COVID-19 : Is the virus evolving into a more virulent form?

In an ongoing pandemic, a key question that arises is the virulence of the virus as it mutates over time and spreads to a wider or new geographic regions. This is because virulence factors determine whether infection occurs in an exposed (potential) host and how severe the resulting viral disease symptoms are.

A recent article in USA TODAY (March 28) concluded that there are 8 strains of the virus and that no one strain of the virus is more deadly than another. This conclusion is based on observing tiny differences between the genomic sequence of the virus strains. It would be interesting to see if currently available data on confirmed cases, deaths and recovered supports the claim of equal virulence.

The ongoing pandemic can be separated into 3 waves: the initial outbreak in Wuhan, China (Wave 1), propagation to Italy/Iran (Wave 2) and onwards to USA (Wave 3). The infections in the USA may be attributed to a mix of Wave 1 and Wave 2 strains which may have evolved further. Since the Wuhan and majority of US infections are well apart in time, the available time for mutation is longer and the viral strains may likely have greater separation. It therefore makes sense to contrast the empirical data for these 2 regions in order to compare the virulence of the strains. The ideal approach would entail contrasting the Wuhan city data with NYC data – but in the absence/unavailability of such information, the below summaries are based on overall China vs US data.

The results are presented at various milestones denoted by # of total cases for each region. An attempt is made to ensure approximately equal number of cases at each data cut and the summary measure of interest is the ratio of number of patients that recovered to the number that died. This ratio is more meaningful in an ongoing epidemic, whereas, the final # deaths divided by the total # infected makes more sense after the epidemic ends.

The data are sourced from worldometers.info and some numbers are imputed from the graphs. There may be slight discrepancies in the actual counts, this should not impact the computed trends and conclusions.

Several points to conclude from the last column.

• The US “recovered to dead” ratios are much smaller compared to the China ratios for approximately the same number of infections.
• The China ratio for March 1 may reflect the “end of epidemic” ratio and may also be driven by the increased testing (due to increased resources) of asymptomatic/mild patients who are more likely to recover and contribute to the numerator of the ratio.
• The rate of infections (or uptake of the virus) is much faster in the US - more than twice the rate in China. This may contribute to the worse ratios for the US since deaths can occur at any point in time but “recoveries” may need a minimum of 21+ days to be confirmed (i.e. retested). This is a limitation of the ratio in a rapid rate of infection setting.
• It is interesting that the absolute # of US deaths is lower than China for the first 4 data cuts even though the ratios are much worse. What is concerning is that patients in the US are not exiting the system (i.e. not recovering) and may end up in the death column as well as straining the healthcare system. This could possibly exacerbate the situation.

In conclusion, one would expect improved and better standard of care at a later stage in the pandemic leading to much improved outcomes. However, empirical data seem to suggest otherwise and a possible culprit may well be a more virulent strain. The data so far do not suggest that the virus strains are similar in virulence.

Limitations of Analysis: The above Table could be refined by stratifying the data for "high-risk" groups, e.g., Age > 70, Gender (earlier public data show Males ~2 times more likely to have fatal outcome than Females), 1 or more Underlying condition with/without Age category. This approach would allow a better comparison of ratios and guard against the fact that the US, in the early stages of the outbreak, is capturing data on more severe patients than China (which is likely given limited testing resources). Cumulative data on number of test kits used or number of patients tested would be very helpful (I have not been able to locate it). And finally, the data are taken at "face value", i.e., assumed to be accurately reported.