An ecological time-series study design was adopted, using secondary data.

The first case of COVID-19 in India was reported from Kerala on January 27, 2020 [25]. The GOI responded to the COVID-19 threat by issuing travel advisories and screening at airports in January and later restricting international travel from March 2020 onward [26,27]. As the cases in India started escalating, the GOI called for a Janata curfew on March 23, 2020, and implemented a national lockdown from March 24 to June 7, 2020, for 75 days [28]. The GOI started phased unlocking from June 8, 2020, onward [28]. COVID-19 and SI data for India are available from January 21, 2020, to March 2022. This study used data from March 4, 2020 (when India crossed 10 cumulative cases), to October 24, 2021.

We used 2 publicly available data sources from the Oxford Covid-19 Government Response Tracker (OxCGRT) [29] and the WHO [4] licensed under the Creative Commons Attribution (CC BY) standard.

Ethical clearance for the study was obtained from the institutional review board of the Postgraduate Institute of Medical Education and Research, Chandigarh, India (vide letter INT/IEC/2020/SPL–1594).

The first case of COVID-19 was detected in India on January 27, 2020. It took 46 days for the infection to spread among 100 people. India had 60 COVID-19–infected individuals on March 11, 2020, when the WHO declared COVID-19 a pandemic. True to its nature, COVID-19 started spreading rapidly, and it took only 15 days to reach 1000 cases from 100. Public health experts and policy makers were on high guard after GOI notification—a fact also reflected by the SI and CHI. Table 2 provides details about different milestones regarding the rise in disease burden. The increase in cases followed a linear trajectory during the lockdown stage. After unlocking, India achieved its first peak in September 2020, which was almost 100,000 cases per day. Subsequently, the second peak in May 2021 was characterized by a maximum caseload of ~400,000 cases per day. Figure 1 shows the change in India’s disease burden from January 2020 to October 2021.

The strict implementation of policies by the government to contain the COVID-19 crisis led to a very high value of CHI and SI in the beginning. Figure 2A shows that the CPM was near the x-axis from the beginning till the end of the fourth lockdown. Subsequently, CHI reduced gradually; in contrast, the CPM swiftly increased after the GOI initiated unlocking from July 2020. Figure 2B shows that the DPM also increased steadily with time—the highest during the unlock phase. Although the CHI and SI reduced marginally, there were spikes during the first and second peaks. The CPM and DPM also displayed similar increasing trends with the SI. Figures 2C and 2D depict CPM and DPM changes with SI over time.

Figures 1 and 2A-2D show that both positive (symptomatic) and deaths (mortality) were nonstationary time series. On the surface, the CPM and DPM followed the same trend. However, the final CPM and DPM model were SARIMA (3,2,1)(1,0,1) and ARIMA (1,1,1), respectively. Multimedia Appendix 3 provides details regarding the steps involved in obtaining the final models. There was no improvement in model performance for CPM: the BIC values for both time (BIC 2.42) and the expanded model (BIC 2.46) were almost the same. Similarly, the BIC values for both time (BIC –1.35) and the expanded model (BIC –1.34) models for DPM were the same. Despite the appearance of an inverse relationship between COVID-19 and policy indicators, neither CHI nor SI were significant predictors of CPM and DPM. Table 3 displays the final model’s output and performance indicators: BIC, stationary R², and root mean square error.

We used the CHI and SI data from the Blavatnik School of Government, University of Oxford, response tracker to evaluate and model COVID-19 morbidity and mortality. There are specific critical findings in our study. COVID-19 displayed the typical characteristic of a highly contagious disease that spread rapidly after a slow start. At the beginning of the pandemic in India, cases were geographically scattered and scarce; thus, the CHI and SI were low as neither the GOI issued any guidelines nor did the people adhere to preventive behaviors. There was a sharp increase in cases after unlocking, and India witnessed 2-clear and visible peaks in September 2020 and May 2021; this was in line with a temporary increase in CHI and SI during peaks. The value of CHI and SI gradually decreased; similarly, however, these were consistently higher at values of >80% points.

For intercountry comparisons, the CPM and DPM were calculated and reported, since these are better measures than the raw number of cases. The CPM and DPM linearly increased when stringency was high during the lockdowns and rapidly increased after the government lifted the restrictions. Despite initial indication, both CHI and SI were (rather surprisingly) not significant predictors of COVID-19 morbidity and mortality.

