|
 
 |
| ORIGINAL ARTICLE |
|
| Year : 2020 | Volume
: 6
| Issue : 2 | Page : 91-95 |
|
A multistate ecological study comparing evolution of cumulative cases (trends) in top eight COVID-19 hit Indian states with regression modeling
Sudhir Bhandari1, Ajit Singh Shaktawat1, Amit Tak2, Bhoopendra Patel3, Kapil Gupta2, Jitendra Gupta2, Shivankan Kakkar4, Amitabh Dube2
1 Department of Medicine, S.M.S. Medical College and Attached Hospitals, Jaipur, Rajasthan, India
2 Department of Physiology, S.M.S. Medical College and Attached Hospitals, Jaipur, Rajasthan, India
3 Department of Physiology, Government Medical College, Barmer, Rajasthan, India
4 Department of Pharmacology, S.M.S. Medical College and Attached Hospitals, Jaipur, Rajasthan, India
| Date of Submission | 14-May-2020 |
| Date of Acceptance | 21-May-2020 |
| Date of Web Publication | 29-Jun-2020 |
Correspondence Address:
Dr. Amit Tak
4, Pushpa Path, Uniara Garden, Moti Dungri Road, Jaipur - 302 004, Rajasthan
India
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/IJAM.IJAM_60_20
Background: The global pandemic of coronavirus disease – 2019 (COVID-19) has wrecked the very fabric of mankind putting its survival at stake. The prior knowledge of trends of cumulative cases helps in management of disease epidemic by optimized allocation of logistics and human resources.
Materials and Methods: An ecological study was undertaken to compare the standardized trends of cumulative cases of top eight highly COVID-19 affected states of India with linear regression modeling. The data were sourced from Kaggle repository and Unique Identification Authority of India. The coefficients of regression of linear regression models of all the eight states were compared using analysis of covariance (ANCOVA).
Results: It was observed that evolution of COVID-19 was the highest in the state of Gujarat (b = 0.186, P < 0.001) followed by Madhya Pradesh (b = 0.166, P < 0.001), Maharashtra (b = 0.159, P < 0.001), Delhi (b = 0.156, P = 0.02), Rajasthan (b = 0.136, P = 0.98), Uttar Pradesh (b = 0.117, P < 0.001), Tamil Nadu (b = 0.091, P < 0.001), and Andhra Pradesh (b = 0.076, P < 0.001) respectively.
Conclusion: It is seen that ranking of states on the basis of trends of evolution and the absolute number of cumulative cases are different. The trends of evolution assist public health authorities and governmental agencies in providing right picture of evolution and help in decision making process during management of epidemic.
The following core competencies are addressed in this article: Medical knowledge, Practice-based learning and improvement, Systems-based practice.
Keywords: Analysis of covariance, coefficient of regression, coronavirus disease 2019, cumulative cases, linear regression modeling
How to cite this article:
Bhandari S, Shaktawat AS, Tak A, Patel B, Gupta K, Gupta J, Kakkar S, Dube A. A multistate ecological study comparing evolution of cumulative cases (trends) in top eight COVID-19 hit Indian states with regression modeling. Int J Acad Med 2020;6:91-5 |
How to cite this URL:
Bhandari S, Shaktawat AS, Tak A, Patel B, Gupta K, Gupta J, Kakkar S, Dube A. A multistate ecological study comparing evolution of cumulative cases (trends) in top eight COVID-19 hit Indian states with regression modeling. Int J Acad Med [serial online] 2020 [cited 2023 Jun 10];6:91-5. Available from: https://www.ijam-web.org/text.asp?2020/6/2/91/287965 |
| Introduction | |  |
The Severe Acute Respiratory Syndrome-Coronavirus-2 outbreak in Wuhan (China) in December 2019 crossed the boundaries and become a pandemic.[1] Coronavirus disease-2019 (COVID-19) has manifest its presence in more than two hundred countries of the world with total number of confirmed cases 3.9 million confirmed cases and 270,765 deaths globally.[2] The researchers across the planet committed to find out new management strategies to lessen the effect of pandemic. A number of drugs and vaccines are under clinical trial. Since India got its first case in Kerala, the number of cases is rising in different states of India. India is the home to 17.7% population of the world and is the seventh largest country of the world. The uncontrolled pandemic in India has grave consequences with respect to trade, economy, and defense. The Indians represent significantly on the world map and assist in strengthening the economy of all the countries of the world.
