Effect of SARS-CoV-2 prior infection and mRNA vaccination on contagiousness and susceptibility to infection

The immunity conferred by SARS-CoV-2 vaccines and infections reduces the transmission of the virus. To answer how the effect of immunity is shared between a reduction of infectiousness and an increased protection against infection, we examined >50,000 positive cases and >110,000 contacts from Geneva, Switzerland (June 2020 to March 2022). We assessed the association between secondary attack rate (i.e. proportion of new cases among contacts) and immunity from natural infection and/or vaccination, stratifying per four SARS-CoV-2 variants and adjusting for index cases and contacts’ socio-demographic characteristics and the propensity of the contacts to be tested. Here we show that immunity protected contacts from infection, rather than reducing infectiousness of index cases. Natural infection conferred the strongest immunity. Hybrid immunity did not surpass recent infection. Although of smaller amplitude, the reduction in infectiousness due to vaccination was less affected by time and by the emergence of new SARS-CoV-2 variants than the susceptibility to infection. These findings support the role of vaccine in reducing infectiousness and underscore the complementary role of interventions reducing SARS-CoV-2 propagation, such as mask use or indoor ventilation.


Testing of contacts
In this subsection we provide additional information concerning the testing of the contacts.Table S1 provide the age category of the contact and of the index for the contact who tested or not during the 10 days following their last encounter with the index case.Figure S1 provides the total number of tests performed by the contacts as a function of the delay in days since their last encounter with their index case.S1: Contact tested or not during the 10 days following their last encounter with the index case, by category of contact and index age The important peak in number of tests observed in Figure S1 at day 7 stems from the fact that from 8th February, 2021 to 31st of December 2021, the quarantine could be shortened with a negative test a day 7 (see methods in the main article).

SAR as a function of the delay
We present here the raw non stratified SAR as a function of the delay between the last encounter between the index and the contact and the positive test of the contact (figure S2) Figure S2: raw SAR as a function of the delay between test result of the index and test result of the contact.
Adjusted analysis with all immune category for index cases  [37.4,44.6]Index Immunity: previously infected < 6months (NVNI) -10.6*** [-14.7,-6.5]-10.2*** [-15.3,-5.1]-5.8 [-12.8,1.3]-4.0 [-10.4,2.5]Index Immunity: previously infected > 6months (NVNI) -7.8** [-13.3,-2.3]-5.0 [-14.0,3.9]-11.6*** [-14.8,-8.4]-4.4** [-7.7,-1.1]Index Immunity: vaccinated < 6 months (NVNI) -5.7* [-10.2,-1.2]-5.5*** [-6.8,-4.2]-6.4*** [-8.1,-4.7]Index Immunity: vaccinated > 6 months (NVNI) -1.3 [-3.5,1.0]-2.6** [-4.5,-0.7]Index Immunity: hybrid (NVNI) 19.3 [-13.3,51.9]-8.5** [-13.7,-3.4]-6.6*** [-9.5,-3.6]Index women (men) -0.3 [-1.2,0.5]0.1 [-1.4  Table S2: Estimated coefficients of the multivariable generalized estimating equation [Confidence Interval], providing the additional effect of each variable on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001 Figure S3: Estimated Adjusted Secondary Attack Rate (circle) and its 95% confidence interval (error bars) stratified per variant (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.Exact values of the estimated can be found in Table 2, and unadjusted estimates are presented in table S2.
Adjusted analysis with interaction term between contact immunity and propensity of contact to perform tests Figure S4 and table S3 present the coefficients of the interaction terms for the multivariable analysis when including an interaction term between the number of tests performed by the contacts the last 3 months and their immunity status.
Figure S4: Adjusted secondary attack rate, displayed per immune status of the contact and number of tests performed by the contact during the 90 days preceding its last encounter with the index case.The estimates are based on 42295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.Table S3: Coefficients [Confidence Interval] of the interaction terms between the immune status of the contact and the number of tests performed by the contacts 3 months prior to the date of the interaction between contact and index, for the adjusted multivariable model.These coefficients provide the additional effect on the reference secondary attack rate (last line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001.NVNI : non vaccinated never infected.