People’s participation is vital to the success of public health interventions. Thus, the GOI tested people’s moods and sentiments with the Janata Curfew (self-restricted curfew) before implementing strict public health measures. After the success of the Janata curfew, the GOI enforced a nationwide lockdown to close schools, parks, transport, offices, and borders to contain the COVID-19 crisis. The daily incidence was relatively flat during the entire lockdown period; the strict response to epidemic control measures reduced daily incidence [26]. However, social control measures mitigate contagious diseases; they do not eradicate them [27,28]. The rate of disease spread increased rapidly after the unlocking process began. Despite the nationwide lockdown, significant intra- and interstate variation was a cause of concern—the majority (~76%) of cases were reported from 10 out of 36 states and union territories in India [29]; the disparity can be partially attributed to constitutional provisions besides factors such as mobility, urbanization, and population density. Moreover, while the central government has the power to make public health laws, the state government develops infrastructure and executes public health policy because health care and hospitals are subject matters in the state list [4].

NPIs such as tests, contact tracing, masking, social distancing, mobility, hygiene, and vaccination form the backbone for controlling contagious diseases. However, in the absence of vaccination, strategic implementation of NPIs helps mitigate the rapid spread of COVID-19 [13-16]. The CHI and SI are more comprehensive than specific mobility indicators—these cover various aspects of NPIs. The CHI and SI attain very high values, and the results are in parallel with those of Ma et al [34], which indicate that countries that implemented nationwide lockdowns in March took strict measures [26]. The high value of CHI and SI during the lockdown reflects public health experts’ and policy makers’ comprehensive and dynamic responses to contain the COVID-19 crisis. Government-enforced lockdowns are significant contributors to restrictions on activities. People also respond to contagious threats by restricting socializing and travelling [35]. Overall, CHI and SI reflected high stringency; this further increased during the peak. Initial investigation reveals a distinct inverse relationship between restriction indicators (CHI and SI indices) and disease burden (morbidity and mortality); this relationship was explicit as the unlocking phase began.

We fitted a time-series model to investigate the relationship between policy and disease indicators. The time-series model with dates explained more than 60% of variability for both CPM and DPM models. The initial investigation indicated the relationship between predictors (CHI and SI) and disease burden (CPM and DPM). However, the expanded inferential models do not increase the models’ performances—neither CHI nor SI contributed significantly explaining the change in CPM and DPM. In other words, stringency measures such as CHI and SI do not explain the change in both COVID-19 morbidity and mortality. The results of our study are in contrast to those of other studies [26,30,31]. However, it may be crucial to note that strict implementation of policies does not necessarily lead to strict compliance [32,33]. A thorough inspection of CHI and SI indicates the government’s proactiveness on different indicators compared to participation. India saw the mass movement of migrant workers [34,36] and multiple attacks on health care workers during the lockdowns [34,37]. Further, COVID-19 was a relatively urban phenomenon; most of India is rural.

A significant strength of our study is that this is the first study to investigate the relationship between policy indicators CHI and SI and epidemiological indicators CPM and DPM. It is the first pan-India study that has used WHO and OxCGRT data to quantify and model COVID-19 transmission. Our study is different from other studies focused on NPIs including mobility or SI but not CHI.

However, a significant limitation of this study, and perhaps most digital epidemiological studies, is the validity and reliability of the data. As already highlighted, CHI and SI are proxy measures that do not reflect strict compliance with the implementation of policies. There may be an inherent bias or reporting error in composite CHI and SI indices at the country level due to within and between heterogeneity at the state and union territory levels. The unweighted indices (CHI and SI) are easy to interpret but make strong assumptions, and a user may obtain different results when using weighted indices. Lastly, correct reporting of the daily incidence of COVID-19 cases depends on a country’s accuracy and testing capacity.

Countries with vast geographical and administrative regions may differ significantly in implementing policies. Therefore, further research is required to develop and validate the metrics to identify whether country- or county-level (state and union territory) metrics are needed. The capacity to test, trace, and treat varies from country to country. Do disease morbidity and mortality reporting reflect the actual scenario? What logical steps and weightage must be given to each region is also a perplexing problem vouching for researchers’ attention. The media plays a vital role in containing public health crises. However, did it aid in signaling by motivating people to adhere to guidelines, or has it added to noise by creating panic among the public? Lastly, are OxCGRT indicators enough to capture people’s participation and policy makers’ execution or do they require appropriate weightage? The aforementioned issues invite attention from health experts and policy makers.

Our study concludes that the policy measures (CHI and SI) do not explain the change in epidemiological indicators (CPM and DPM). The study reiterates our understanding that strict policies do not necessarily lead to better compliance but may overwhelm the overstretched physical health systems. Twenty-first–century problems thus demand 21st-century solutions. The digital ecosystem was instrumental in the timely collection, curation, cloud storage, and data communication. Thus, digital epidemiology can and should be successfully integrated into existing surveillance systems for better disease monitoring, management, and evaluation. An OxCGRT policy metric is a novel innovation to assess government actions during the epidemic, which have the potential for future use and refinement. Therefore, policy makers, public health experts, and programmers must start collaborating to design a hybrid health system that can borrow from the strengths of the existing physical surveillance system and the ever-expanding digital ecosystem.