Daily new infections in Bangladesh estimated to be at least 8 times higher than official numbers

Daily new infections in Bangladesh estimated to be at least 8 times higher than official numbers — with growth in new infections increasing by around 20% every four days.
Jun 10 · 3 min read

June 9th marks the end of the 12th week since the Bangladesh government announced on March 18th its first official death from the Covid-19 virus.

There were a total of 266 deaths in the 12th week — and as one can see from the diagram above, Bangladesh is on a clear upward trajectory.

It is important to note that these deaths involve infections that took place about 20 days ago.

According to Imperial College modelling set out in its “Situation Report for COVID-19: Bangladesh, 2020–06–06”* (most recently published at time of publication)* in the 28 days between May 9th and and June 6th,** it is estimated there would have been a total of 373,707 new Covid-19 infections.*** In the same period, Bangladesh government detected, through its own testing, 47,257 positive cases of infection. Imperial College’s modelling therefore suggests that the actual number of people gaining new infections in this period was around 8 times the number that are formally identified through testing. This, of course, is a minimum as the Imperial College modelling of infections is based around the numbers of officially announced deaths only. If the unofficial Covid-19 deaths were taken into account the number of infections would be estimated to be higher.

Significantly, Imperial College modelling also suggests that the estimated number of infections is increasing day by day. This makes sense since the reproduction number is thought to be above 1 — at 1.53 at present. (To read about the reproduction number see here). The 373,707 estimated new Covid-19 infections is one fifth more than the 314,032 new infections that were estimated to have taken place in the 28 days before the Situation Report published on June 2nd, four days earlier.

Since about 1 in 100 infected people do not recover, of those 373,707 estimated to have been infected in the 28 days between May 9th and June 6th, about about 3,737 people will subsequently die over the 20 or so days from when they were infected — this is on average, over 186 deaths per day in that 20 day period. This explains why the modelling suggests that on 4 July, in 28 days, the model estimates that the number of deaths may be as high as about 425. (A week ago this was estimated to be around 700 — so this is a decrease)

Clearly this is not reflected at present in the official number of deaths — suggesting either that many deaths resulting from these infections are not being officially counted or that there is something wrong with the modelling.

//DB

* Note that this link gets updated every few days with new estimates. The figures given in the article are taken from the Situation Report dated June 6th.

* Note that all dates in Imperial College projections refer to death data given in Bangladesh of the previous day. So June 6th refers to death data announced on June 5th

* The numbers of new infections are estimated through modelling back from the official numbers of deaths. The Situation Report states: “Our current working estimate is that 1 death indicates that approximately 100 people will have been infected with the other 99 recovering (based on an infection fatality ratio of ~1%). These infections will have happened approximately 21 days previously — capturing a 5-day period from infection to onset of symptoms (the incubation period), 4 days from onset of symptoms to hospitalisation, and 12 days in hospital before death. With a 3-day doubling time, 100 infections that occurred 15 days ago will have generated 200 infections 12 days ago, 400 infections 9 days ago, 800 infections 6 days ago and 1,600 infections 3 days ago resulting in approximately 3,200 infections at the time the first death is observed…..We have also included the impact of interventions that have been put in place and their effect on human mobility and transmission based on the Google COVID-19 Community Mobility Reports.”