Analysis of COVID-19 Case Fatality Rates in the States and Union Territories of India

Although the Covid-19 Case Fatality Rate (CFR) in the Indian States and UTs has been changing with time, some states constantly appear to show signicantly higher CFR than the national average. Our objective is to calculate the CFR of all the states/UTs of India and analyse the possible factors behind the disparities in it. Research papers and news articles on Covid-19 were explored to understand the factors responsible for the CFR disparities in the States/UTs. State-wise CFR was calculated and Correlated with Covid-19 Testing Rates and data from Demographic & Healthcare factors, using Spearman’s Rank Correlation Coecient Methodology. The overall Covid-19 CFR in India was among the lowest (1.76%) in the world but varied vastly from one state to another. Where the states like Punjab and Maharashtra constantly have the highest CFR in the country, states like Assam, Kerala, and Bihar have the lowest. In the correlation analysis, a weak agreement (+0.33) between state-wise CFR and ‘Test Positive Rate’ was found. CFR and ‘Life Expectancy at 60’ showed a moderate agreement (+0.49). Healthcare components like ‘Number of Doctors Per Million People’ and ‘Number of Hospital Beds’ showed very weak agreement with CFR. Where the higher Life Expectancy and Test Positive Rates clearly tend to increase CFR, Healthcare Facilities had surprisingly little effect on it. Analyses of various news articles suggested that Comorbidities, Availability of Essential Drugs, Trained Manpower, Contact Tracings, and Hospital Referral Time were also some of the major factors affecting CFR.


Introduction
Coronavirus Disease 2019 (Covid- 19) is an infectious disease caused by a newly discovered coronavirus called Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) 1 . The rst case of Covid-19 in India was reported on 30 th January 2020 in the state of Kerala 2 . By October 9, 2020, the total number of con rmed Covid-19 (SARS-CoV-2) cases in India stood approximately 70 lakhs, of which nearly 1,07,000 people, unfortunately, lost their lives 3 . Although the number of active cases in India started declining from the second half of September, there were still approximately 75,000 new cases and 1000 deaths daily due to Covid-19 in India 3 . The cumulative number of Covid-19 Cases and the Recoveries with time is shown in Before going for any analysis, we need to agree upon a de nition of Case Fatality Rate (CFR) for an ongoing pandemic and understand Test Positive Rate (TP). For an ongoing pandemic, it may be a little confusing to precisely calculate its 'Case Fatality Rate' (CFR). According to Piotr Spychalski et al. 4 , during an epidemic, cases might be de ned as either total con rmed cases, or by cases which had an outcome (deceased + recovered). Hence, the denominator for calculating CFR might be either of these numbers. In the initial phase of the epidemic, the 'number of cases with outcome' is relatively small, so CFR calculated per 'cases with outcome' will be an overestimate. But CFR calculated per total cases will be an underestimate as the numerator will be underestimated. But since we are well past the initial phase of the Covid-19 pandemic in India on October 9, 2020, as per a WHO scienti c brief 5 it would be more appropriate for us to be using 'cases with outcome' in the denominator to calculate CFR. Hence, we have used the following formula to calculate the CFR of Covid-19 in India: Here, 'Total number of cases with Outcome' = 'Total number of patients recovered' + 'Total number of patients deceased' till given date.
'Test Positive Rate' shows the proportion of the total number of people who were tested positive for a virus to the total number of people who got tested. The Test Positive Rate (TP) till a given date can be calculated using the following formula: The CFR of India was calculated based on the above-mentioned CFR formula, and plotted with time as shown in Figure 3. We can observe a continuously decreasing trend in the CFR with time, primarily due to the improvements in recovery rates and increased number of testing, resulting in more positive cases with mild or no symptoms.
Since our primary objective is to calculate the Covid-19 CFR in the different States and UTs of India and analyse the various possible causes for the difference in their CFR, we have analysed recent research papers and news articles on Covid-19 to know about the possible factors affecting CFR.
Jennifer Beam Dowd et al. 6 highlighted the role of demographics, particularly the age structure of a population in determining the Covid-19 death rate. They suggested that the progression of Covid-19 and its mortality rate are strongly connected to the age structure of the population affected. It was found after analysing the Covid-19 data from countries like China, South Korea, Italy, and Germany, etc. that mortality risk was highly concentrated at older ages, particularly for people aged above 80. John P.A. Ioannidis 7 concluded that infection fatality rates inferred from seroprevalence studies tend to be much lower than earlier speculations during the initial days of the pandemic. He also pointed out that the infection fatality rate of Covid-19 can vary substantially across different locations depending on the population age structure, case mix of infected and deceased patients as well as various other factors.
Arghadip Samaddar et al. 8 , showed that the severity of Covid-19 in India was among the lowest in the world with low fatality rates, ICU admissions rate, and need for ventilators. In this paper, they suspected several factors having a role in reducing the susceptibility of Indians to Covid-19. These factors included several ongoing mutations in the circulating SARS-CoV-2 strains with lesser virulence, host factors like innate immunity, genetic diversity in immune responses, epigenetic factors, genetic polymorphisms of ACE2 receptors, micro RNAs and universal BCG vaccination, and environmental factors like temperature and humidity. Graziano Onder, Giovanni Rezza, and Silvio Brusaferro in their paper 9 , after analysing the early data of Covid-19 In Italy, found that deaths were concentrated in older male patients with multiple comorbidities. Shahbaz A. Shams, Abid Haleem, and Mohd Javaid 10 compared the Covid-19 data of the top 18 worst affected countries to conclude that there is a positive relationship between average life expectancy and fatality rates due to Covid-19. This was mainly because the vulnerability to this disease increases exponentially with age, especially after the age of 60, due to weak respiratory and immune systems and other comorbidities that occur at later ages.
Sourendu Gupta 11 found that younger adults in India have more Covid-19 infection as compared to their proportion in the population. He also suggested that women are half as likely to be infected by Covid-19 as men with signi cantly lower infection rates for women between puberty and menopause. Similarly, Manisha Mandal and Shyamapada Mandal 12 found that the relative susceptibility of developing symptoms (RSODS) and relative susceptibility of developing an infection (RSODI) was almost 33 times higher among younger people aged below 45 years. William Joe et al. 13 found that males share a higher rate of Covid-19 infection than women, but the infection is evenly distributed in under 5 and elderly age groups.
The study found a CFR of 14.3% for people aged above 60 as compared to an overall CFR of 3.2% in India.

