Systematic reviews using evidence synthesis methods

Impact story 9 Jan 2019
118

Until the early 1990s, evidence synthesis was not part of medical currency and certainly had not touched tropical medicine. LSTM Professor Paul Garner, along with colleagues Hellen Gelband who was with the Health Technology Assessment Office in Washington and Piero Olliaro from the TDR, part of the World Health Organization (WHO), set up and registered the Cochrane Infectious Diseases Group (CIDG) in 1994. This group committed to preparing systematic reviews using Cochrane evidence synthesis methods in health problems of the tropics.

Early reviews, drawing on unpublished studies and francophone literature, broke new ground. One review showed that amodiaquine resulted in higher cure rates compared to chloroquine for treating malaria. A meta-analysis of ‘low osmolarity’ oral rehydration salt (ORS) solution led to a global change in the formula of ORS worldwide. Suddenly, CIDG had demonstrated the power of these methods in disease global policies.

CIDG’s co-ordinating editor, Professor Paul Garner, saw that there were several cluster-randomised trials of insecticide treated mosquito nets for preventing death in children being conducted, facilitated by the WHO. He persuaded Christian Lengeler from the Swiss Tropical Institute to take the Cochrane approach and synthesize the evidence across all the trials in a single Cochrane review. At the time, there were no statistical methods for dealing with the design effects in cluster-randomized trials, but the CIDG engaged specialist statisticians in health services research in Canada who had started on these methods. The review was published in 1998 and updated in 2004 once the methods of ‘reverse variance’ had been developed so confidence intervals around the meta-analysis were possible (now termed the ‘inverse variance’ method). The results were extraordinarily important, demonstrating a clear impact on mortality: 5.5 lives were saved every year per 1000 children protected. The confidence intervals and the certainty around the estimates were pivotal in underpinning the investment in insecticide-treated nets (ITNs) in the following decade, and the review has been cited 1767 times.

As always with Cochrane, the editorial team advancing the methods and helping people apply them stayed in the background. Cochrane has never been about centre-staging the staff and academics that help get the reviews done - it looks for modesty and generosity in its partners; to help a research field summarize research in a scientifically defensible way, free of conflicts of interest and the spin that may happen when specialists review their own research. Whilst the Cochrane team have gone on to advance synthesis and support (or sometimes challenge) policies in global health, vector control stayed out of the mainstream evidence synthesis flow, mainly as the specialists thought it was not relevant to them. The CIDG team in Liverpool however continued their work, collaborating with LSTM Director, Professor Janet Hemingway, on reviews of sonic repellents; and another larger review examining the impact of entomological resistance on mortality and behaviour of mosquitoes.

However, over the last few years the tide has begun to turn. The WHO Global Malaria Programme (WHO GMP), which has worked with the CIDG for 15 years, has recently teamed up with the Entomology & Vector Control Unit (EVCU) at the WHO GMP to help ensure vector control guidelines are more in line with current evidence-informed scientific approaches. The CIDG were asked to assist, given their recognized excellence in the methods of evidence synthesis and knowledge in malaria. A whole suite of reviews was prepared using Cochrane methods on topics including: indoor residual spraying added to ITNs; insecticide space spraying; piperonyl butoxide (PBO) nets’ effectiveness in the presence of insecticide resistance; and an update of existing Cochrane reviews on larviciding and the use of larvivorous fish. In addition, the original 2004 ITN review was updated using advanced GRADE methods for assessing the certainty of the evidence. The young research team, Joseph Pryce and Leslie Choi, along with CIDG statistician Marty Richardson, used innovative methods to estimate confidence intervals for the space spraying review. This is the first time this has been achieved with a disease that showed seasonality in interrupted time series analysis designs.

“These reviews support WHO’s strategic decision in vector control to move away from expert opinion to evidence-based recommendations”, said Jan Kolaczinski, EVCU Coordinator. The initial suite of seven Cochrane reviews underpin the first edition of the WHO malaria vector control guidelines to be published in 2019.