Medics wearing protective suits collect swab samples for the COVID-19 test during the nationwide lockdown, in wake of the coronavirus pandemic, in Chennai, Friday, May 1, 2020. Photo: PTI
Even as the lockdown has been extended beyond May 3, it is imperative that India’s public health strategy for dealing with the COVID-19 pandemic starts shifting towards testing, tracing and isolating. Lockdown is just not sustainable beyond a point in a poor country like India, and, as has already become amply clear, it imposes enormous costs on the most vulnerable sections of the population.
Testing will need to replace lockdown as one of the central elements of the alternative strategy. It will involve adequate testing to detect positive cases as they emerge, tracing contacts of individuals who test positive and conducting tests on them, and isolating infected persons either at home or at quarantine facilities.
If this strategy is implemented properly, it will keep transmission rates from spiking again in the post-lockdown period. Keeping transmission rates low, in turn, will be necessary to prevent case counts from increasing rapidly and overwhelming India’s weak health infrastructure.
The key component of this post-lockdown strategy is adequate testing. To implement this strategy, it is necessary to have reliable estimates of the magnitude of “adequate” testing. Comparing the benchmark figures for “adequate” testing with actual testing right now, we can arrive at rates by which COVID-19 testing needs to be ramped up.
Using such a method, this article presents estimates of ‘adequate’ number of tests on May 15 across 16 Indian states that have seen the worst outbreak of the COVID-19 epidemic.
A caveat is in order before we proceed. The numbers reported in this article rely on the predicted case count on May 15. The predicted case count comes from a statistical model. Like all models, its results come with possibilities of error – either in the upward or lower direction. The possible error in the predicted case count directly feeds into my calculations of ‘adequate’ number of tests. Hence, the projections for ‘adequate’ number of tests could be mistaken, to the precise extent to which the predicted case count is mistaken.
Moreover, the COV-19-IND study predictions, which I have used, are updated daily – as new data becomes available. When this article is published, the predicted case count for May 15 will be different from, and most probably lower than, what I have used for the calculations. Even if the projected case count on May 15 is half of what I have reported it will still mean that almost all states have to ramp up tests at a rapid rate. That finding will remain unchanged – only the magnitude of the ramping will change.
Adequate testing rate
According to epidemiologists and public health experts, an indicator of adequate testing is a consistently low test positivity rate (ratio of positives and total tests conducted). The World Health Organisation (WHO) recommends a test positivity rate of 10% (1 positive for 10 tests conducted) as a good benchmark. Countries that have successfully dealt with the pandemic so far, like South Korea and Germany, have attained lower test positivity rates (TPR) of 2% and 6%, respectively.
According to data available on covid19india, India’s test positivity rate on April 29 stood at 4.28%. This is a relatively low TPR and provides some evidence that India’s testing regime, thus far, has been successful by global standards. But there is a large variation between states. On the one hand, there are states like Kerala (1.91%) and Odisha (0.4%), which have low TPRs. On the other hand, many states have relatively high TPRs: Maharashtra (7.3%), Gujarat (7.4%), Madhya Pradesh (6.3%).
Given these observed trends across Indian states, I will define three benchmarks for adequate testing in any Indian state: a target TPR of 2%, 3% and 4%. Attaining a target TPR of 2% in each state is the best possible outcome. It will mean attaining Kerala’s testing effectiveness in every state, a highly desirable outcome, given Kerala’s success in dealing with the pandemic.
At the other extreme, hitting a target TPR of 4% is the worst outcome. It will mean that each state attains test effectiveness that is just about equal to India’s overall average right now. In a previous article, I had argued that India’s TPR at 4% was not to be interpreted in a positive light. This is because India’s TPR of 4%, even though low by international standards, has not yet started falling. A stable TPR, in the case of India, suggests that adequate numbers of tests are not being conducted. With an adequate number of tests, the ambit of testing will move from high to medium to low risk groups – so that the TPR should start falling.
The third benchmark, a target TPR of 3% in each state, is a medium-range outcome. It will mean a performance that lies between the two extremes. Given Kerala’s experience, we should clearly prefer the 2% TPR benchmark.
Adequate number of tests
To convert adequate test positivity rates into adequate numbers of tests, I will adopt a method used in a recent study by Harvard scholars to predict testing rates in US states. To compute the adequate number of tests on some future date, we just need to divide the predicted number of cases on that date by a target TPR. For instance, if the predicted case count on a future date is 500 and our target TPR is 2%, then the adequate number of tests on that day would be 25,000.
