The results from this large multicentric study indicate that both AZD1222/Covishield and BBV152/Covaxin significantly reduce the risk of severe COVID-19 and against the Delta variant among the Indian population aged 45 years and above, more so with two doses, albeit higher for AZD1222/Covishield than BBV152/Covaxin. Highest reduction in risk of severe COVID-19 was documented for 6-8 week interval between the two doses.

Our estimates of vaccine effectiveness were expectedly lower than the efficacy estimates from vaccine trials. BBV152/Covaxin was introduced based on high safety, tolerability and immune responses among the Indian participants ( Ella et al., 2021a, 2021b). Subsequently, the efficacy in the phase III trial was estimated to be 93% against severe COVID-19 (Ella et al., 2021c) Real world estimates of VE should be considered in the context of programmatic issues such as storage and cold chain maintenance, and off-schedule and incomplete delivery of doses (World Health Organization, 2021). Besides, the field performance of vaccines were influenced by newly emerging SARS-CoV-2 variants. The VE of two doses of Oxford-ChAdOx1-S against severe infection with the Delta variant in United Kingdom was 67% (Lopez Bernal et al., 2021). Indian studies indicated reduced neutralization capability of BBV152/Covaxin and AZD1222/Covishield vaccine against Delta variant (Yadav et al., 2021; Sapkal et al., 2021). Our study was conducted when B.1.617 lineages were dominating in India (INSACOG, 2021).

Our results are consistent with the VE reports of either AZD1222/Covishield or BBV152/Covaxin against hospitalization to be 77% for full and 70% for partial vaccination among healthcare workers in South India (Victor et al., 2021). Our VE estimates for AZD1222/Covishield against severe diease are higher than 67% (95% CI: 44-81%) for moderate to severe disease and 76% (95% CI: 37-89%) for supplemental-oxygen-therapy among a chort study of hospital employees in New Delhi (Satwik et al., 2021). The disparity could be on account of study design, different outcome measures between the two studies as also the fact that hospital employees although at higher risk for SARS-CoV-2 infection are also more likely to be vaccinated.

The interval between the two doses can also influence the VE in programmatic condition. We estimated a maximum effectiveness for a 6-8 week interval for AZD1222/Covishield, with a gradual decline beyond 12 weeks. Unlike our observation, other published studies of Oxford-ChAdOx1-S indicated higher efficacy or immunogenicity for >12 weeks interval between two vaccine doses (Flaxman et al., 2021; Voysey et al., 2021). During the high transmission period, when we conducted this study, asymptomatic or known SARS-CoV-2 infection few weeks prior to the first or second dose could have boosted the vaccine-induced immunogenicity manifesting as improved vaccine performance at a shorter interval. Our finding has programmatic implications on rolling out of the vaccination to reach the maximum eligible Indian population at the earliest, where after an initially interval of four weeks for both BBV152/Covaxin and AZD1222/Covishield, the interval was increased to 12-16 weeks for AZD1222/Covishield (Perappadan, 2021). The policy on interval between the two doses vary across countries and needs further evidence towards its standardization (Bobdey et al., 2021).

We highlight the strengths of our study. The study sites included 11 hospitals spread across all parts of the country, thereby strengthening the generalisability of VE estimates. We used a robust case-control study design along with an approach to multiple logistic regression guided by the application of causal framework using directed acyclic graphs to identify and adjust for appropriate known and measured confounders. Further, we used a random effects model for multiple logistic regression to take into account the variation in VE by hospital sites. Cases and controls in each study site were recruited from the same hospital. Hence, they are likely to belong to the same catchment area of the particular hospital and hence coming from the same source population. This minimizes the chance of selection bias influencing VE estimates. Although we could not achieve the desired sample size, our study had adequate power to estimate VE for both the vaccines separately. Such power could be attributed to substantially higher vaccine coverage than the assumed coverage for calculating the sample size. Molecular characterization of SARS-CoV-2 enabled us to estimated the VE specifically for Delta variant and its sub-lineages.

