Many rural areas were hit hard during India’s devastating recent Covid-19 surge. The scale of this rural epidemic remains largely hidden in official figures. But a flood of news reports tell a tale of infection sweeping rapidly through villages, high mortality, minimal testing, and health systems unable to keep up.

We collected a total of 61 case-studies from the first three weeks of May, each describing at least five suspected Covid-19 deaths in one or more villages. We searched the Hindi press for such reports, and hence the focus is on Hindi-speaking states.

Of the studies, 26 were from Uttar Pradesh, nine from Haryana, eight from Bihar and six each from Madhya Pradesh, Jharkhand, and Rajasthan. The details and links to the media reports are in the accompanying technical documentation.

A surge in mortality

The reports describe a total of 1,297 deaths from villages with an estimated combined population of around 480,000. This means that in these villages taken together around 0.27% of the population died of suspected Covid-19. In individual reports, this figure varied between 0.05% and 1% with a median value of 0.31%. This means that a typical village in this sample with 5,000 inhabitants would have seen around 15 deaths in the space of roughly as many days. Indeed, many reports described roughly one death a day.

There are several uncertainties. For example, some reports name all the deceased while others give an approximate figure reported by the village head. The reports cover periods ranging from under a week to over a month; but it is not always clear whether they describe all deaths in the village in these periods, or only those suspected to be from Covid-19. The death tolls may also be incomplete – outbreaks were often ongoing when journalists filed their reports, and sometimes follow-up reports give higher death counts.

Given the uncertainties, we can look at each location and ask how many deaths we would expect to occur in normal times in the period described in the report. Using crude death rates from 2018, we should expect a total of around 174 deaths in these villages during the periods covered in the reports. The reports thus describe around 1,123 “excess deaths” in these villages.

Put another way, the mortality described is more than seven times expected. It is little wonder that many of the reporters who visited the villages describe panic, confusion, and sometimes a feeling of abandonment.

Putting the numbers in perspective

We’ll call the ratio of excess deaths to the total population excess mortality. Taken together, the reports describe excess mortality of 0.23%, or 2.3 excess deaths per thousand population. In individual reports, excess mortality varies from 0% to 0.95% with a median value was 0.29%.

Thirteen of the reports describe excess mortality of over 0.5%. It seems that during this wave when a village was hit by a Covid-19 outbreak, it was not very uncommon for one in every 200 villagers to die in a month or less.

To see that these figures are remarkably high, consider Mumbai, one of the cities hit hardest by the pandemic. To date, around 0.12% of Mumbai’s population have officially died of Covid-19, while according to data from the municipal corporation, excess mortality during 2020 was around 0.17%.

We shouldn’t rush to assume that the high excess mortality documented in these village reports reflects the situation across all rural areas of the states covered. The reports focus on villages where many deaths occurred, and the overall picture is undoubtedly more varied. But what the reports do indicate is that disease can spread rapidly in rural areas, with tragic consequences.

Were all of the deaths from Covid-19?

The great majority of deaths described in the reports were not from confirmed Covid-19. However, it seems likely that most were, in fact, from the disease.

Firstly, the timing of the reports coincides with surging cases in the six states from where the reports came. News reporters often turned up in the villages precisely because of a spurt of unexplained deaths consistent with a Covid-19 epidemic.

Moreover, most reports describe symptoms consistent with the disease amongst village residents. The reports frequently mention breathlessness, fever and “Covid symptoms” – although it is not always clear whether all of the deceased had these symptoms.

In some cases – generally after most of the deaths – district health teams arrived and tested a sample of villagers. Testing usually confirmed that many villagers were, indeed, infected with SARS-CoV-2, the virus responsible for Covid-19.

Very low testing and reporting

While testing did sometimes occur, it seems that only a small minority of those who died were ever tested for the disease. Some reports are explicit that none of the deceased were tested, while in others this is strongly implied. In some instances, the villagers refer to the deaths as “mysterious”, or even as being from other diseases such as typhoid or malaria. Some reports describe a reluctance to get tested, and even denial that the epidemic was real.

A few reports give an official figure, perhaps from a local health official, for the number of patients who were tested, and for the number of deaths confirmed to be from the novel coronavirus. Based on the reports it appears that fewer than 10% of the deaths described were officially recorded as Covid-19 deaths.

The typical situation could be even worse: sometimes health teams arrived and did some testing precisely because of the scale of mortality and perhaps the media attention. Thus the villages from which such reports emerge might have seen higher testing and reporting than typical.

Estimating fatality rates

Infection fatality rate refers to the fraction of SARS-CoV-2 infections that result in death. Limited data on the spread of infection and weak death recording mean that credible IFR estimates from rural India are lacking.

We can try to infer from the reports some minimum estimates for Covid-19 IFR in these areas during the second surge. If we were to assume that all the excess deaths were from Covid-19, and that everyone in a village got the disease, then excess mortality would indeed equal IFR. While the assumption that all the excess deaths were from Covid-19 is likely close to the truth, it is unlikely that all villagers were infected in most of the studies; hence, excess mortality will, in general, underestimate IFR.

The median excess mortality of 0.29% can thus be considered a conservative estimate of rural IFR during this surge. More pessimistically, the excess mortality of over 0.5% described in over a fifth of the reports could be closer to the true value of IFR in these areas during this surge. Note that this is roughly twice the median estimate for Covid-19 IFR in Mumbai during 2020.

Preventable deaths very possibly pushed up IFR in these parts of rural India: the unavailability of medical care, and in particular of oxygen, is described in several of the reports. It may also be that more lethal variants of the virus increased the death rate.

Conclusions

The latest Covid-19 crisis has caused a surge in mortality in parts of rural India. News reports do not allow us to infer the scale of the mortality crisis – for this, death registration data or careful surveying will be needed. But they suggest that if disease enters villages, it can sweep rapidly through, and cause many deaths.

What triggered these village outbreaks? We can speculate that high active infection levels coincided with high mobility to cause such devastation. Several of the reports from Uttar Pradesh associate the rural outbreaks with people moving around because of the local elections. Some reports describe a lack of awareness or precautions, or a particular superspreading event. It is striking how fast many of the outbreaks progressed: more transmissible variants of the virus may have accelerated the spread.

According to the reports most of the deceased were not tested, and many did not receive medical attention. Poor rural disease surveillance and health infrastructure are frequent themes. Some of the deaths were likely preventable. It seems that inadequate public health messaging led to generally poor awareness about the pandemic in rural areas. This is consistent with several reports of vaccine hesitancy in rural areas.

The reports give no reason to believe that Covid-19 IFR is low in rural India. During the first wave, there were regions in rural India which saw high levels of spread according to seroprevalence surveys, but almost no recorded Covid-19 fatalities (examples from Bihar in this piece, and a similar picture emerges in Jharkhand).

If the current data is anything to go by, the low death figures reflect poor testing and recording rather than some natural “protection” from severe disease in rural India. It is likely that the great majority of rural Covid-19 deaths in many parts of India have gone unrecorded.

Murad Banaji is a mathematician at Middlesex University, London.
Aashish Gupta is a PhD Candidate in Demography and Sociology at the University of Pennsylvania.
Leena Kumarappan is an independent researcher

Links to all the news reports, and all the data used for calculations, are in an accompanying more detailed version of this piece.