The COV-19-IND study group, an interdisciplinary group of scientists, have used epidemiological models to predict the number of cases (daily and cumulative) for Indian states. They predict the case count for three different scenarios: no intervention, social distancing and travel ban (but no lockdown), lockdown. The ‘no intervention’ scenario gives the largest predicted case counts; the ‘lockdown’ scenario gives the smallest predicted number of cases; and predictions for the ‘social distancing and travel ban (but no lockdown)’ scenario falls in between.
India obviously intervened. So the ‘no intervention’ scenario is not relevant. So, the choice is between the lockdown and social distancing (without lockdown) scenarios. To choose between the two, I will see which scenario’s predictions come closer to observed numbers.
Using data till April 1, the COV-19-IND study had predicted that the cumulative case count on April 30 would be 35,796 under the ‘social distancing and travel ban (but no lockdown)’ scenario, and it would be 3,951 under the ‘lockdown’ scenario. The cumulative case count on April 30 is 34,866, a number that is very close to the social distancing scenario prediction from the COV-19-IND study, and it is very far from the lockdown scenario prediction (in this case, the prediction is off by a factor of close to 9). For this reason, I will use projected case counts for the scenario with social distancing and travel ban (without lockdown).
Using observed data till April 30, the COV-19-IND study group has made projections until May 31. Using the projected numbers for May 15 and dividing by the target TPR (2% or 3% or 4%), I arrive at projections for the adequate number of tests on May 15 – for India as a whole and for different states.
Projections for the number of adequate tests at the all India level
For the country as a whole, the projected daily case count on May 15, according to the COV-19-IND study group, is 13,590. If testing must attain a test positivity rate of 2% (the best case scenario), the total number of tests conducted on May 15 have to be 6,79,500 (which is 13590/0.02). If the target TPR is to be 3%, the total number of tests on May 15 should be 4,53,000, and if the target TPR is 4%, then the total number of tests on May 15 need to be 3,39,750.
On April 29, India conducted 54,031 COVID-19 tests. Hence, if a target TPR of 2% is to be attained by May 15, daily tests have to increase 13-fold within 16 days. This means an increase in daily testing at about 17.4% every day.
If a target TPR of 3% is to be attained, daily tests have to increase by a multiple of 8 (implying a growth rate in daily testing of 14% every day); and if a target TPR of 4% is to be attained, which is the worst possible of the three scenarios, daily tests have to increase by 6 times (implying a growth rate of daily testing by 12% every day).
Projections for the number of adequate tests at the state level
I carry out the same exercise for 16 states that have the highest case counts of COVID-19 and for which case count projections are available from the COV-19-IND study group. The results of this analysis are presented in Figure 1 and Figure 2, and clearly highlight the enormous differences across states. In fact, we can divide these 16 states into three groups.
In the first group is Telangana, which is the clear outlier (hence, I have not included this state in Figure 1 or 2). It has the largest deficit in the number of tests and will need to ramp up testing at the fastest rate. To attain a 2% TPR, Telangana will need to increase daily testing 63-fold within 16 days.
The second group, which is shown in Figure 1, consists of states that will require high rates of growth of daily testing. Gujarat, Maharashtra, Madhya Pradesh, Uttar Pradesh and West Bengal belong to this group. For each of these states, attaining a target TPR of 2% will require increasing daily testing by close to or more than 20-fold between April 29 and May 15 (implying a daily growth rate of 21%).
The third group of states, which are part of Figure 2, will require relatively lower growth in daily testing. In this group, we have Andhra Pradesh, Bihar, Delhi, Haryana, Jharkhand, Karnataka, Kerala, Odisha, Punjab, Rajasthan and Tamil Nadu. It is worth noting that even in these better performing states, daily testing needs to be ramped up significantly – by more than five-fold if a 2% TPR is to be attained by May 15 (other than in Jharkhand, Odisha and Tamil Nadu).
The main conclusion that should be drawn from these projections is that there is a need to increase testing in most states, and this need is most urgent in the following six states: Gujarat, Maharashtra, Madhya Pradesh, Telangana, Uttar Pradesh, and West Bengal.
Deepankar Basu is an associate professor in the Department of Economics, University of Massachusetts Amherst.