Our study had several limitations. Firstly, misclassification bias could have affected the VE estimates in several ways. (1) It is possible that vaccinated individuals could have had higher risk of COVID-19 due to (a) potential transmission during travel for vaccination, (b) crowding at the vaccination centres, and (c) risky behaviors post-vaccination due to the self-perceived vaccine protection. This could have led to underestimation of the VE. (2) Low sensitivity or specificity of rRT-PCR testing and asymptomatic status of COVID-19 could have led to differential misclassification of the control status (Tahamtan and Ardebili, 2020). To reduce such bias, we did confirm the rRT-PCR status among negative controls after a week. However, we could not confirm all the negative controls and hence, we could have underestimated the VE. (3) The status of IgG used as a surrogate marker for recent infection, though done, could not differentiate the antibodies generated by the vaccine or the infection. Accordingly, we could have misclassified the infection status of those infected prior to joining the study but not tested positive at the time of enrolment and thus leading to underestimation of VE. (4) We anticipated differential misclassification about the interval between two doses of vaccine, as cases might have recalled vaccination dates better than controls- thereby causing a biased estimate of VE by dosing interval, in either direction. (5) The vaccination status, including dates thereof, was based on recall by nearly half of the study participants, and more so among the cases than the controls. This could have led to differential misclassification in vaccination status resulting in VE estimate biased in either direction. On sensitivity analysis of VE estimate by mode of ascertainment of vaccination status, record vs. recall, we found that there was a difference in VE for partial vaccination with AZD1222/Covishield and for full vaccination with BBV152/Covaxin (Supplementary Table S4). Secondly, prior SARS-CoV-2 infection and conferred immunity thereof, could have influenced the VE estimates (Lipsitch et al., 2020). Such prior infection could reduce the chance of re-infection. While being aware of a prior infection can reduce the likelihood of vaccination, unknown prior infection (e.g., asymptomatic) is unlikely to have influenced the decision to vaccinate. Although the bias in estimating VE due to unknown prior infection is reported to be minimal in case-control studies, it is not known whether this could be true during periods of high transmission (World Health Organization, 2021). Thirdly, the sample size could have influenced the VE estimates. Four-fifth of the study participants had received AZD1222/Covishield and hence recruited study size was inadequate for VE estimates for BBV152/Covaxin. Estimation of VE for interval of more than 9 weeks was under-powered and hence needs to be interpreted cautiously. Further, it did not allow us to statistically compare VE across intervals. WHO suggets that any useful COVID-19 vaccine should have VE estimates with lower bound of the 95% CI above 50% (World Health Organization, 2021). Fourthly, although we adjusted the odds ratio for known and relevant confounders identified on the basis of causal framework analysis, the bias-indicator measuring VE among recently vaccinated individuals (within 6 and 14 days after one dose; results not shown) indicates that our adjusted odds ratio and the resulting VE estimates may remain biased due to unknown, unmeasured and/or residual confounding (Hitchings et al., 2021). However, with moderate to high background seroprevalence, hitherto unknown prior infection may result in protection from a single dose of vaccine during a shorter period post-vaccination when compared to the efficacy trial results (Saadat et al., 2021). Finally, the external validity of the VE findings beyond 45 years needs to be considered cautiously. The younger adults (18-44 years) had recently become eligible for vaccination and hence, we expect similar VE results in the younger age group as well.

Our findings highlight significant real-word protection with two vaccine doses against severe COVID-19 and specifically against the currently dominant Delta variant in India. The substantial effectiveness of only one dose, more so for AZD1222/Covishield, supports the policy decision from a public health perspective to initially maximize coverage with single -dose in the country. Our study finding for effectiveness of single dose of AZD1222/Covishield is in contrast to other studies that report either lack of effectiveness, (Satwik et al., 2021) or significant but lower effectiveness (Victor et al., 2021). This could be due to difference in study design, including difference in the primary outcome measure of disease severity as well as potential difference in background seroprevalence among the study population of the different studies referenced above. Vaccine effectiveness below 100% suggests the possibility of severe disease among vaccinated and thus further likely transmission in the absence of adequate control measures (Lee et al., 2022; Pritchard et al., 2021).

Our finding of high variation in the Delta lineage across the country is suggestive of fast mutations in Delta due to its immune escape ability in host genome. Further, it has been suggested that the dominance of Delta variant and its sub-lineages with higher transmissibility and potential for immune escape can make the task of achieving a robust protection against infection even with near universal vaccination coverage much more difficult. (Gupta et al., 2021; Lazarevic et al., 2021) Nevertheless, acceleration of the two-dose vaccination coverage can be critical for an effective and timely reduction in the burden of severe COVID-19 in India.