The management of any disaster is limited by human resources and materials. The planning of disaster management requires the inputs for the experts in operational research. To optimize use of national resources and manpower is the key to contain such epidemic. There are 29 states and 7 union territories in India. The vast diversity between different states with respect to climate, education, culture, language, population density and others, make it more difficult for any unified strategy to work. The mechanisms of transmission of corona virus and affect of various environmental and other factors are not completely understood and the effects of lockdown and isolation measures are evaluated on the empirical basis. After observing the pattern of COVID-19 pandemic in various countries, it can be presume that this epidemic is not short lived. Therefore, there is urgent requirement to make decisions on scientific basis, so there is need to quantitate various situations. Since the notification of first case, various public health measures has been implemented in the country. The quarantine measures were implemented on all the airports of India. The initial trends of various states emerge from various factors like number of tourist visits, population density, presence of airports, and quarantine measures.
In the present study, we aimed to compare trends of COVID-19 evolution in the top eight COVID-19 hit states of India. The study hypothesized that the rate of evolution of cumulative cases in various states of India are different and not corresponds to absolute number of cumulative cases. To compare the rate of evolution statistically, comparison of coefficients of regression (slopes) of regression lines after logarithmic transformation of cumulative cases per day of various states were performed.[3] Most of the states in India have enforced public health measures such as lockdown, contact tracing, testing of suspects, personal protective measures to contain the virus. The study reflects the performance of various strategies taken by state governments.
| Materials and Methods | | |
An ecological study was designed to compare trends of COVID-19 evolution in top eight states of India having highest number of COVID-19 patients (n = 8). The bivariate data (time, number of cumulative COVID-19 cases) of all the eight states from the day when the hundredth case were reported until the fifteen hundredth case was reported had been collected. The number of bivariate data points might be different for different states, as days required to reach from the hundredth to fifteen hundredth case may vary. The time series data of cumulative cases of COVID-19 were sourced from Kaggle [4] repository (www.kaggle.com/sudalairajkumar/covid19-in-india/activity) and the population of Indian states were sourced from Unique Identification Authority of India (www.uidai.gov.in/images/state-wise-aadhaar-saturation.pdf).[5],[6] The Kaggle is a subsidiary of Google LLC that is an online community of data scientists and machine learning practitioners. Kaggle provides a cloud based workbench for data science and artificial intelligence applications. The datasets and competitions organized by Kaggle were published in many journals.
Data analysis
Logarithmic transformation of number of cumulative cases was performed to compare the trends of various states. The dataset was fitted to a linear regression model, with dependent variable as number of cumulative cases and regressor as time. The linear regression model can be defined as follows:
yi = ai + bit + e
Where ai is the intercept and bi is the coefficient of regression and e is the error term of the models. The estimation of parameters of various different linear models was performed with respective dataset of states. The coefficient of regression reflects the evolution of trends. The higher the value of coefficient of regression means higher rate of evolution.
Statistical analysis
To compare the coefficient of regression of eight states, analysis of covariance (ANCOVA) was performed. The ANCOVA evaluates whether the means of regression coefficient (dependent variable) are equal across the eight states (independent variable often called treatment). The statistical level of significance was considered at 5%. Microsoft Excel 2010 was used for creating database and analysis is performed on MATLAB [7] 2016a (9.0.0.341360).
