Summary

Using the same model as the previous post, (a model that takes into account delay in recovery and death), and assuming the infection fatality rate (chance of fatality if infected) is \(0.66^{+0.13}_{-0.13}\%\) [1], I found that the reported case is only \(2.3^{+0.5}_{-0.5}\%\) of the total infection. With the total 790 reported case (as of 25 March 2020), there might be about 34,000 infections currently (which most of them are mild).

NOTE: to avoid unnecessary panic, it is now known that most infection cases are mild and some cases are even asymptomatic.

Assumptions

  • The median of confirmed-to-death time is assumed to be between 9 to 13 days
  • The infection fatality rate is \(0.66^{+0.13}_{-0.13}\%\), based on the estimate from [1]

Introduction

The analysis results from the previous post shows that the case survival rate is about 71% or the case fatality rate (CFR) about 29%. I need to reiterate that this number is defined as the chance of fatality if being confirmed positive in Indonesia. This is different from the infection fatality rate (IFR) which is the chance of fatality if being infected in Indonesia. The high calculated case fatality rate is most probably due to the reported cases being heavily skewed towards severe cases and underreporting the mild case.

Results

Calculating the infection fatality rate accurately is much harder as it requires mass-testing for everyone, including the ones with mild or even no symptoms. This is still a highly challenging task. However, there has been an estimate of the infection fatality rate in China, which is about \(0.66^{+0.13}_{-0.13}\%\) (86% confidence interval). By assuming the infection fatality rate in Indonesia is similar to that number, we can estimate how many unreported infections in Indonesia.

With the infection fatality rate about 0.66%, most of the infected patients will be recovered or even do not develop the symptoms. So we can assume that the unreported infection in Indonesia is mostly dominated by patients who will survive. With this assumption, the reported fraction can be calculated simply as

\[\begin{equation} \mathrm{reported\_fraction} = \frac{\mathrm{estimated\_IFR}}{\mathrm{estimated\_CFR}} \end{equation}\]

By putting \(\mathrm{CFR} = 29^{+3}_{-3}\%\) and the assumed \(\mathrm{IFR} = 0.66^{+0.13}_{-0.13}\), we obtain the reported fraction is only \(2.3^{+0.5}_{-0.5}\%\) of the total infection. This estimate is similar to the results of the study here. With the total 790 reported case (as of 25 March 2020), the number translates to 34,000 number of infections currently.