Vaccines have been an unparalleled public-health success: they have eradicated smallpox and driven polio to near extinction, and routine childhood immunization saves millions of children a year from death from diseases such as measles, diphtheria, tetanus and whooping cough. So it is not surprising that the public tend to view vaccines as synonymous with elimination, or near elimination, of our microbial foes.

The vaccine falls short of the target set by the World Health Organization.

This may help to explain last week’s extensive and often upbeat media coverage of the 18-month results of a huge phase III trial of the malaria vaccine candidate RTS,S/AS01 in more than 15,000 children across 7 African countries. In the United Kingdom, for example, the front page of The Guardian stated that the vaccine “could save lives of millions of children”. Unfortunately, however, it won’t. The 18-month results only confirm the disappointing results seen after 12 months.

The RTS,S vaccine is not what most people would think of as a vaccine. It provides only partial protection and most of those vaccinated, particularly those in areas with moderate to high malaria transmission rates, will eventually contract the disease. There is also confusion over its efficacy. Many media reports concluded that although the vaccine did not give the 90%-plus efficacy levels of most childhood vaccines, it might nonetheless be satisfactory, with a reported 46% reduction in cases in children vaccinated when they were aged 5 to 17 months, and 27% in 6–12-week-old babies.

Not so. The efficacy figures given for RTS,S are not directly comparable with those usually given for vaccines. The conventional measurement of a vaccine’s success is how may people remain protected after a given period, such as 12 months. Because RTS,S is only partially protective, a different measurement of efficacy is used — a complex statistical model that computes hazard ratios on the basis of the first clinical episodes of malaria. As the designers of the method themselves concede, “a shortcoming of the vaccine efficacy calculated from hazard ratios could be that it is not intuitively understood”. Too true. In the hands of experts, and regulatory agencies, this hazards-ratio model offers a valid measurement of the efficacy of a partially protective vaccine, but it can be easily misinterpreted by the media, politicians and policy-makers.

It is not possible for outside scientists to deduce a more conventional efficacy estimate from the 18-month data, as it was described only briefly in press releases from the vaccine’s sponsors, the PATH Malaria Vaccine Initiative (MVI) based in Seattle, Washington, and GlaxoSmithKline (GSK), headquartered in Brentford, UK. (The paper and supporting data are under review at a journal.) But applying a conventional measurement of vaccine success to the published figures for 12-month estimates — for which detailed data are available — reduces the vaccine’s efficacy by more than one-third (see Nature 478, 439–440; 2011). Its protective effect also seems to begin fading after about six months.

Perhaps more promising are the reductions seen in cases of severe malaria, which are reported in the conventional manner. However, although a 36% reduction was reported in children of 5–17 months, the 15% reduction seen in 6–12-week-old babies was not significant — and this age group was the main target of the trial because for logistical reasons it is likely that any malaria vaccine would need to be given alongside routine immunizations at this age.

Many vaccine trial participants had access to other anti-malarial measures — including insecticide-treated bednets and effective drug treatment — so it is possible that the vaccine might offer greater benefit to people more exposed to malaria. Nonetheless, the vaccine falls short of the target for a partially protective malaria vaccine set in 2006 by the World Health Organization, which stated that it should have a “protective efficacy of more than 50% against severe disease and death” that “lasts longer than one year”.

The work will continue. Data on the effects of a booster dose given after 18 months will not be available until next year, and RTS,S is also due to be tested in combination with a vaccine developed by researchers at the University of Oxford, UK, in an early-stage clinical trial. Meanwhile, the RTS,S trials are to be applauded for having left a lasting legacy in the unprecedented collaboration with African scientists who led the study, and a first-class clinical-trials infrastructure on the continent.

RTS,S has been in the works for almost 30 years. Since 2001, the MVI has put some US$200 million into it, and GSK more than $350 million, with a further $260 million earmarked to complete its development. The huge past impact of vaccines risks fuelling illusions over the impact of having a malaria ‘vaccine’. But the modest efficacy of RTS,S means that it falls squarely in competition with other malaria control measures, many of which might be more cost-effective. Care must be taken not to build excessive expectations that can only lead to disappointment over its potentially limited public-health impact.