The authors acknowledge the contributions of all the study participants, and the clinical and laboratory staff from all the collaborating institutions. We acknowledge Michaelraj E, Amanda Rozario GA, Gayathri K, Fathima Shireen and Mohana Balan Parivallal for their support in data management. We appreciate the technical inputs towards the design and implementation of the study provided by the members of the Epidemiology and Surveillance Working Group of the ICMR, constituted by the COVID-19 National Task Force of Government of India and the WHO India team.

References

Afifi A, Clark V, May S. Regression analysis with multicollinearity (4th ed.). Boca Raton, FL: Chapman & Hall/CRC 2004.

Bobdey S, Kaushik SK, Menon AS. The conundrum of two-dose interval of ChAdOx1 nCOV-19 corona virus vaccine: Way ahead. Med J Armed Forces India 2021;77:S250–3. https://doi.org/10.1016/j.mjafi.2021.05.024.

Ella R, Reddy S, Blackwelder W, Potdar V, Yadav P, Sarangi V, et al. Efficacy, safety, and lot to lot immunogenicity of an inactivated SARS-CoV-2 vaccine (BBV152): a, double-blind, randomised, controlled phase 3 trial. Lancet 2021a. https://doi.org/10.1101/2021.06.30.21259439.

Ella R, Reddy S, Jogdand H, Sarangi V, Ganneru B, Prasad S, et al. Safety and immunogenicity of an inactivated SARS-CoV-2 vaccine, BBV152: interim results from a double-blind, randomised, multicentre, phase 2 trial, and 3-month follow-up of a double-blind, randomised phase 1 trial. The Lancet Infectious Diseases 2021b;21:950–61.

Ella R, Vadrevu KM, Jogdand H, Prasad S, Reddy S, Sarangi V, et al. Safety and immunogenicity of an inactivated SARS-CoV-2 vaccine, BBV152: a double-blind, randomised, phase 1 trial. The Lancet Infectious Diseases 2021c;21:637–46. https://doi.org/10.1016/S1473-3099(20)30942-7.

Flaxman A, Marchevsky NG, Jenkin D, Aboagye J, Aley PK, Angus B, et al. Reactogenicity and immunogenicity after a late second dose or a third dose of ChAdOx1 nCoV-19 in the UK: a substudy of two randomised controlled trials (COV001 and COV002). The Lancet 2021;398:981–90.

Ghosh S, Shankar S, Chatterjee Kaustuv, Chatterjee Kaushik, Yadav AK, Pandya K, et al. COVISHIELD (AZD1222) VaccINe effectiveness among healthcare and frontline Workers of INdian Armed Forces: Interim results of VIN-WIN cohort study. Med J Armed Forces India 2021;77:S264–70.

Gupta N, Kaur H, Yadav PD, Mukhopadhyay L, Sahay RR, Kumar A, et al. Clinical Characterization and Genomic Analysis of Samples from COVID-19 Breakthrough Infections during the Second Wave among the Various States of India. Viruses 2021;13:1782.

Hitchings MDT, Lewnard JA, Dean NE, Ko AI, Ranzani OT, Andrews JR, et al. Use of recently vaccinated individuals to detect bias in test-negative case-control studies of COVID-19 vaccine effectiveness. medRxiv 2021:2021.06.23.21259415.

INSACOG. Bulletin. 28 June 2021 New Delhi: Department of Biotechnology, Ministry of Science and Technology, Government of India. Available from: https://dbtindia.gov.in/sites/default/files/INSACOG%20WEEKLY%20BULLETIN%20June%2030.pdf

Jaiswal A, Subbaraj V, Vivian Thangaraj JW, et al.. COVID-19 vaccine effectiveness in preventing deaths among high-risk sgroups in Tamil Nadu, India. The Indian Journal of Medical Research 2021;153. https://doi.org/10.4103/ijmr.ijmr_1671_21.

Lazarevic I, Pravica V, Miljanovic D, Cupic M. Immune Evasion of SARS-CoV-2 Emerging Variants: What Have We Learnt So Far? Viruses 2021;13:1192.

Lee LYW, Rozmanowski S, Pang M, Charlett A, Anderson C, Hughes GJ, et al. Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infectivity by Viral Load, S Gene Variants and Demographic Factors, and the Utility of Lateral Flow Devices to Prevent Transmission. Clin Infect Dis 2022;74:407–15.

