Excess deaths methodology used by Newsminute for Kerala has holes

 The calculation of actual number of deaths due to COVID-19 has been an ongoing struggle for academics around the world.

There are several entirely legitimate reasons why COVID deaths are mis-identified. These are (from an academic paper below):

-some direct deaths attributable to COVID-19 may be assigned to other causes of death due to an absence of widespread testing and low rates of diagnoses at the time of death

- direct deaths from unfamiliar complications of COVID-19 such as coagulopathy, myocarditis, inflammatory processes, and arrhythmias may have caused confusion and led to attributions of death to other causes, especially early in the pandemic and among persons with comorbid conditions

- there are increases in mortality resulting from reductions in access to and use of healthcare services and psychosocial consequences of stay-at-home orders

- Economic hardship, housing insecurity, and food insecurity may cause indirect deaths, especially among those living with chronic illnesses or who face acute heath emergencies and cannot afford medicines or medical supplies

- the pandemic may reduce mortality as a result of reductions in travel and associated motor vehicle mortality, lower air pollution levels, or the possible benefits of COVID-19 mitigation efforts (i.e., mask wearing and physical distancing) on reducing influenza spread

- It is also possible that COVID-19 deaths are over-recorded in some instances, e.g., because some deaths that should have been assigned to influenza were instead assigned to COVID-19.

So, first, COVID deaths being mis-attributed is something that has occured in every nation; and it is not because of a desire to hide numbers from the public.

Now, in order to calculate what is the possible real number of deaths, in UK and in other countries, they have tried to find the difference between two numbers:

1. The number of deaths that is "normal" in a nation/state if COVID had not occured

2. The number of actual deaths

The difference between these two is called "excess deaths".

Not all excess deaths are due to COVID. That is, if you had a projected "normal" death count of 200; and deaths in 2020 is 300 - the difference is 200-100 = 100. But you cannot assume immediately that all of these are due to COVID.

Further excess deaths calculations done at a nation/state level may not reflect the actual variation at the local body level.

In order to find the "normal" deaths that would have occured WITHOUT the pandemic, several methods have been followed. These involve complex statistical projections to simpler methods.

The study mentioned in the first comment does this:

"To compute an average historical death rate for 2013 to 2018, we divided the sum of deaths from 2013 to 2018 by the total population from 2013 to 2018."

This excess deaths based analysis was done by many journalists in India during the April-May-June period. Based on their studies, a number of articles were written in Scroll, Wire, Newsminute etc, IN JUNE ITSELF.

You can find that this may not be correct, because there is a lag between the occurence of a death; and its registration by the local body. This lag is high in Kerala (greater than 21 days for 40% of the deaths WITHOUT the pandemic). But even for other states, there is a lag of weeks between death date and registration; and this must have been higher during the pandemic.

The study below takes a lag of 10 weeks - for deaths in December 2020, they give time until March 12, 2021. Then they used that data to publish the results now.

Therefore, the first mistake committed by these journalists was their rush to calculate deaths while the surge was going on.

The second was that they did not emphasize enough that the excess deaths may not be all attributable to COVID. Instead they sensationalized the numbers in headline.

The study below calculates that only around 20% of the excess deaths could be attributed to COVID-19, as an average, in the US. That is, in the above example, if the excess deaths is 100, then around 20 of them could be COVID. That means that the total deaths due to COVID could actually be 220 instead of 200 - this is not what you get from the headlines.

Now, let us go to the Newsminute article about Kerala excess deaths calculation.

The article uses the following formula to find the projected death rate in 2021 without the pandemic:

"TNM calculated excess deaths based on the 2015-2019 data from the state records. The state registered an average of 98,387 deaths from January 1 to May 31 every year from 2015 and 2019."

But, how was this average calculated? I tried calculating it for the whole year from 2015 to 2019, and I got a different number (as I have mentioned in a previous article).

Did the Newsminute calculate this number after taking into account population increases, as the study below does?

If yes, how did it get the population of the state from 2015 to 2019? Was that a projected figure too?

We do not know. My suspicion from their wording is that they simply took the death numbers from 2015 to 2019 and then averaged them. This is not the correct method, though.

My question is, how were these people qualified to do this analysis? Why did they rush to do it while the surge was going on, when we would have inaccurate numbers?

This excess death study is just evolving, and it should have been left to academics.

I have no doubt there were excess deaths; and that there was misreporting. This was true of Kerala and it was true of UP.

By rushing to use an academic tool in the middle of the surge to sensationalise, and (in the case of Burka Dutt and others) cast aspersions on govt officials, I believe the journalistic community was highly irresponsible.

References:

US Study: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1003571

Newsminute article on Kerala: https://www.thenewsminute.com/article/tnm-exclusive-kerala-reports-almost-14000-excess-deaths-till-end-may-2021-151124