The Group banner image
Home | About RA | The Board | Contact Us
First Previous

Rebecca Culshaw and David Crowe Challenge Julio Montaner (but Lancet won't print it)


August 22 Letter to Lancet

To Lancet:

A recent article in Lancet by Julio Montaner and colleagues [1] may lead to the unfounded perception that a mathematical model has shown that prescribing AIDS drugs to healthy, HIV-positive people will result in a massive reduction in the worldwide prevalence of HIV.

The biggest problem with this model is that scientifically, it does not exist. The algorithm, variables and parameters are not described in this paper nor is a citation given to another source for the information. It has thus not been subject to review or replication by other scientists and mathematicians.

Even in the absence of details there are several indications that the model is unrealistic.

The researchers claim, for example, that it is reasonable to assume “a 50% or so yearly reduction in new HIV cases” based on data they cite from Taiwan [2] and British Columbia (unpublished) that shows only a one-time reduction of about 50% in, respectively, 5 years and 3 years.

The weakness of their model is also shown by their use of the term “notion” (which we interpret as “wishful thinking”) to describe the assumption that most new HIV infections are produced by people who are not on HAART.

We initially took this claim of Montaner et al on faith, but when we read the Taiwan study we realized that Montaner had seriously misquoted the data. Montaner says:
These concerns have been alleviated by an ecological study from Taiwan, which provided compelling evidence about the effect of HAART on HIV transmission. The study showed a 53% reduction in new positive HIV tests after the introduction of free access to HAART.
This is not true. What the Taiwan paper showed (via a mathematical model of course) was that “The reduction in the average HIV transmission rate in Taiwan was thus…53%.”, the transmission rate being the probability that an HIV+ person would infect another person.

What about new positive tests? Well, they went up year over year, from about 125 in a four month period of 1997 to about 300 for the last four month period at the end of 2002 (We don't have the exact numbers as the paper just includes a graph).

The graph of new HIV/AIDS cases from 1984 to 2002 shows (via eyeball) a slightly more than linear growth although as the number of “people living with AIDS” increases over time, the calculated rate of transmission (number of new cases (incidence) divided by the number of infected people (prevalence)) would decline. Looking at the graph there is no obvious change around 1997, the upward trend continued uninterrupted.

The Taiwanese reasearcher’s methodology is a bit strange too. They work off the natural logarithm of the “numbers of all newly detected cases every four months” which obviously compresses the upper end of the curve. They do linear regression on the 1990-97 natural logarithms and 1997-2002, resulting in two different slopes, a slope that is lower at later dates with natural logarithms but would have been higher with the raw numbers. And they chose the break point of 1997 (because that’s when free HAART began to be supplied) it didn’t emerge naturally from the data.

One would think that if they were studying transmission they would calculate transmission rates assuming that the number of new cases (which they estimate were detected 2-1/2 years after infection) as a function of the cases existing at the time of infection (which would be the total number of cases, less deaths) although perhaps working off the number of known cases at the time of detection would be a good approximation.

The Montaner paper claims that regional differences in the rate of new infections per 100 currently infected people correlates inversely with regional availability of HAART, but information on the availability of HAART in each region is not provided. It does not seem reasonable to assume that Sub-Saharan Africa has better access to AIDS drugs than the Caribbean, East Asia, North Africa, the Middle East, East Europe and Central Asia, which is what their transmission rate data implies.

The authors indirectly recommend Atripla as a fixed dose combination that is “simple, safe, and well-tolerated” [3]. They may have ignored the black box warning (”Lactic acidosis and severe hepatomegaly with steatosis, including fatal cases, have been reported with the use of nuceloside analogs alone or in combination with other antiretrovirals”). Other side effects listed in the manufacturer’s prescribing information include nervous system symptoms, renal failure, skin rash (including Stevens-Johnson Syndrome), significant liver toxicity, reduced bone density, convulsions, fat redistribution and Immune Reconstitution Syndrome (infections or other conditions that occur after initiation of HAART and that are claimed to be a signed of a repaired immune system). Even a rare side effect, such as Stevens-Johnson Syndrome, at 0.1%, could be a big problem when multiplied by the 38 million target patients estimated by Montaner et al, producing about 38,000 victims. Rates of side effects could be higher in practice if poorer and less well nourished people prove to be more susceptible.

The modellers estimate costs at about $365 per year, but this appears to be just the cost of a combination pill that they cite as $1 per day. This does not include the costs of two forms of contraception recommended by the manufacturer due to the risk of birth defects, the cost of tests for kidney, liver and other disorders, nor the cost of HIV testing everyone in the world on a routine basis to ensure that they are put on AIDS drugs shortly after they become HIV-positive. They also ignore the cost of long-term disability in some people caused by side effects of the drug.

Clearly the recommended combination pill is not simple, safe and well-tolerated. This makes it surprising that the modellers also ignored the effects of non-compliance.

While the Montaner et al article is categorized as “Viewpoint” by Lancet, that subtlety may be lost on people who use it to promote vastly more widespread use of AIDS drugs. With the health of many millions of people at stake it is unfortunate that these authors used the pages of Lancet to indulge in poorly informed and highly optimistic speculation.

Rebecca Culshaw PhD
Assistant Professor
Mathematics Department
University of Texas at Tyler

David Crowe, HBSc
Alberta Reappraising AIDS Society
Calgary, Canada


  1. Montaner JSG et al. The case for expanding access to highly active antiretroviral therapy to curb the growth of the HIV epidemic. Lancet. 2006 Aug 5; 368(9534): 531-6.

  2. Fang CT et al. Decreased HIV transmission after a policy of providing free access to highly active antiretroviral therapy in Taiwan. J Infect Dis. 2004 Sep 1; 190(5): 879-85.

  3. Atripla (efavirenz 600mg / emtricitabine 200 mg / tenofovir disproxil fumarate 300 mg). Gilead. 2006 Mar.

Sept. 1 Rejection by Lancet

Dear Doctor Culshaw,
Thank you for submitting your letter on To The Lancet..   After in-house review, I’m afraid we have decided not to accept it for publication. We regret that we are unable to write a personal note for every letter we turn down, but the following common reasons for rejection may help you with future submissions: lateness (ie, more than 2 weeks after publication of the article on which you are commenting), inclusion of original research (the section is not peer reviewed, so we cannot publish such work here), submission of case reports (we have a separate section for these), reiteration of points made by another correspondent, and inappropriate length (limits are 250 words and 5 references). If none of these apply to your letter, please be assured that we have nevertheless considered it carefully and probably had to refuse it because we have simply received too much good material.
Yours sincerely

Zoë Mullan
Senior Editor
[This is the standard Lancet form letter]

© Copyright May 18, 2009 by Rethinking AIDS.