Effectiveness of BNT162b2 and ChAdOx1 against SARS-CoV-2 household transmission - a prospective cohort study in England

Background: The ability of SARS-CoV-2 vaccines to protect against infection and onward transmission determines whether immunisation can control global circulation. We estimated effectiveness of BNT162b2 and ChAdOx1 vaccines against acquisition and transmission of the Alpha and Delta variants in a prospective household study in England. Methods: Adult index cases in the community and their household contacts took oral-nasal swabs on days 1, 3 and 7 after enrolment. Swabs were tested by RT-qPCR with genomic sequencing conducted on a subset. We used Bayesian logistic regression to infer vaccine effectiveness against acquisition and transmission, adjusted for age, vaccination history and variant. Findings: Between 2 February 2021 and 10 September 2021 213 index cases and 312 contacts were followed up. After excluding households lacking genomic proximity (N=2) or with unlikely serial intervals (N=16), 195 households with 278 contacts remained of whom 113 (41%) became PCR positive. Delta lineages were 4.6 times (95% Credible Interval: 1.5 - 20.1) more transmissible than Alpha; contacts older than 18 years were 2.0 times (1.4 - 3.3) more likely to acquire infection than children. Effectiveness of two doses of BNT162b2 against transmission of Delta was 31% (-3%, 61%) and 42% (14%, 69%) for ChAdOx1, similar to their effectiveness for Alpha. Protection against infection with Alpha was higher than for Delta, 71% (12%,95%) vs 24% (-2%, 64%) respectively for BNT162b2 and 26% (-39%, 73%) vs 14% (-5%, 46%) respectively for ChAdOx1. Interpretation: BNT162b2 and ChAdOx1 reduce transmission of the Delta variant from breakthrough infections in the household setting though their protection against infection is low. Funding: This study was funded by the UK Health Security Agency (formerly Public Health England) as part of the COVID-19 response.

Introduction 38 secondary attack rates (SARs). For such the SARs were sampled during the MCMC sampling, for 107 each combination of variant and case and contact vaccine status (1 or 2 doses for each product) 108 and age group, against a baseline of that case-contact pair and variant in the absence of any 109 vaccination. 110 Lineage 111 cases and contacts, with pseudo-absences generated to simulate the time of infecting exposure. 135 Comparison is made for each combination of vaccine product, number of doses, and variant 136 against the corresponding unvaccinated group. Details of this modelling can be found in the 137

Appendix. 138
Identification of non-household transmission 139 As per the study design, the index case for each household was by default considered to be the 140 individual who presented for Pillar 2 testing. To reduce the risk of misclassification bias we 141 excluded from the analyses all households where both the index case and an infected household 142 contact were symptomatic and the index case's symptoms appeared more than two days after the 143 contact's symptoms. 144 To further reduce the potential for misclassification bias, a phylogenetic approach was used to 145 identify apparent secondary cases in the household who were in fact infected elsewhere. If none 146 of the sequences from a contact clustered with at least one of the sequences from the household's 147 index case, then this was considered as evidence for an infection acquired outside of the 148 household; therefore, the contact was excluded from the downstream analysis. Details of the 149 phylogenetic approach can be found in the Appendix. 150 Role of funding source 151 The study sponsors had no role in the collection, analysis, and interpretation of data; in the 152 writing of the report; and in the decision to submit the paper for publication 153

154
By September 10th, 2021, a total of 213 index cases and 312 contacts had been recruited and met 155 the criteria for inclusion at that time. Two contacts were removed due to lack of genomic 156 proximity (outlined below), which resulted in the removal of each of their households as there The phylogeny provided evidence that in two households the contact of the recruited index case 195 had acquired infection elsewhere (Figure 1, households HH002 and HH007). Five households 196 that did not form unique clusters in the phylogeny did not meet the exclusion criteria: in two a 197 sequence from an index case did not cluster with the remaining household sequences but another 198 sequence from the same index case did (HH004 and HH006) while the other three households 199 did not have sufficient bootstrap support to be a part of a cluster (HH001, HH003, and 200 HH005). Of the remaining households, 72 (91%), formed unique, household-specific clusters 201 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

212
Either one or two doses of BNT162b2 provide a protective effect against infection from a 213 symptomatic index case with Alpha variant SARS-CoV-2 with a vaccine effectiveness of 53%, 214 (95% credible interval: 7%, 83%) and 71% (95% CrI: 12%, 95%), respectively (Table 1, Figure A4). against infection with the Delta variant was lower than against Alpha and was similar to the 217 effectiveness offered by ChAdOx1 to either variant (Table A5) which, after two doses, had 218 effectiveness against Alpha of 26% (-39%, 73%) and against Delta of 14% (-5%, 46%). 219 We estimate that the effectiveness of one and two doses of BNT162b2 against onward 220 transmission if infected with the Alpha variant was 26% (-11%, 54%) and 57% (5%, 85%) and for 221 Delta variant one and two doses reduce transmission by 9% (-16%, 49%) and 37% (4%, 65%). RRs 222 for the protective effect of BNT162b2 over ChAdOx1 for one and two doses of against both 223 Alpha and Delta variants indicate that at 95% credibility there is no difference between the 224 effectiveness of the two vaccine products (Table A5). Specifically, two doses of ChAdOx1 reduce 225 transmission from an Alpha variant case by 35% (-26%, 74%) and from a Delta variant case by 226 42% (14%, 69%). 227 Table 1: Median vaccine effectiveness (VE) and 95% credible intervals for infection protection 228 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. has an SAR of 30% (13%, 57%). SARs for each combination of contact and case age, vaccine 245 history and variant lineage are given in the appendix (Figures A5, A6). Notably, the reduced 246 susceptibility to infection of (unvaccinated) under-18s results in SARs which are no greater than 247 those seen in adult contacts who have received two doses of ChAdOx1. 248