| Results | | |
The linear regression models of various states were fitted with corresponding dataset as shown in [Figure 1]. The mean of coefficients of regression of eight states are shown in [Table 1]. The mean of coefficients of regression are statistically significant except for the state of Rajasthan. The comparison of coefficients of regression of different states was performed with analysis of covariance (F (7,147) = 748.25, P < 0.001). The plot of confidence intervals of various coefficients of regression is shown in [Figure 2]. We observed that the evolution of COVID-19 was highest in the state of Gujarat (b = 0.136, P value < 0.001), followed by Madhya Pradesh (b = 0.166, P < 0.001), Maharashtra (b = 0.159, P value < 0.001), Delhi (b = 0.156, P = 0.02), Rajasthan (b = 0.136, P = 0.98), Uttar Pradesh (b = 0.117, P < 0.001), Tamil Nadu (b = 0.091, P < 0.001), and Andhra Pradesh (b = 0.076, P < 0.001). There is no significant differences between the coefficients of regression of Gujarat, Maharashtra, Delhi, and Madhya Pradesh. Similarly, there is no significant differences between coefficients of regression of Andhra Pradesh and Tamil Nadu. The Uttar Pradesh coefficient of regression is significantly different from above two groups of states and the value lie in between them as shown in [Figure 2]. [Table 2] outlines the position of the each state based on absolute number of cumulative cases and evolution trends. It can be seen that on the basis of number of cumulative cases, Maharashtra stood first, followed by Gujarat, Delhi, Tamil Nadu, and Madhya Pradesh. The evolving trends in Maharashtra, Gujarat, Delhi, and Madhya Pradesh are similar. Tamil Nadu has shown decreased evolving trends as compared with above states. | Figure 1: Shows linear regression models of top eight COVID-19 hit states of India. The slope of lines corresponds to coefficient of regression. The slope of regression line shows different evolving rate COVID-19 of various states of India
Click here to view |
 | Table 1: Estimates of coefficients of regression of top eight coronavirus-19 disease hit Indian States
Click here to view |
 | Figure 2: Shows the mean (red open dots) and 95% confidence intervals (red horizontal lines) of coefficients of regression, with Gujarat and Andhra Pradesh securing the highest and lowest coefficients of regression
Click here to view |
 | Table 2: Depicts the ranking of top eight coronavirus-19 disease hit states of India
Click here to view |
| Discussion | | |
As per the World Health Organization,[8] the coronavirus is highly contagious and increased more than tenfold in merely 10 days. As per the Ministry of Health and Family Welfare,[9] Government of India, as of 08 May 2020, 08:00 (IST), there are 37916 active cases, 15539 cured/recovered cases and 1886 deaths have occurred in India. The highest number of confirmed cases of COVID-19 is in Maharashtra. The top eight states with confirmed cases include Maharashtra – 17974, Gujarat – 7012, Delhi – 5980, Tamil Nadu – 5409, Rajasthan – 3427, Uttar Pradesh – 3071, and Andhra Pradesh – 1847. The management decision in pandemic based on not only on the absolute numbers of cases, but evolving trends. In the present study, we observed that the evolution of COVID-19 was highest in the state of Gujarat (b = 0.187, P < 0.001), followed by Madhya Pradesh (b = 0.166, P < 0.001), Maharashtra (b = 0.159, P < 0.001), Delhi (b = 0.156, P = 0.02), Rajasthan (b = 0.136, P = 0.98), Uttar Pradesh (b = 0.117, P < 0.001), Tamil Nadu (b = −0.091, P < 0.001), and Andhra Pradesh (b = 0.076, P < 0.001). It can be seen that there is difference in the total number of cumulative cases and rate of change of cumulative cases (trends) in various states [Table 2]. There is significant difference between the coefficients of regression of various states (F (7,147) = 748.25, P < 0.001). There are no significant differences between the evolution rates of Gujarat, Maharashtra, Madhya Pradesh and Delhi. There is no significant differences between Tamil Nadu and Andhra Pradesh. The state with higher evolution trends has the potential to override the states with highest number of cumulative cases in future.
Gupta [10] suggested effects of temperature and humidity in United States. The location of various states and their trends does not favor effect of weather in Indian scenario. Singh and Adhikari [11] modeled the effect of age and social contact structure through SIR (susceptible, infected and recovered) model. The social contact structure varies in various states of India. They also investigated the role of social distancing measures across time duration and suggest sustained lockdown with periodic relaxations. Sardar et al.[12] modeled transmission of COVID-19 and incorporate the lockdown effect and investigate the transmission between symptomatic and asymptomatic populations and consider former as fast spreader. They assessed the 21 days lockdown effects in three states, viz., Maharashtra, Delhi, and Telangana in terms of reduced cases and deaths and proposed no effect of lockdown in Maharashtra, while some effect in Delhi and Telangana. Bhattacharyya et al.[13] in their study use curve fitting to trace the epidemic curves of USA, Italy, Spain, and India and mentioned the number of cases at the announcement of lockdown lead to higher coefficients of fourth degree fitted polynomial. Large value of coefficients leads to increased evolution rates. Mandal et al.[14] used mathematical models to evaluate to what extent the quarantine of symptomatic patients will prevent or delay the outbreak. They estimate that if 50% of symptomatic individuals are quarantined within 3 days of developing symptoms, the reproduction number would be 1.5 and if asymptomatic individuals lacking any infectiousness the cumulative incidence would reduce by 62%. Belfin et al.[15] find out the effect of lockdown and peak of epidemic curve in India. They use third-degree polynomial to estimate the basic reproduction number that comes out to be 3.3 (95% confidence interval, 3.1–3.5), and the epidemic peak will be arrived on September 28, 2020.
