Listen to this article:
The Wire is grateful to Murad Banaji for his help with preparing this report.
Data collected and analysed by Citizens for Justice and Peace, a human rights organisation, and independent experts shows deaths in the Purvanchal region of eastern Uttar Pradesh, home to Prime Minister Narendra Modi’s constituency of Varanasi, surged in shocking fashion as the COVID-19 epidemic raged in the state.
From January 2020 to August 2021, the areas surveyed saw around 60% more recorded deaths than expected from 2019 records as well as from government data on the death rate in the state before the pandemic.
If the rise in mortality in these areas was repeated across all of Uttar Pradesh, the state as a whole could have had around 14 lakh (1.4 million) excess deaths during the pandemic – roughly 60-times Uttar Pradesh’s official COVID-19 death toll of 23,000.
This would make Uttar Pradesh one of the hardest-hit states in the country during the pandemic – and also one where death recording has been the weakest.
Polling for the state’s assembly elections began on February 10.
Across India, the COVID-19 epidemic has taken a heavy toll. Official figures, however, reflect only part of the suffering. We know now from many independent reports and studies that COVID-19 deaths in India have been grossly under-reported. Academic studies using civil registration statistics and data from various surveys have revealed that India has likely seen over 3 million excess pandemic deaths.
However, mortality data from Uttar Pradesh, India’s largest state, has been very limited. Official COVID-19 deaths in Uttar Pradesh number around 23,000. A report from IIT Kanpur claimed that thanks to the state government’s effective response to the outbreak, COVID-19 has had limited impact on the state.
On the other hand, especially during the second wave, around May 2021, the media was awash with stories of death sweeping through the state, of bodies swelling the Ganga and of patients dying avoidable deaths without oxygen.
Official claims and ground reports appear to tell entirely different stories. Where does the truth lie?
We decided to dig deeper, hoping to use data available in the public domain to understand the scale of COVID-19 deaths in Uttar Pradesh.
During November-December 2021, the Uttar Pradesh team of Citizens for Justice and Peace (CJP) undertook a mammoth exercise to collect records of deaths from 2017 to August 2021, as maintained by local offices in villages and urban areas in the Purvanchal region of eastern Uttar Pradesh. We succeeded in obtaining complete records from 129 areas, largely in Varanasi and Ghazipur districts.
A basic analysis of this data reveals a big spike in recorded deaths in 2020-2021 compared to previous years as well as to expectations based on government data on Uttar Pradesh’s death rate.
While not all deaths can be conclusively attributed to COVID-19, it is reasonable to conclude that the pandemic played a key role in the huge increase in mortality.
Let us dive in.
Our teams faced many challenges when collecting death records in the regions we visited. Although we attempted to obtain data from many areas in four districts – Varanasi, Ghazipur, Jaunpur and Chandauli – the majority of data we were eventually able to collect came from Varanasi and Ghazipur districts.
Over the course of gathering the data, our team approached Nagar Nigam offices, gram panchayat offices and crematoria and burial grounds, and sought the help of village heads and secretaries, as well as ASHA workers. In many areas, we were unable to access any data – either because death records were not maintained or because they were not shared.
In the end, we were able to obtain partial or complete death records from 2017 up to August 2021, from 147 rural and urban areas. Village officials and ASHA workers were key to our success in obtaining this data, especially in the Varanasi district.
To assess trends in the data, we used only numbers from areas with complete records going back to 2017. We removed from the analysis 17 areas that did not have records for every year, and one area for which we could not find population data. Of the remaining 129 areas whose data we analysed, 79 were rural and 50 were urban. Of these, 104 areas were in Varanasi, 23 were in Ghazipur, and one each was in Jaunpur and Chandauli districts.
The 129 areas we used in our analysis had an estimated total population of around 2.8 lakh in 2021. Of them, around 43,000 people resided in urban mohallas and the remainder were in rural areas. That is, the surveyed population was roughly 85% rural. This is an approximate reflection of Uttar Pradesh as a whole: according to government projections, around 76% of the population resides in rural areas.
