On August 10, The Hindu carried an op-ed, “ COVID-19 deaths may be higher than reported ”, by two community physicians, Hemant Deepak Shewade and Giridara Gopal Parameswaran. They argued that the COVID-19 death figures for India may have to be adjusted for low coverage of medical certification of cause of death (MCCD). They were criticised for estimating COVID-19 deaths by relying on the fact that only 22% of deaths are medically certified in India. This exercise amounted to fudging of data, some said.
Flagging issues surrounding data
I would like to give the piece some context before I evaluate the complaints against it. The Hindu has been flagging issues of data suppression and data manipulation since the Central Statistics Office started re-basing GDP estimates. Economist R. Nagaraj has explained in more than one lead article why there is distrust in the GDP figures. Writers have spoken of how the integrity of numbers should be restored. In this column, I wrote about the disastrous effects of data manipulation by quoting Nobel Prize-winning economist Ronald Coase: “If you torture the data long enough, it will confess to anything.”
With reference to COVID-19 numbers, there are many speculative reports from government bodies themselves. For instance, in June, a PTI report quoted scientists associated with the Indian Council of Medical Research (ICMR) saying that the pandemic in India would peak around November 2020. When The Hindu carried the report, the ICMR issued a clarification on Twitter: “The news reports attributing this study to ICMR are misleading. This refers to a non-peer reviewed modelling, not carried out by ICMR and does not reflect the official position of ICMR.” Another report by PTI, “ Facing criticism for ‘under reporting’ of COVID-19 deaths, Delhi issues SOP for hospitals ” (May 10), flagged the issue of under-reporting. Since the break of the pandemic, this newspaper has carried analytical articles as well as ground reports. The ‘Ground Zero’, “ Telangana’s COVID trials and tribulations ” (July 18), was about lack of transparency of data and inadequate testing. It is in this context that one has to read the recent op-ed on under-reporting of COVID-19 deaths.
Giridhara R. Babu, Professor of Epidemiology at the Indian Institute of Public Health, Bengaluru, in his lead article, “ The challenges in counting the dead ” (August 14), explained the complexities involved in calculating COVID-19 deaths. He wrote: “To obtain unbiased, reliable estimates, nationally representative verbal autopsy surveys such as the Million Death Study can be helpful. In the meantime, India has to adopt a strategy of transparency in data flow and reporting, with increased investments in innovations for improving data collection and reporting.”
A fair assumption
The August 10 article flagged major lacunae in the Indian system of collecting data relating to death. Critics pointed out that the analysis was based on the MCCD report for 2017 when the report for 2018 had been released a few weeks prior to the publication of the article. I agree that the writers should have revised their article based on the latest report; their arguments would have stood ground even with the 2018 report. The article’s assumption was that COVID-19 death reporting is similar to non-COVID-19 death reporting in the absence of specific data from the Indian government pertaining to COVID-19. This seems to be a fair assumption; they were not fuelling conspiracy theories.
We have reached a situation where efforts to make sense of numbers, in the absence of rigorous methods from government institutions, are characterised as partisan readings. In a May 20 report, " Is India undercounting its COVID-19 deaths? ", The Hindu had raised the same question based on available data. If that report is read along with “ Coronavirus deaths increasing significantly in several States ” (June 16), it is clear that the problem with the data conundrum lies with the government and not with researchers who are trying to make sense of numbers by stitching together pieces of available information. One can differ with the writers’ methodology and their estimate, which is based on a raw factor of multiplication and could have been refined with other variables that could affect the count, but that does not diminish their plea for a robust health data system.