India’s excess deaths not derived from global model: WHO team
NEW DELHI: Senior members of World Health Organisation (WHO) team tasked with calculating the global death toll from the coronavirus pandemic responded Monday to the Union government’s statement that criticised the model they used for arriving at their death estimates for India, and said their numbers for excess mortality in India were solely based on data from within the country — and not on some global formula.
Researchers working in the Technical Advisory Group (TAG) of the WHO Covid-19 Mortality Assessment Group shared an extract from their yet-to-be-published research paper titled “Estimating Country-specific Excess Mortality During the COVID-19 Pandemic” detailing their methodology for countries that did not have official national all-cause mortality data and explained how they arrived at individual mathematical models to reach their estimates.
The controversy about the volume of “excess deaths” in India kicked off on April 16, when a report in The New York Times claimed that an effort by WHO to calculate the real global death toll from the pandemic “has been delayed for months because of objections from India, which disputes the calculation of how many of its citizens died and has tried to keep it from becoming public”. The article put India’s death toll from the pandemic at nearly four million.
Hours later, the Union health ministry responded to the article and said India has been in “regular and in-depth technical exchange with WHO on the issue”. “India’s basic objection has not been with the result (whatever they might have been) but rather the methodology adopted for the same,” the government said in a statement.
The “excess death” calculation at WHO is being performed by TAG. An advisory body to WHO, its primary role is to support efforts to assist the UN agency and UN member states in obtaining accurate estimates of numbers of deaths attributable to the direct and indirect impacts of the pandemic.
“Excess death” is a term that refers to the total number of deaths occurring due to all causes during a crisis that is above and beyond what would have been expected under regular conditions. To be sure, not all such deaths may be due to Covid-19, but during a pandemic an abrupt rise in fatalities is likely to be either directly or indirectly caused by the outbreak.
The full paper, which has been submitted for publication, will be available shortly, said Professor Jon Wakefield, TAG member and professor of statistics and biostatistics at the University of Washington.
The extract of the paper explained that for a handful of countries (which included India), the team did not have the national all-cause mortality (ACM) data and that researchers constructed different statistical models to estimate country-specific death tolls. Other than India, this list of nations included Argentina, China, Indonesia and Turkey. Their model expanded a proportionality assumption method (which assumes that hazards are proportional over time) previously proposed by Ariel Karlinsky (2022) to analyse excess mortality in Argentina, it stated.
“For India we have data from up to 17 states and union territories... over the pandemic period (out of 36), but this number varies by month,” it said.
The state/ut-level data for India was gathered by a series of official reports, and Right to Information (RTI) data requests by journalists. “For India, we use a variety of sources for registered number of deaths at the state and Union territory-level. The information was either reported directly by the states through official reports and automatic vital registration, or by journalists who obtained death registration information through Right To Information requests,” the statement explained.
To be sure, several news organisations, HT included, have tried to estimate all-cause mortality figures for which they have generally used data from Civil Registration System (CRS) — a national system of recording all births and deaths, under the Office of the Registrar General of India and implemented on the ground by state governments — to count the number of excess deaths. A wide range of undercount has appeared in such analysis by several news organisations — ranging from as little as 0.42 times in Kerala to 48 times in Bihar. Some experts have previously argued that CRS may not be the best database to estimate all cause deaths.
The WHO researchers also stressed that for the India analysis, they did not use a global model.
“We stress that for India the global predictive covariate model is not used and so the estimates of excess mortality are based on data from India only,” they wrote.
One of the key concerns raised by India’s ministry of health was regarding the covariates used by the WHO team. “Of the covariates used for analysis, a binary measure for income has been used instead of a more realistic graded variable. Using a binary variable for such an important measure may lend itself to amplifying the magnitude of the variable. WHO has conveyed that a combination of these variables was found to be most accurate for predicting excess mortality for a sample of 90 countries and 18 months,” the health ministry said in its April 16 statement, adding that the Indian government was yet to receive from WHO a detailed justification of how the combination of these variables is found to be most accurate.
But the researchers themselves admitted to the limitations to their approach in the summary findings.
“If, over all (local) regions, there are significant changes in the proportions of deaths in the regions as compared to the national total, or large changes in the populations within the regions over time, then the approach will be imprecise,” they wrote. However, they added: “We have carried out sensitivity (for example, we remove different subsets of states and run the model) and cross-validation analyses” to address these shortcomings.
Several issues raised by the Indian government (such as inverse relationship between monthly temperature and monthly average deaths, which it said had “no scientific backing”) in its April 16 statement found no mention in the excerpt tweeted by Wakefield.
When asked about the Indian statement regarding the New York Times article, a spokesperson from WHO said their full report concerning excess deaths would be released in the “near future”. “We are planning the release of the Excess Mortality Data in the near future. Once there is a date of release, we will let you know,” the body said in an emailed response.
To be sure, several other researchers have tried to estimate the total number of excess deaths that took place in India during the pandemic, and have arrived at figures that suggest a significant undercounting. The excess deaths figures in these studies have ranged from 2 million to 4.9 million.
One such scientist, Prabhat Jha, director of the Centre for Global Health Research at St Michael’s Hospital in Toronto, whose research paper published in the journal Science estimated India’s Covid deaths by September 2021 to be “six to seven times higher than reported officially” said that his findings were similar to the numbers from WHO. “Our paper in Science using additional data sources reached similar conclusions to the WHO,” he said about his paper which put the excess death toll in India till June 2021 to over 3 million.
These studies have used a variety of sources and methods ranging from household surveys to CRS data, to infection fatality rate estimates for Covid-19. As of
Monday morning, the Indian government’s official death toll from the pandemic is 521,965.
To be sure, a precise picture on the real mortality numbers can be a near-impossible task in a country the size of India, especially over a large period of time as has been in the case of the global pandemic, which spanned over two years.
Experts stressed that it is still prudent for India to set up a system which records precise number of deaths that registers a definitive scientific cause as well.
“In the time of an epidemic when excess deaths due to all causes are reported with no explanation, it can be attributable to the disease that has caused the outbreak. It is an assumption, nonetheless, since there is not an adequate system in place that can give the real picture. What is required in our public health system is the need to set-up a proper system to determine exact death numbers on daily or weekly basis in each city or town with a definitive cause of death determined by scientific method; and adequate actions in that control the cause. It is doable. European cities did it 300 years ago. It is not rocket science. India must invest in such a system,” said Dr Dileep Mavalankar, director, Indian Institute of Public Health, Gandhinagar.
Rhythma Kaul contributed to this report