The number of deaths recorded in the 129 areas from 2017 to August 2021 is set out in the following table.
We observe first that there were significant increases in the number of deaths recorded year-on-year in these areas prior to the pandemic. In fact, recorded deaths increased by 9% between 2017 and 2018, and a further 23% between 2018 and 2019. It is likely that much of this increase is a consequence of improved record-keeping.
Indeed, some local officials told our team that they do not maintain a record of all deaths in the village. They only register the deaths of those people who approach them to get a death certificate. And it is usually for reasons connected to pension benefits, life insurance claims, property settlements and bank accounts that families wish to get a death certificate or a deceased relative’s name recorded in the death register.
It is likely that, over time, more and more families chose to report deaths to local officials.
During the pandemic period, we see further sharp rises in recorded deaths. Could further improvements in record-keeping, with more families needing death certificates, explain the rise in deaths between 2019 and 2020, and again between 2020 and August 2021? A closer look reveals that this is highly implausible. Let us see why.
From our data, we can calculate an annualised crude death rate (CDR) for each time period – namely, the number of recorded deaths occurring per 1,000 population per year in the surveyed population. This gives us the numbers in the following table.
We see that the CDR, according to death records, was indeed rising prior to the pandemic in 2017-2019 – no doubt reflecting increasing levels of death registration, and probably not increasing mortality.
According to government estimates in the 2019 Sample Registration System (SRS) bulletin, the actual CDR for Uttar Pradesh as a whole is 6.5 (6.9 in rural areas and 5.3 in urban areas). Based on the rural-urban make-up of the surveyed population, we would expect the CDR in the surveyed areas to be around 6.7. We see that by 2019, the recorded CDR of 6.4 in this population was in fact close to state-level expectations. There was not much room for further improvement in reporting to push the numbers up.
It is important that, in these areas, the CDR in 2020 is already around 15-20% higher than expected either from 2019 data or from the state-wide CDR estimates from the annual SRS. And the death rate during January-August 2021 was, shockingly, more than double the expectation.
Even if we assume that the SRS significantly underestimated pre-pandemic yearly deaths in Uttar Pradesh, and further assume that record-keeping improved to perfection during the pandemic period (highly unlikely, especially given major disruption to death registration during the lockdown), we still find deaths during the pandemic period to be greatly above expectations.
It is inconceivable that such a big surge in recorded deaths reflects anything other than many more people dying.
We can try to gain more insight into the scale of the mortality surge by looking separately at rural and urban areas. Analysing rural and urban data, we find the following annualised CDR values.
We see that in the surveyed areas, the rise in CDR during 2020 was largely driven by a sharp rise in urban deaths. But in case it is claimed that 2020 had no increase in rural deaths – we note that there is in fact a more modest, but nevertheless clear, rise in rural deaths reported from 2019 to 2020.
The dramatic increase in deaths during January-August 2021 came from both rural and urban areas. Indeed, deaths in rural areas during the first eight months of 2021 were double that expected from the 2019 data – and around 80% higher than that expected based on the SRS-estimated rural CDR.
The surge is consistent with reports during the second wave of large numbers of COVID-19 deaths among the state’s teachers and reports in the Hindi press of the death and destruction wreaked by the virus in Uttar Pradesh’s villages.
Estimating excess pandemic deaths across UP
We cannot be sure if the situation in the parts of Purvanchal that we visited was reflected across the state. But we can nevertheless ask: if the mortality surge seen in the surveyed areas was repeated across the state, what would this imply about the scale of pandemic deaths in Uttar Pradesh?
The surveyed areas had 55-60% more deaths during the 20 months from January 2020 to August 2021 than expected in this period. Across Uttar Pradesh, a 55-60% surge over 20 months would amount to around 14 lakh (1.4 million) excess deaths in this period.
