Coronavirus | India may have 8 to 10 lakh cases a day in mid-May, says Michigan University epidemiologist Bhramar Mukherjee – The Hindu
April 24, 2021
Complacency with a false sense of security has led to the spike in cases, she says.
With the daily infections accelerating at a blazing speed to reach 3,45,103 on April 23, and the daily deaths stubbornly remaining above 2,000 and rising since April 20, the second wave is growing at an alarming rate resulting in health-care facilities bursting at the seams. The second wave is expected to peak in May. Bhramar Mukherjee, Professor of Epidemiology at University of Michigan in an email says there will be 810 lakhs cases a day in mid-May when it peaks, and 4,500 deaths around May 23. Edited excerpts:
Since April 1, the number of daily cases has been accelerating at a rapid speed. Can it be any reason other than more infectious variants?
We have to be cautious here. Causality can sometimes be established by elimination of alternative explanations. Let us try that argument here.
We all agree that it is not a single factor but a confluence of different factors all coinciding to create the perfect transmission inferno in India. Lack of covid-appropriate behaviour at a time when the country was fully reopening, massive rallies, religious gatherings, cricket matches, use of public transportations, all were taking place largely without proper face covering, throwing caution to the wind. Indoor facilities with air-conditionings like malls, theatres, restaurants were buzzing with people.
We were complacent with a false sense of security, thinking we have conquered COVID-19. Instead of anticipating the silent footsteps of this insidious virus, we let it run wild without any surveillance. Even when we saw the uptick in mid-February, we were dismissive and continued with data denial. The nonchalance, negligence, complacency and hubris cannot be ignored. Colossal mistakes were made by not accelerating vaccination when the virus curve was at its nadir.
Even with all of those features factored in, and allowing for a certain rate of re-infection consistent with existing literature (84% protection from past infections at seven months), the growth rate that we are seeing with cases growing by 8-folds, deaths increasing by 9-fold, and eight States having a reproduction number (R0) around 2 cannot be adequately explained without entertaining the possibility of an intrinsically more transmissible variant. We have data now from different Indian States showing that the double mutant or the UK variant have quickly become dominant strains in Maharashtra or Punjab for example. The increasing number of reports of cluster/family level infections also point to this hypothesis. However, without proper sequencing data over geography and time and proper epidemiological investigations, this evidence is still circumstantial.
Bhramar Mukherjee
Even if the rise is due to new highly transmissive variants, why are we seeing a sudden acceleration since April 1?
This is the nature of exponential growth, the virus creeps in silently and explodes astronomically. The rate parameter of the growth is startling, but the pattern is explainable. This is a feature of the last surges for example in the US and UK. During the 1918 Influenza pandemic, India saw a similar pattern.
We started imposing lockdowns only recently to slow down transmission. Before then we were having not one or two isolated superspreader events but a continuous flow of numerous superspreader events.
The reproduction number is over 2.5 in Uttar Pradesh and Bihar, and above 2 in Delhi, Rajasthan and West Bengal for a few days now. At this high reproduction number, are these States reporting the expected daily cases?
Our papers have consistently estimated underreporting factors for reporting cases nationally around 10-20. The IHME model is projecting 45 lakhs daily new infections today in India, pointing to a daily underreporting factor of about 15. This factor widely varies across States. Even with inaccurate numbers the relative trends are clear. From all I know, the reality on the ground is much starker than what the numbers show.
I would like to reiterate that suppressing the truth or having artificially deflated numbers does not help anyone. It hinders prudent policymaking, prevents estimating true healthcare needs or need for oxygen supply/ICU beds accurately. This pandemic has turned into this confusing policy pandemonium partially because the data and science have not been presented transparently to the public.
Based on the high reproduction number in these Uttar Pradesh, Bihar, West Bengal, Rajasthan, Madhya Pradesh, Gujarat for days, are we seeing the expected number of deaths now?
We have estimated death underreporting by a factor of 2-5 in the first wave. Now with the surge, the reporting infrastructure has probably eclipsed dramatically. So I expect the underreporting of deaths to be massive right now. All reports from burial grounds and crematoriums strongly suggest this possibility.
The fact is, we have a relative idea of the growth but we have no idea about the absolute numbers. I tell my students that this India modelling exercise is to teach them to adopt best statistical practices with the worst possible data. Finally, even if we believe the reported death numbers, the IHME is projecting 664,000 reported deaths by August 1 for India. Each number is a person and I am so heartbroken to see the loss of countless human lives that could have been saved, particularly when in a few months we may have copious vaccine supply.
Misclassification of COVID-19 deaths and attributing the cause of death to other comorbidities has happened to some extent in every country. The excess mortality calculations can provide a holistic evaluation of COVID-related deaths, comparing say year 2020 to historic data. For example, in the U.S. there have been 23% excess deaths than expected from March 2020-January 2, 2021 and 73% of those are attributed to COVID-19.