There is an urgent need to investigate into the strategies and factors responsible for observed trends in various states. Planning is for tomorrow, and management is for today. Planning involves opinion of experts from various fields beyond the public health like economics, statistics, sociology, management. In view of the present situation, the polices of the government should aimed at achieving the balanced social and economic development along with public health.
| Conclusion | | |
The study concludes that while managing the COVID-19 pandemic, the strategies developed by various states should consider the evolving trends along with the absolute numbers of cumulative cases. The states having higher trends should be managed on priority and various factors responsible for such trends should be evaluated and steps should be taken to avert such scenarios.
Limitations of the study: In our study, we have taken the sequence of cumulative cases per day since reporting of the hundredth case to fifteen hundredth case that may lie in unequal time periods. Though these unequal time periods lies during the first lockdown phase of India. Second, number of the sample of time points is small.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
Research quality and ethics statement
The present study complies with the reporting quality, formatting, and reproducibility guidelines set forth by the EQUATOR Network. The data used were secondary in nature because it is not generated at first hand by the investigators, the dataset is maintained by Kaggle (a subsidiary of Google LLC) which further outsourced from Ministry of Health and Family Welfare, Government of India, which make the data available for general public.[9] Data are deidentified before release and consent of study subject is reasonably presumed, so does not require Institutional Review Board/Ethics Committee review.
| References | | |
| 1. | Kaushik S, Kaushik S, Sharma Y, Kumar R, Yadav JP. The Indian perspective of COVID-19 outbreak. Virusdisease 2020; pp 1-8. https://doi.org/10.1007/s13337-020-00587-x.  |
| 2. | |
| 3. | Maindonald JH. Time series analysis with applications in R, by Jonathan D. Cryer, Kung-Sik Chan. Int Stat Rev 2009;77:300-1. |
| 4. | |
| 5. | |
| 6. | |
| 7. | Matlab. Version 9.0.0.341360 (R 2016a). Natick, Massachusetts: The Mathworks Inc.;Statistics and Machine Learning Toolbox, ANOVA, Analysis of Variance and Covariance Copyright: 1994-2016. |
| 8. | Srivastava N, Baxi P, Rat ho RK, Saxena SK. Global trends in epidemiology of coronavirus disease 2019 (COVID-19). In: Coronavirus Disease 2019 (COVID-19). Singapore: Springer 2020. p. 9-21. |
| 9. | |
| 10. | Gupta S, Raghuwanshi GS, Chanda A. Effect of weather on COVID-19 spread in the US: A prediction model for India in 2020. Sci Total Environ 2020;728:pp-138860. https://doi.org/10.1016/j.scitotenv. 2020.138860 |
| 11. | Singh R, Adhikari R. Age-structured impact of social distancing on the COVID-19 epidemic in India. arXiv Preprint arXiv: 2003;pp:12055. doi: https://arxiv.org/abs/2003.12055v1 |
| 12. | Sardar T, Nadim SS, Chattopadhyay J. Assessment of 21 days lockdown effect in some states and overall India: A predictive mathematical study on COVID-19 outbreak. arXiv Preprint arXiv 2004;pp:03487. https://arxiv.org/abs/2004.03487v1. |
| 13. | Bhattacharyya A, Bhowmik D, Mukherjee J. Forecast and interpretation of daily affected people during 21 days lockdown due to COVID 19 pandemic in India. medRxiv 2020;pp-20075572. doi: https://doi.org/10.1101/2020.04.22.20075572. |
| 14. | Mandal S, Bhatnagar T, Arinaminpathy N, Agarwal A, Chowdhury A, Murhekar M, et al. Prudent public health intervention strategies to control the coronavirus disease 2019 transmission in India: A mathematical model-based approach. Indian J Med Res 2020;151:190-9. [ PUBMED] [Full text] |
| 15. | Belfin RV, Brodka P, Radhakrishnan BL, Rejula V. COVID-19 peak estimation and effect of nationwide lockdown in India. medRxiv 2020;pp-20095919 doi: https://doi.org/10.1101/2020.05.09.20095919. |
[Figure 1], [Figure 2]
[Table 1], [Table 2]
|
Search Pubmed for