To understand this figure, note that, based on SRS and civil registration data, Uttar Pradesh expects to have around 15 lakh deaths in a normal year. The pandemic excess death toll would then amount to almost a full year’s extra deaths in the state.
Another perspective on these stunning numbers is that 14 lakh people amounts to around 0.6% of Uttar Pradesh’s estimated 2021 population, of around 23 crore. This means 0.6% of the state’s population possibly lost their lives prematurely as a result of the pandemic.
How does this compare to the pandemic toll in other parts of India?
It is on the very high end. Andhra Pradesh has good civil registration data available, and which has reported the highest pandemic excess mortality in India. In fact, estimated excess deaths in Andhra Pradesh amount to a little over 0.5% of the state’s population. Our survey thus suggests that when it comes to pandemic mortality, Uttar Pradesh could be worse-hit than Andhra Pradesh – and possibly the hardest-hit state in all India.
Understanding the results
As it emerged, the data confirmed our worst fears: COVID-19 had indeed left a gory trail of death in every town and village in Purvanchal.
The story in the data was confirmed by multiple testimonies. Muniza Khan, a social science researcher who compiled the data, said, “We were shocked by the sheer number of deaths in these four districts. ASHA workers and Panchayat heads said they did not know a single house where a person had not died due to COVID-19.”
Khan recalled how anguished families would throng Adikeshav ghat, where bodies of people who had died with COVID-19 were sent for their last rites. The place was packed with families waiting their turn to bid their loved ones goodbye, while funeral services workers struggled to find firewood for cremations. “They finally ran out of firewood. The bodies that didn’t burn properly were simply thrown into the Ganga,” she said. “Still the long lines continued.”
Not just hospitals and crematoria – there were long lines at burial grounds for Muslims and other minorities as well. Employees were working overtime, digging graves even as space ran out and people waited outside. One worker at Ramnagar, near the city area, told the CJP team, “I have been digging all day long but the lines outside the burial ground won’t shorten.”
It is also noteworthy that many burial grounds across India for Muslims, Christians and other minorities, who bury their dead, were found to be saturated with bodies to the point that older graves had to be ‘recycled’ to accommodate the growing number of the dead.
The tragedy was compounded by the weak or absent health and welfare infrastructure. Already during relief work, the CJP team observed how healthcare centres in rural areas were often shut. Families approached the CJP headquarters to ask for medicines and oximeters. Both in towns and in villages, education and three-meals-a-day had become everyday challenges. One elderly individual told Dr Khan, “The last two years have been the most traumatic years of my life. I don’t even want to remember this time.”
Our investigation found a staggering increase in the number of deaths during the pandemic in Varanasi and nearby areas. This discovery is very different from the official story of a well-managed epidemic with limited impact. In fact, local people in some areas told us that every household had lost a family member to the virus and poor healthcare.
While the bulk of the rise in deaths occurred during 2021, presumably during the second wave of the outbreak, there was also significant excess mortality during 2020. And although additional deaths during 2020 seemed to be concentrated in urban areas, even the rural areas we surveyed saw around 20% more deaths during 2020 than during 2019.
There is little convincing evidence that this reflects improved record-keeping. On the contrary, data from many parts of the country shows death registration plunging during the first part of 2020 as several offices shut during the national lockdown.
Across our sample, deaths from January to August 2021 were more than twice as many as we could expect from the 2019 data, and SRS estimates. In a period of eight months, in this population, we would expect to have around 1,200 recorded deaths based on 2019 records or the state’s death rate as estimated in the SRS. Instead, we found 2,570 deaths.
While the population we surveyed was concentrated mainly around the Varanasi district, our data suggests that Uttar Pradesh as a whole was hit very hard, especially during the second wave. It is clearly crucial to survey more parts of the state – to assess the true impact of the COVID-19 epidemic as well as to bring the lives lost in the state out from behind its government’s veil of propaganda.
A more complete report, including all the data used in this piece and further methodological details, will shortly be made available on the CJP website.