But in India, medical reporting of deaths and cause of deaths is already a very porous system so it is challenging to do such calculation reliably to quantify COVID-related fatalities in an indirect way. The data deficient infrastructure in India is really hurting us right now.
The seven-day average test positivity rate (TPR) nationally on April 23 was 18.5%. Delhi (30.5%), Chhattisgarh (30.1%), Maharashtra (24.6%), Madhya Pradesh (23.8%), Andhra Pradesh (22%) and West Bengal (20.4%) are reporting higher TPR than the national average. Are the daily fresh cases reported from these States in concordance with the test positivity rate?
These high levels of TPR can capture both increasing prevalence or limited testing. I think in this case it is a combination of both and impossible to unconfound one from the other. Again, I think all arrows point that cases are severely underreported.
How much should the daily tests be in these States to detect cases early and to bring down the TPR?
The testing shortfall can be estimated by setting a target TPR, if you set it at 5% say, that indicates it should be 4-5 times more than current level. You can also be clever with testing strategies by repeated testing with rapid tests instead of all RT-PCR tests to avoid testing bottleneck. India should also allow the home testing kit that we have in the U.S. now produced by Abbott which is inexpensive, easy to use and accurate. You can be clever with all of these strategies, there are so many papers now on optimal allocation of tests with limited budget. You have to innovate and be open to using new efficient tools.
Why are we seeing low TPR in Uttar Pradesh (12.5%), despite the number of tests done being less than in Maharashtra? What are the reasons for this?
You are asking me about a ratio where I neither believe the reported numerator nor the denominator. It could be that patients with obvious COVID-19 symptoms are not even being tested. Selection bias in testing can distort the numbers you get. We have worked on this issue of selective testing. I would like to add that some RT-PCR tests have a high false negative and they may not have the same accuracy to detect new variants if they are optimized for the original strain.
You had tweeted saying Uttar Pradesh's growth in spread is alarming. Our models are failing at this high rate of growth to come up with sensible predictions. Is the growth in spread alarming only in Uttar Pradesh?
No, not just Uttar Pradesh. Uttar Pradesh, West Bengal, Bihar and Delhi are on top of my high alert list. Then comes Andhra Pradesh, Rajasthan, Madhya Pradesh, Kerala, Gujarat, and Karnataka. Kerala is again starting to look worrisome. I feel West Bengal, Uttar Pradesh, Bihar and Kerala will need lockdown at some point. Odisha and Assam also have a high R0 value but the projected number of total cases is lower.
When do you think the second wave in India will peak and what will be the daily fresh cases reported at the time it peaks nationally?
All models are projecting a peak for infections in May right now. Deaths will be a lagged indicator by 7-10 days. The IHME is projecting early-May and we are projecting mid-May for infections to peak. We are projecting reported cases at 8-10 lakhs a day with 4,500 deaths, whereas the IHME predicts about 50 lakh infections (reported plus unreported) and 5,500 deaths at the peak of the two curves.
Do you expect a third wave in India? Are we anywhere close to reaching the daily vaccinations needed to avert a third wave?
Depends on how fast we vaccinate. We need to get to 10M vaccines a day (with the assumption of two dose vaccines). To vaccinate 800M adults it will then take another 5 months. If we could procure one dose vaccines like the J & J that will be best.
Very plausible that this will not be the last wave, this will not be the last variant we are seeing. We need to have an agile public health alert system to deal with this situation governed by data, science and humanity. We need to continue to build healthcare capacity, oxygen supply, ICU beds. Sequence reinfections, breakthrough infections.
Preparation and anticipation is the key to prevention. We have had a sluggish start to the vaccination. I am hoping with the new policies (like opening up to 18+ from May 1, approving multiple other vaccines with EUA) we can ramp up and have copious supplies by the summer.
Despite the increasing number of deaths seen since April 1 (from less than 500 daily deaths to over 2,000 on April 20) the case fatality rate is continuously dipping. How do you explain this?
Case fatality rate (CRF) is calculated by taking the ratio of deaths to cases. It could be that deaths have been growing but cases are growing at a faster rate, but please remember that death also is a lagged indicator. This lag is not incorporated in the current calculation. We should really calculate CFR by the number of deaths divided by the number of recovered plus deaths as we do not know what proportion of the active cases will die. A fair comparison could also be dividing todays death by cases reported two weeks ago.
I really want to advocate to look at the absolute numbers of active cases here. It is your number of active cases that determines what proportion will need oxygen/ventilators and drives your plan for gauging the need for healthcare capacity.
In general we do see a lower overall mortality in more recent surges in the U.S. as younger people are infected who have less co-morbidities. We should really compare age-specific mortalities across two waves, not overall mortalities here.
It seems like the released data by the government does not indicate that younger people are more infected in the second wave, though it seems from the same briefing that there is enrichment in disease severity in younger age groups compared to the first wave. I would love to get individual level or age-sex stratified data to study this.
Read the rest here:
Coronavirus | India may have 8 to 10 lakh cases a day in mid-May, says Michigan University epidemiologist Bhramar Mukherjee - The Hindu
