India has confirmed more than 450,000 cases of COVID-19 so far, making it the world’s fourth-worst-hit country. Major cities such as Delhi and Mumbai are particularly badly affected, with hospitals struggling to accommodate critically ill patients .

The current surge in infections follows a two-and-a-half-month India-wide lockdown that began on 25 March and severely disrupted the economy and livelihoods. Some researchers say the government failed to take advantage of this time to prepare the country’s struggling health infrastructure.

Even as India struggles, the true scale of the epidemic there might not be apparent. The country has an incomplete death-registration system, which means that not all deaths are recorded and the documented cause is often incorrect.

This raises questions about India’s COVID-19 mortality rate, which is officially 11 deaths per million people in the population — among the lowest in the world. By comparison, the United Kingdom has seen 635 deaths per million people, and the United States has seen 376.

Jayaprakash Muliyil, an epidemiologist at the Christian Medical College in Vellore in the state of Tamil Nadu, has been advising the Indian government on COVID-19 surveillance and helped to design the government’s first sero-surveillance survey, of 26,400 people, to estimate the proportion of the population with viral antibodies. He talks to Nature about some of the factors affecting India’s epidemic, and discusses why officials in some badly affected cities seem reluctant to say that outbreaks are being driven by community transmission — where cases cannot be linked to known sources.

Do you think the outbreak in India is charting a different path from outbreaks in other badly hit countries, such as the United States, Italy or Spain?

It is. It is spreading much faster here, and the infection rates are higher. The general population’s anxiety about the disease is low. People will willingly go out into the market today, and take fewer precautions to protect themselves. Consequently, at least in cities, the epidemic is growing very rapidly. And we know it is spreading in rural areas too. The whole trajectory of the infection is moving upwards more sharply than in many other countries. What happened in many Western countries is that when a big city like London was affected, other cities reacted strongly and reduced transmission. So, everywhere else, the doubling time got longer, but in some Indian cities it is short.

India is reporting that its mortality rate is among the lowest in the world. Is that accurate?

The mortality per million people in India is expected to be lower because of the low average age of India’s population. (Older people are more likely to die from this infection.) So, we can take some comfort in the fact that deaths are fewer, especially in the rural population.

But the problem with death as an indicator is that a COVID-19 death has to be certified as such. The only way to do this is through an RT-PCR test (a reverse-transcription polymerase chain reaction test, which looks for viral genetic material in nose and throat samples). And with a population of 1.3 billion, what do you think is the proportion of people that has access to this kind of testing? It is very low.

So, it is very difficult to count all deaths due to COVID-19. There is no way you can get it done, unless rapid tests become more widely available. Remember that at least half of all deaths will happen in rural villages — around 66% of our population. And there are no real mechanisms to ascertain causes of deaths in these villages.

What do you think of the Indian government’s response to the epidemic so far?

The lockdown all over the country was not the right response. It brought misery to untold numbers of people and destroyed lives. And we haven’t been able to repair its consequences for society. That was unfortunate. If we had planned the lockdown better, we would have still had losses, but they wouldn’t have been greater than what we are experiencing now. The excuse we had for the lockdown was to gain time. But what we achieved in that time, I don’t think is commensurate with the damage from the lockdown.

The lockdown was announced with only four hours’ notice. Would advance warning have helped?

The generalized lockdown did have one benefit: everyone became aware of this thing called COVID-19. It is not easy to communicate this to everyone in India, with its many remote regions, but because of the lockdown, people heard about it. The concept of an infectious disease is not easy for many to understand. In many rural areas, measles is considered to be caused by a goddess visiting a village. So is chicken pox. There, when you introduce the term virus, it doesn’t make sense to many groups of people.

But we should have introduced the lockdown after adequate warning. And in consultation with people, telling them what to expect and when it would end. Instead, there was absolute surprise when it was announced. The livelihoods of migrant workers and daily-wage earners — who are a very large section of our population — were lost, and people were traumatized. This damage may take time to repair, because these are the quiet people who suffer in silence.

At the same time, the ramping up of health infrastructure occurred only in bits and pieces. Several regions simply didn’t respond.

Why do you think some cities seems reluctant to say that the outbreak is now being driven by community transmission?

Someone must have told them that community transmission is a sign of failure. I have been trying to question this idea from the beginning — the virus spreading in the community is not the government’s fault. It is the way the virus works. Because the transmission is being driven in part by people without overt symptoms, it is very difficult to stop the virus, especially in a densely populated country such as India.

The government's first sero-survey estimated that about 0.73% of India's population had antibodies. Now, if we apply that percentage to the population in rural India excluding tribal areas, some 750 million people, that means about 5 million people have been infected and recovered in these areas.

The only way we can explain this, given that most rural Indians don’t travel abroad, is community transmission. And this sero-survey happened more than three weeks ago. The numbers would since have increased, because each person would have infected somebody else. (The study has not yet been peer-reviewed or published, and sero- surveys can be unreliable.)

What should cities with large outbreaks do now?

Many cities are quarantining people returning from COVID-19-affected states or countries in public facilities and hotels. I would say that should stop, and these people should quarantine at home. Most of them won’t know whether they have been infected, because they might not have been tested. And when the number of infected people is already high in the community, quarantining incoming travellers in public facilities, which is very labour-intensive, is not economical.

Instead, we should focus on two things. The first is a reverse quarantine for elderly people — where the old and the vulnerable are quarantined from others to protect them. The second is to put all our money in hospitals, and provide oxygen for patients. That manoeuvre will save lives.