Correlation Methodology
To nd the correlation between state-wise CFR and other data sets, we have applied the "Spearman's Rank Correlation Coe cient Methodology". The value of the correlation coe cient (r s ) varies between -1 to +1, where +1 suggests perfect correlation and -1 suggests perfectly correlated in the opposite order, and 0 (or near 0) suggests uncorrelated. The formula for calculating correlation coe cient (r s ) is given bellow: Here d i = X i -Y i is the difference between the ranks of each data sets and n is the number of elements in each of the data sets. Here X i and Y i are the rank data columns of rst and second data sets after sorting them in ascending order (or   Figure 4, dividing them into 5 different categories of CFR from Very Low (0-0.671%) to Very High (2.688-3.360%). The gure was prepared using ArcGIS 10.5 software.
The different CFR categories from Very Low to Very High and Indian states and UTs falling in those categories are shown in Table 1. The correlation between CFR and the Expectation of Life at the age of 60 in the Indian States/UTs is graphically shown in Figure 6.
We may infer from the above results that the higher CFR moderately agrees with the higher life expectancies and median age as states with higher life expectancies and median ages will have more older people who, as the studies 6 suggest may be more susceptible to the Covid-19 disease.

Correlation between CFR and Healthcare Data:
The The correlation coe cient for CFR and the number of 'Government Allopathic Doctors Per Million People' in the Indian states and UTs was calculated r s = -0.04, suggesting almost no relation between them. The correlation between CFR and 'Government Allopathic Doctors Per Million Population' in the Indian States/UTs is graphically shown in Figure 7.
To correlate CFR with healthcare infrastructure, we calculated correlation coe cients for CFR and the total number of Hospital beds, ICU beds and Ventilators per 1000 Covid-19 Cases in the different states and Union Territories of India. The correlation coe cients for Hospital beds, ICU Beds and Ventilators per 1000 Covid-19 Cases with CFR were all found to be around r s = +0.07, suggesting almost no relation between CFR and the availability of healthcare infrastructure items like Hospital beds, ICU Beds and Ventilators. The correlation between CFR and 'Hospital Beds per 1000 Covid-19 cases' in the Indian States/UTs is graphically shown in Figure 8.
We can infer from the above analysis related to the healthcare data indicates that the state-wise CFR is surprisingly not related to the availability of healthcare infrastructure and government doctors. This indicates that other factors overwhelm the availability of healthcare facilities in deciding the Covid-19 CFR in the states and UTs of India.
All of the above data analysis related to Testing Rates, Demographics, and Healthcare exhibit a moderate to no relation with the Case Fatality Rates (CFR) which points towards the possibility of other factors, that may be responsible for the vast differences in the CFR of Covid-19 in the States and UTs of India. Because of the lack of updated data for any other possible factor for correlation analysis, we analysed several news articles about Covid-19 CFR in India in some of the leading News Papers to know about the other possible factors affecting the Covid-19 CFR. Due to the lack of updated data for any other possible factor for correlation analysis, analysis of several news articles about Covid-19 CFR in some of the leading newspapers was done, which suggested that pre-existing comorbidities, availability of essential drugs, trained manpower, medical expertise, effective contact tracing, and timely referral to hospitals were also some of the major factors affecting Covid-19 CFR in the States and UTs of India.

Declaration of competing interest
The authors declare that they have no known competing nancial or non-nancial interests or personal relationships that could have appeared to in uence the work reported in this paper.