Univariable and multivariable analysis
Some of the variables had a different effect when performing a univariable analysis (see supplementary table S4) or when adjusting for all co-variate.The main differences concerned the effect of symptoms during the alpha wave (+12.3pp in the univariable model and +7.0pp in the adjusted model) and in a lesser extend during the EU1 wave (4.8 pp in the univariable model and 2pp in the adjusted model).This difference was caused by the adjustment for the cough and for demographic variables.
The Immune status of the contact had different effects between the univariable and adjusted model.The effect of being vaccinated less than 6 months ago for the contact was greater in the adjusted analysis, especially for the omicron wave.
The SAR reduction associated with the fact of being previously infected was larger in the adjusted analysis than in the univariable analysis, especially during the EU1, delta and omicron.These differences were mainly due to the adjustment for the propensity of the contacts to perform tests, measured by the number of COVID-19 tests performed during the last 3 months.In a sensitivity analysis including an interaction term between the immune status of the contact and the number of tests performed the last 3 months (see supplementary table S3), we observed a highly significant interaction between these two variables.This interaction effect is mainly driven by the fact that nonvaccinated and vaccinated more than 6 months ago individuals who didn't do any tests in the last 3 months had low SAR compared to those with a higher propensity to test.In contrast, previously infected individuals had very low SAR irrespective of their number of previous tests.
On the opposite, the protection of being previously infected for the index case was lowered in the adjusted analysis when compared with the univariable analysis.The adjustment for symptoms and cough explained 2pp of this difference, and the adjustment immune status of the contact explained the rest of this difference.Adjusted analysis stratified per gender Figure S5a: Estimated Adjusted Secondary Attack Rate (diamond) and its 95% confidence interval (error bars) stratified per variant (EU1, alpha, delta and omicron) considering only men, with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 19,756 index-contact relations for the EU variant, 9,337 for the alpha variant, 12,830 for the delta variant and 9,760 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.Exact values of the estimated can be found in Table 2, and unadjusted estimates are presented in table S2.
Figure S5b: Estimated Adjusted Secondary Attack Rate (diamond) and its 95% confidence interval (error bars) stratified per variant (EU1, alpha, delta and omicron) considering only women, with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 22,520 index-contact relations for the EU variant, 10,938 for the alpha variant, 14,325 for the delta variant and 11,955 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.Exact values of the estimated can be found in Table 2, and unadjusted estimates are presented in table S2.

