The London School of Hygiene and Tropical Medicine, with the support of the RSS Medical Statistics Section, hosted the Bradford Hill Memorial Lecture on the 6 June 2018.
Professor Nicholas Jewell spoke on ‘Statistical Challenges in Infectious Disease Research: from HIV to Dengue’. Professor Jewell is currently professor of biostatistics and statistics at the University of California, Berkeley, but will shortly join LSHTM, and this was also his inaugural lecture. His talk was preceded by an introduction by the deputy director of LSHTM, Professor Anne Mills, who spoke about the life and work of Austin (Tony) Bradford Hill, regarded as the greatest medical statistician of the 20th century.
Professor Jewell started his talk by pointing out that infectious disease research needs statisticians, especially given the growing number of threats – he mentioned just a few recent disease outbreaks, including the Plague in Madagascar, the Zika virus, and Ebola. He spoke in detail on some studies in his experience that benefitted from careful thought about appropriate statistical methodologies.
He described his involvement in HIV research in its earliest days, when one of the first challenges was to estimate the incubation period of the virus. The only data available were right-truncated transfusion cases. A Weibull model seemed to fit the data well, but the data were confined to the left-hand tail of the distribution, and extrapolating to estimate the incubation period was inaccurate. It turned out that the Weibull model was not appropriate.
In the context of the MIRA trial of diaphragms in women to reduce the number of HIV infections, Professor Jewell explained an instance of how ethical considerations need to be taken into account in designing a trial. For ethics reasons, all women in the trial were given a supply of condoms, and it turned out that women in the no-diaphragm arm were much more likely to use condoms, which are well known to decrease the risk of HIV transmission. In the intention-to-treat analysis, there was no evidence of a difference in infection rates between the arms of the trial. Methods of causal inference for observational data were used to compare the difference in infection rates in women who did not use condoms at all during the study period; however the number of such women in the non-intervention arm was small and the results inconclusive. Interestingly, The Lancet did not accept this modified analysis in the original paper, but the results were published later.
Professor Jewell also described a current cluster-randomised trial where swarms of mosquitoes, which are infected with bacteria called Wolbachia that supresses replication of the dengue virus, have been released in some areas in Yogyakarta, Indonesia. Outcomes are being collected from clinics in areas with and without the infected mosquitoes. Both test-positive and test-negative cases for dengue are recorded, as not everyone in the population shows up to the clinics.
Professor Neil Pearce, Professor of Epidemiology and Biostatistics at LSHTM, thanked Professor Jewell for his talk, and emphasised the importance of methodology in infectious disease research and other global health research, as bad methods can lead to bad policies.
After the talk, Professor Stuart Pocock said a few words in memory of Doug Altman and his profound influence on the quality of statistics in medical research.