The epidemiology of MERS-CoV.

“The oldest and strongest emotion of mankind is fear, and the oldest and strongest kind of fear is fear of the unknown”

Howard Phillips Lovecraft

Coping with, and trying to understand, emerging epidemics has been part of the human experience since time immemorial. The concern that attends the emergence of a novel infectious disease might be substantial, but fears often subside as the disease becomes better understood, partly because initial findings tend to be biased towards cases and case clusters with more severe outcomes.1, 2, 3 Epidemiologists and microbiologists are currently working to better characterise the outbreak of Middle Eastern respiratory syndrome (MERS) coronavirus infection that made its initial appearance in a health-care-based outbreak in Jordan in 2012. The understanding of the natural history and epidemiology of an emerging infectious disease allows us to predict its behaviour and identify control strategies. In The Lancet Infectious Diseases, the study by Simon Cauchemez and colleagues is timely and important, since it provides us with such information about the emerging MERS outbreak.

We were impressed by the extent to which Cauchemez and co-authors did their analyses using limited publicly available data. Whereas decision makers often ask infectious disease modellers to provide an epidemiological crystal ball that shows what will happen, these investigators use modelling more appropriately, as a method for the quantification and management of uncertainty. What does their analysis tell us?

A key epidemiological parameter early in an emerging epidemic is the basic reproductive number (R 0), which describes the average number of new cases of infection generated by one primary case in a susceptible population. R 0 affects the growth rate of an outbreak and the total number of people infected by the end of the outbreak. When R 0 is lower than 1, a sustained epidemic will not occur. Using epidemiological and genetic data, the authors estimated that R 0 was small (consistent with earlier estimates),7, 8 but might be slightly greater than 1. However, R 0 estimates based on disease cluster sizes were lower than 1, suggesting that cluster identification leads to application of successful control measures. Using data on MERS-infected travellers returning from the Middle East, the investigators estimated MERS incidence, and back-calculated the likely extent of case underreporting; they estimated that most cases have been undetected.

These findings suggest that mildly symptomatic cases are common, with implications for the effectiveness of infection control measures. They also show us that the MERS outbreak represents a dynamic process, with human incidence possibly portraying transmission from animal to man occurring in the context of an as yet unconfirmed epizootic.

The distance between the raw, publicly available, epidemiological and virological data obtained by the authors, and the insights provided by them in their report can represent the distance between abundant information and actionable knowledge. The quality of data available to these authors is poor, but should that have prevented them from proceeding with analysis until better data were available? We do not think so: inferences based on the best available data, even if those data are imperfect, allow decision makers to follow optimum courses of action based on what is known at a given point in time. Could the estimates presented here be refined if more complete data were to become available? Undoubtedly, and it seems likely that these investigators and others will continue to do so as this disease is further studied.

The widespread, transparent, and trans-jurisdictional sharing of epidemiological data in the context of public health emergencies could improve the accuracy of early efforts at epidemiological synthesis, such as those presented here by Cauchemez and colleagues. However, many disincentives for such sharing by jurisdictions experiencing outbreaks remain, including lost travel and tourism, concerns about data security, and interest in scientific publication. Creative epidemiologists and computer scientists have shown various means to infer levels of disease activity with indirect Internet data mining techniques.9, 10, 11 The ability to draw inferences about diseases from non-traditional data sources will hopefully both provide alternate means of characterising epidemics, and diminish the temptation towards non-transparency in traditional public health authorities. The resultant improvements in rapid epidemiological risk estimates would benefit all of us.