Sensitivity analysis
249 Sensitivity analysis was conducted by including the 16 index case-contact pairs with serial 250 intervals less than -2 days. This did not qualitatively change our results (Table A4). The absence 251 of informative priors on the protective vaccine effects against infection led some of the vaccine 252 effectiveness against infection in our study to be re-attributed to effectiveness against onward 253 transmission or to age effects. Figure A4  was slightly less pronounced. 264 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  was however low; 14% (-5%, 46%) and 24% (-2%, 64%) after two doses of ChAdOx1 and 274 BNT162b2 respectively. This is lower than that estimated from cases presenting for Pillar 2 275 testing in the community for which the effectiveness of two doses of ChAdOx1 against 276 symptomatic infection is estimated as 67.0% (61.3%, 71.8%) and 88.0% (85.3%, 90.1%) for 277 BNT162b2. 18 Effectiveness against acquisition of an Alpha infection in the household was 278 substantially higher in our study than that against Delta but still lower than that estimated from 279 Pillar 2 community testing. The lower protection against acquisition in the household likely 280 reflects the prolonged and intense exposure that occurs in this setting. Similarly, although the 281 effectiveness estimates against transmission were moderate at 42% (14%, 69%) and 31% (-3%, 61%) 282 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted November 24, 2021. ; https://doi.org/10.1101/2021.11.24.21266401 doi: medRxiv preprint after 2 doses of ChadOx1 and BNT16b2 respectively, the protective effect in those with 283 breakthrough infections may be higher in the community where exposure is less intense and of 284 shorter duration. The reduction in duration of PCR positivity in breakthrough infections 285 (average of 4 days shorter for the Delta variant for those infected after two doses of BNT162b2 286 and around 2-3 days for ChAdOx1) will also have more of an impact in the community than in 287 the household setting where generation times between infections are short -around 3.5 days for 288 the Delta variant. 19 Our household contacts were actively followed up with repeated swabbing 289 and showed the high secondary attack rates that occur in this setting; 81% for Delta infections in 290 unvaccinated households but that reduced to 25-40% in households where both index case and 291 contacts were fully vaccinated. 292 Our finding of a moderate level of protection against onward transmission from fully vaccinated 293 individuals ,with either vaccine and against either variant, is in apparent contrast to a study that 294 similarly followed up contacts reported by the UK test and trace system prospectively, about 90% 295 of whom were in the same household as the index case 20 . The study estimated a moderate effect of 296 vaccination against infection but no difference in secondary attack rates with the delta variant 297 between fully vaccinated and unvaccinated index cases (24% and 23% respectively). However, 298 such estimates were neither controlled for age nor vaccination status of the contact. Interestingly 299 only 4 out of 17 (24%) unvaccinated contacts were infected by fully vaccinated index cases, 300 whereas 8 out of 20 (40%) unvaccinated contacts were infected by unvaccinated index cases; a 301 reduction in transmission of 41% albeit based on very small numbers. Vaccine effectiveness 302 against onward transmission of 40 to 80% has been suggested by several retrospective 303 observational studies using either information on the household structure 4 or contact tracing 5 6 304 in combination with routine national COVID-19 notification systems to estimate reductions in 305 secondary attack rates from breakthrough infections. While observational studies are prone to 306 the confounded age and vaccine history effects which arose due to vaccine product prioritisation 340 and were exacerbated by low counts for case-contact vaccine history combinations. 341 Our findings provide robust evidence from a prospective study that vaccination with either 342 BNT162b2 or ChAdOx1 can help to substantially reduce, but not completely prevent, 343 household transmission with SARS-CoV-2. This highlights the importance of vaccines to limit 344 circulation of SARS-CoV-2 particularly in close and prolonged contact indoor settings. The 345 effectiveness of booster doses to further enhance protection against transmission will need to be 346 evaluated to better understand the extent to which we can rely on vaccination for the control of Data sharing 370 All analysis code is available from https://github.com/cmmid/hhSAR. The data necessary to 371 replicate results is available from the authors on request, subject to a data sharing agreement. 372

373
The household surveillance protocol was approved by the UKHSA Research Ethics and 374 Governance Group as part of the portfolio of the UKHSA's enhanced surveillance activities in 375 response to the pandemic. Oral informed consent for sampling and follow up was obtained by 376 the nurses from household members who were free to decline to participate in the surveillance at 377 any time. Consent for children was obtained by a parent or legal guardian. Only anonymised data 378 were provided to non-UKHSA authors. 379