Lots of discussion today on the future of the coronavirus outbreak (2019_nCoV) on social media. The R_0 estimates reported yesterday, see, raised interesting comments. Some, apparently, thought that the end was near, where others criticized modelling if based on few cases only. Terms as “irresponsible” science were posted. The opposite is true.

Predicting the faith of a new outbreak is essential for a proper public health response. Establishing an opinion (and response) on just looking at the data as they come in, is as good as hand waving. Mathematical modellers establish their estimates on a mechanistic model, reflecting key characteristics of the disease, and on observed data and explicitly formulated assumptions where data are lacking. Everything is (or should be) transparent and can be reproduced by others. The mathematical parts may not be understandable to all physicians (including me), but so are the technological parts of an MRI scan or deep-sequencing (that we blindly trust). If you don’t trust mathematics ask an expert for help. The other important aspect of modelling is that the uncertainties are quantified, and that the robustness of estimates of, for instance R_0, can be tested with sensitivity analyses for parameters where uncertainty is large. Modellers (usually) are fully aware of these uncertainties and extensively discuss them. If ignorant readers communicate the results without the nuances, shoot the messenger, not the modeller!

So, back to the model estimates of today. There was an update by the MRC Centre for Global Infectious Disease Analysis from Imperial College of London (@MRC_outbreak), a team few would not consider to be highly authoratative. Their R_0 estimate of the day was 2.6 (95% CI 1.5-3.5). That may seem less dramatic than 3.6 (the catch of yesterday, by others), but would still require that infection control measures must block well >60% of transmission to be successful.

So far, the case-fatality rate seems low, and many subjects have mild symptoms. That sounds good, but could also be bad news if the latter are capable to transmit the virus. To me (and the modellers, I hope), that is the big unknown yet. The family history published in Lancet yesterday is not reassuring. One of six family members had no disease symptoms at all, but appeared to have similar pulmonary CT abnormalities as the others when tested on requested of the– worried – family members. The presence of 2019_nCoV in respiratory samples was not tested in this asymptomatic subject. Therefore, no proof of transmissibility in the absence of symptoms, but negative tests would have been more reassuring.

Interesting times, to be continued.

Some additonal information, as new data and interpretations are emerging as rapidly as (or even faster than) the virus. In case of an outbreak, one of the most wanted  numbers is the R_0, defined as the average number of secondary cases resulting from an infected subject surrounded by susceptibles only, and in the absence of infection control measures. If R_0 is <1, you most probably won’t hear of it, as the disease dies out. If R0>1 there is a chance that an outbreak becomes big, as it will grow as long as there are sufficient susceptible subjects around. The goal of infection control is to bring down an R_0 bigger than 1 to an effective number <1, and keep it there.

As it is a new virus, the whole global population is susceptible. I trust many research groups around the globe are searching internet for epi data to fit their models and to estimate R_0. The first report I saw appeared today and comes from the UK. Published on biomedRXiv, so it did not yet undergo peer review. As it is Friday night, I just quote some statements from the abstract. Data were used until “21 January to estimate key epidemiological measures, and to predict the possible course of the epidemic, as the potential impact of travel restrictions into and from Wuhan.” And their estimate is:

R_0 = 3.8 (95% CI ,3.6-4.0)

If so, indeed the globe is at igh risk for a pandemic. It also indicates “that 72-75% of transmissions must be prevented by control measures for infections to stop increasing.”, i.e. to bring down R_0 to <1.

They also “estimate that only 5.1% (95%CI, 4.8-5.5) of infections in Wuhan are identified, and by 21 January a total of 11,341 people (prediction interval, 9,217-14,245) had been infected in Wuhan since the start of the year.”

Naturally, all predictions are difficult, many uncertainties remain and “findings are critically dependent on the assumptions underpinning our model, and the timing and reporting of confirmed cases, and there is considerable uncertainty associated with the outbreak at this early stage.”

So, what will determine, if this R_0 turns out to be correct, if it can be controlled. A vaccin will do, but will probably be too late. Till then isolation infectious subjects will be key. But whom to isolate? The good news of SARS was that transmission only occurred after onset of symptoms. So immediate isolation at symptom onset probably worked. If the opposite occurs (transmission before symptoms, as in influenza) it might be unstoppable. Eagerly awaiting data on this aspect.

The dramatic measures taken by the Chinese government today and the 1000-bed hospital to build in 10 days, makes me think that we may not know everything yet and, thus, that these estimates might not be that unrealistic. Interesting times.