Lipsitch M, Kahn R, Mina MJ. Antibody testing will enhance the power and accuracy of COVID-19-prevention trials. Nat Med 2020;26:818–9.

Lopez Bernal J, Andrews N, Gower C, Gallagher E, Simmons R, Thelwall S, et al. Effectiveness of Covid-19 Vaccines against the B.1.617.2 (Delta) Variant. New England Journal of Medicine 2021;385:585–94.

Ministry of Health and Family Welfare. Frequently Asked Questions – COVID vaccination. Government of India 2021a. [cited 2021 Sep 17]. Available from: https://www.mohfw.gov.in/covid_vaccination/vaccination/faqs.html

Ministry of Health and Family Welfare. Co-WIN dashboard 2021b. [cited 2021 Oct 20]. Available from: https://dashboard.cowin.gov.in/

O'Neill RT. On sample sizes to estimate the protective efficacy of a vaccine. Stat Med 1988;7:1279–88.

Perappadan BS. Gap between two doses of Covishield extended to 12-16 weeks, says government. The Hindu 2021.

Pritchard E, Matthews PC, Stoesser N, Eyre DW, Gethings O, Vihta K-D, et al. Impact of vaccination on new SARS-CoV-2 infections in the United Kingdom. Nat Med 2021;27:1370–8.

Saadat S, Rikhtegaran Tehrani Z, Logue J, Newman M, Frieman MB, Harris AD, et al. Binding and Neutralization Antibody Titers After a Single Vaccine Dose in Health Care Workers Previously Infected With SARS-CoV-2. JAMA 2021;325:1467–9.

Sapkal GN, Yadav PD, Sahay RR, Deshpande G, Gupta N, Nyayanit DA, et al. Neutralization of Delta variant with sera of CovishieldTM vaccinees and COVID-19-recovered vaccinated individuals. J Travel Med 2021;28:taab119.

Satwik R, Satwik A, Katoch S, Saluja S. ChAdOx1 nCoV-19 effectiveness during an unprecedented surge in SARS COV-2 infections. Eur J Intern Med 2021;93:112–3.

Serum Institute of India. Product Insert: ChAdOx1 nCoV- 19 Corona Virus Vaccine (Recombinant) – COVISHIELD. 2021. Available from: https://www.seruminstitute.com/pdf/covishield_ChAdOx1_nCoV19_corona_virus_vaccine_insert.pdf

Tahamtan A, Ardebili A. Real-time RT-PCR in COVID-19 detection: issues affecting the results. Expert Rev Mol Diagn 2020:1–2.

The Hindu. India approves COVID-19 vaccines Covishield and Covaxin for emergency use. 2021 Jan 3 [cited 2021 Sep 17]; Available from: https://www.thehindu.com/news/national/drug-controller-general-approves-covishield-and-covaxin-in-india-for-emergency-use/article33485539.ece

Victor PJ, Mathews KP, Paul H, Mammen JJ, Murugesan M. Protective Effect of COVID-19 Vaccine Among Health Care Workers During the Second Wave of the Pandemic in India. Mayo Clin Proc 2021;96:2493–4.

Voysey M, Clemens SAC, Madhi SA, Weckx LY, Folegatti PM, Aley PK, et al. Single-dose administration and the influence of the timing of the booster dose on immunogenicity and efficacy of ChAdOx1 nCoV-19 (AZD1222) vaccine: a pooled analysis of four randomised trials. The Lancet 2021;397:881–91.

Westreich D, Greenland S. The Table 2 Fallacy: Presenting and Interpreting Confounder and Modifier Coefficients. American Journal of Epidemiology 2013;177:292–8.

World Health Organization. Evaluation of COVID-19 vaccine effectiveness. Interim Guidance 2021. https://www.who.int/publications-detail-redirect/WHO-2019-nCoV-vaccine_effectiveness-measurement-2021.1 (accessed September 17, 2021).

World Health Organization. Clinical management of COVID-19. Interm guidance. 2020.

Yadav PD, Sapkal GN, Ella R, et al. Neutralization of Beta and Delta variant with sera of COVID-19 recovered cases and vaccinees of inactivated COVID-19 vaccine BBV152/Covaxin. J Travel Med. 2021 Oct 11;28(7):taab104.