HIV Treatment Bulletin

Efavirenz associated with suicide risk in analysis from START study

AIDS 2016 combined logo

Simon Collins, HIV i-Base

Suicide-related side effects are an important concern with efavirenz. Given how widely efavirenz continues to be used the main safety concern is whether current prescribing practice is sufficiently accurate to minimise these risks.

Earlier studies have reported that the risk is significantly higher than with other ARVs but also that doctors are careful to use alternative options for people who have a history of anxiety, depression or other psychiatric symptoms. [1, 2]

An oral presentation at AIDS 2016 provided information on both actual risk and prescriber awareness from a new analysis from the large international START study. [3]

START randomised more than 4600 treatment naive individuals with CD4 counts >500 cells/mm3 to either immediate ART or deferred ART (until the CD4 counts dropped to 350 cells/mm3). Importantly, before randomisation doctors were asked to pre-specify the choice of ART for all participants, irrespective of which group they would later join.

This design enabled the START researchers to identify appropriate control groups for participants who were either judged to be at risk or not at risk from suicide-related side effects in a way that could highlight the impact of efavirenz and also the underlying risk independent of ART.

START enrolled a largely young healthy group in early HIV infection. Baseline demographics have already been reported. [4, 5] Additionally, 270/4685 (5.8%) participants had a prior psychiatric diagnoses.

Efavirenz was pre-specified for 3516 participants (75%). This was less often in those with psychiatric diagnosis (40%) than without (77%). Although characteristics were similar between participants for whom efavirenz was pre-specified or not, prior psychiatric disease was less frequent (3.1% vs 13.9%) in the those pre-specified to use efavirenz, as was current use of psychiatric treatment (4.0% vs 15.1%). Also significant, was the higher used of efavirenz in low- and middle-income vs high-income countries (65% vs 35%).

In the study overall, there was no difference in suicidal behaviour between the early vs deferred arms (27 vs 24 events; HR 1.15 (0.66 to 1.99), p=0.63). Of the 52 events, 30 were suicide attempts and 16 events where people wanted to commit suicide. There were single cases of self harm and self harm ideation. The three people who died from suicide were all in the deferred arm and this occurred after starting treatment.

The majority of events occurred in those with a previous psychiatric history. The rate was 10-fold higher in people in whom efavirenz was prespecified and 3-fold higher in those pre-specified to use other ART, in people with a previous diagnosis (see Table 1). This, together with the higher rate of events in people pre-specified to use other ART (overall rates 1.28 vs 0.63 per 100 PY) show a generally high awareness not to prescribe efavirenz in people at highest risk.

Table 1: Suicide-related events by prespecified efavirenz use and psychiatric history
Psychiatric history No psychiatric history
EFV pre-specified 72.0 /100PY 220.2 /100PY
Other ART 81.7 /100PY 140.5 /100PY

A subgroup analysis that censored participants in the deferred group at the start of ART was then able to look more specifically at the role of efavirenz in people not thought to be at risk (ie who had been assigned to use efavirenz) and also at the roll of psychiatric-related events in people not exposed to efavirenz.

In this analysis there were 17 vs 3 events in the immediate vas deferred group assigned to receive efavirenz (HR 4.16; 95%CI 1.2 to 14.4), p=0.02) showing that in people without a psychiatric history, efavirenz was significantly related to a risk of suidcide-related events.

Similarly, in the people whose medical history might be associated with suicide-related events irrespective of ART, there were 9 vs 8 events (HR 1.04; p5%CI 0.4 to 2.7, p=0.93) for non-efavirenz versus their ART-naïve controls – indicating that other ART had no impact on people judged to be at high risk due to past history.

Both these analyses were protected by randomisation and therefore likely to provide high quality evidence. Also importantly, the interaction between these two groups was also statistically signficant (p=0.05 for difference in HRs), see Table 2.

In multivariate analysis, the factors that were significantly associated with risk of suicide-related events were: previous psychiatric diagnosis (HR 12.8; 95%CI 4.7 to 34.9, p<0.001), heavy alcohol use (HR 6.1; 95%CI 1.9 to 19.6, p = 0.003) and ever having used recreational drugs (HR 2.9; 95%CI 1.0 to 7.9, p=0.04).

Table 2: Suicidal and harming events by randomisation group
n Immediate ART Deferred ART HR 95%CI p P
Events Rate Events Rate
Intention to treat (ITT) analysis
EFV pre-specified 3516 18 0.34 11 0.21 1.42 0.6 to 1.9 0.37 0.23
EFV not pre-specified 1169 9 0.53 13 0.72 0.74 0.3 to 1.8 0.5
Censoring deferred arm participants at ART initiation
EFV pre-specified * 3516 17 0.35 3 0.08 4.16 1.2 to 14.4 0.02 0.05
EFV not pre-specified ** 1137 9 0.59 8 0.69 1.04 0.4 to 2.7 0.93

* 6/17 vs 0/3 were in people with psychiatric history.
** 5/9 vs 2/8 were in people with psychiatric history.

Simon Collins is a community representative on the START group working on this analysis.


The good news from START is that even with widespread use of efavirenz, especially in people who either started immediate treatment there were few serious reports of suicide-related behaviour and only three suicides.

Even with few events – a good thing – efavirenz was still associated with a significantly higher risk of suicide related complications, especially in those with a history of depression or other psychiatric conditions.

These results support the continue screening of patients for such history before starting efavirenz.


  1. Mollan et al. Association between efavirenz as initial therapy for HIV-1 infection and increased risk for suicidal ideation or attempted or completed suicide: an analysis of trial data. Ann Intern Med 2014.161(1):1-10. doi:10.7326/M14-0293.
  2. Smith C et al. Lack of association between use of efavirenz and death from suicide: the D:A:D Study. HIV Drug Therapy Glasgow Congress, 2-6 November 2014. Oral abstract O315. Journal of the International AIDS Society 2014, 17(Suppl 3):19512. (Abstract) (PDF)
  3. Arenas-Pinto A et al. Increased risk of suicidal behaviour with use of efavirenz: results from the START trial. AIDS2016. Durban, South Africa. 18-22 July 2016. Oral abstract THAB0202. (Abstract) (Webcast)
  4. Annual DSMB open reports 2009 – 2015.
  5. HIV Medicine supplement. The START Trial Characteristics at Study Entry. HIV Medicine Special Issue: April 2015; 16(S1):1-146.