with efavirenz, the highest mortality risk was for patients who initiated HAART by receiving nevirapine.”

None of the patients included in the study had taken anti-HIV drugs before, so these findings were not complicated by prior exposure to antiretrovirals. However, the investigators’ analysis was limited by the observational nature of the component studies: since they were not randomised, some of the differences in outcomes between different drugs could be due to the patient’s disease stage or other characteristics before they started anti-HIV treatment.

For example, the differences observed between efavirenz and nevirapine do not match those of a recent head-to-head study comparing the two drugs, which found similar anti-HIV effects of the two drugs. “These differences could be attributed to the superiority of efavirenz or might reflect underlying unmeasured differences in the ways in which physicians prescribe and patients use efavirenz, compared with other … drugs,” the investigators explain.

“Large clinical trials that are powered to assess differences in clinical outcomes are required to produce a more definitive answer, and we strongly urge funding bodies to support such large, long-term trials in the future,” they write.

As discussed in an accompanying editorial written by Michael Hughes of Harvard University, the initial virological responses seen in randomised trials in the short term do not necessarily predict the longer-term impact of drugs on disease progression and death in the real world. Dr Hughes calls for all randomised trials to continue to follow up their participants after the end of the study, to try to gather more information on the long-term impact of initial choice of treatment.

“For randomised trials that use surrogate end points to compare treatments, it is important to evaluate effects over much longer that 48 weeks and to follow all randomised patients so that outcomes in patients who experience failure with the initial randomised regimen are evaluated,” he writes.

“A policy requiring follow-up of all randomised patients for long-term vital status would facilitate cross-trial analyses that might help to address important questions about treatment management and, by building on the randomisation, to reduce the potential for confounding by unmeasured factors that is inherent in cohort studies,” he adds.


The Antiretroviral Therapy Cohort Collaboration. Rates of disease progression according to initial highly active antiretroviral therapy regimen: a collaborative analysis of 12 prospective cohort studies. J Infect Dis 194: 612-622, 2006.

Hughes MD et al. Initial treatment of HIV infection: randomized trials with clinical end points are still needed. J Infect Dis 194: 542-544, 2006.