Sensitivity analysis
Adjusted analysis using complete case analysis

Multivariable analysis
We present here the result of the multivariable analysis but considering only the complete case dataset, which is of 48'468 lines.Figure S6 and table S5 present the result of this analysis.Figure S7 and table S6 present the coefficients of the interaction term for the same analysis when considering an interaction term between the number of tests performed by the contacts the last 3 months and their immunity status.
Figure S6: Adjusted Secondary Attack Rate (SAR) stratified per variant (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis), for the complete case analysis (no missing value for any covariate).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 14,790 index-contact relations for the EU variant, 16,356 for the alpha variant, 14,801 for the delta variant and 2,521 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.S5: Estimated coefficients of the multivariable generalized estimating equation [Confidence Interval] when considering only complete case (i.e.data with no missing covariates), providing the additional effect of each variable on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 14,790 index-contact relations for the EU variant, 16,356 for the alpha variant, 14,801 for the delta variant and 2,521 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Multivariable analysis with interaction term
We present the results of the multivariable analysis for the complete case analysis considering an interaction term between the propensity to perform test and the immunity of the contact in Figure S7 and table S6.
Figure S7: Adjusted secondary attack rate, displayed per immune status of the contact and number of tests performed by the contact during the 90 days preceding its last encounter with the index case, for the complete case (i.e.data with no missing covariates).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 14,790 index-contact relations for the EU variant, 16,356 for the alpha variant, 14,801 for the delta variant and 2,521 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.Table S6: Coefficients [Confidence Interval] of the interaction terms between the immune status of the contact and the number of tests performed by the contacts 3 months prior to the date of the interaction between contact and index, for the adjusted multivariable model when considering complete case (i.e.data with no missing caovraites).These coefficients provide the additional effect on the reference secondary attack rate (last line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.The estimates are based on 14,790 index-contact relations for the EU variant, 16,356 for the alpha variant, 14,801 for the delta variant and 2,521 for the omicron variant.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the three months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 3 month preceding the contact.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001.NVNI : non vaccinated never infected.
Adjusted analysis using the number of tests performed in the 180 days before the index-contact interaction as a proxy for testing In order to test the robustness of the number of tests performed the 3 months (90 days) preceding the last index-contact interaction as a proxy of the contact to get tested, we propose a sensitivity analysis using instead the number of tests performed by the contact the last 180 days instead of the last 90 days, discretized in 4 categories: 0, 1, 2, and 3+.Descriptive statistics of this new variable can be found in table S7, and the result of the multivariable analysis in Figure S8 and table S8.The results of the adjusted analysis with the interaction between immune status and the propensity to test is displayed Figure S9 and table S9.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact., providing the additional effect of each variable on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001 Figure S9: Adjusted secondary attack rate, displayed per immune status of the contact and number of tests performed by the contact during the 180 days preceding its last encounter with the index case, for the complete case (i.e.data with no missing covariates).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.-7.9 [-20.3,4.4]12.8*** [5.5,20.2]Table S9: Coefficients [Confidence Interval] of the interaction terms between the immune status of the contact and the number of tests performed by the contacts 6 months prior to the date of the interaction between contact and index, for the adjusted multivariable model when considering complete case (i.e.data with no missing caovraites).These coefficients provide the additional effect on the reference secondary attack rate (last line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, their immunity status, and an interaction between immunity status and number of tests performed.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001.NVNI : non vaccinated never infected.
Adjusted analysis using periods with variant dominance of more than 90% We test here the sensibility of our analysis to the definitions of the periods of variant dominance.In the main text, the periods of variant dominance were defined as the periods where one VoC is >50% of the variant detected.
We propose here the same analysis using the periods with >90% (see figure below).The results of the multivariable analysis are provided in Figure S10 and table S10. Figure S10: Adjusted Secondary Attack Rate (SAR) stratified per variant when considering period of predominance above 90% of variant share (EU1, alpha, delta and omicron), the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 38,148 index-contact relations for the EU variant, 15,662 for the alpha variant, 21,360 for the delta variant and 17,692 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socioeconomic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.
table S10: Coefficients of the multivariate generalized estimating equation [Confidence Interval]-when considering period of predominance of VoCs above 90%, providing the additional effect on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 38,148 index-contact relations for the EU variant, 15,662 for the alpha variant, 21,360 for the delta variant and 17,692 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Misclassification of community cases
In order to control for potential misclassification of community cases, we reproduced the multivariable analysis but considering that the contact is positive if it is tested positive between 4 and 10 days after the last encounter with its index case.This sensitivity analysis has for objective to test for a potential effect of index-contact case misclassification: in the main analysis, we suppose that the index case is the one receiving the positive test the first, and the positive contact the one receiving the positive test after.But the reality could be the opposite.By considering contact positive only with a delay of at least 3 days, we lower this risk of misclassification.The results are presented in Figure S11 and table S11.The reference SAR is of course lower than in the main study, because less contacts comply with the new restrictive positivity criteria, but the relative influence of the factor remains similar to the main analysis.
Figure S11: Adjusted Secondary Attack Rate (SAR) stratified per variant when considering that the contact is positive if it is tested positive between 4 and 10 days after the last encounter with its index case, stratified per variant dominance (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected between 4 and 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.
Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.0.3 [-1.3,1.8]0.0 [-1.8,1.9]1.1 [-0.2,2.5]-0.3 [-1.6,1.0]table S11: Coefficients of the multivariate generalized estimating equation [Confidence Interval] when considering that the contact is positive if it is tested positive between 4 and 10 days after the last encounter with its index case, providing the additional effect on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected between 4 and 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Misclassification of primary case
In order to control for potential misclassification of primary case, we reproduced the multivariable analysis but restricting the data to household with only one contact.This sensitivity analysis has for objective to test for a potential effect misclassification of primary case: in the main analysis, we suppose that the index case is the one receiving the positive test the first, and the positive contact the one receiving the positive test after, when the contact could actually be contaminated by a third party, contact of the index case.By restricting the analysis to households with only one contact, we avoid this risk of misclassification, and obtain 8669 index-contact relations.The results are presented in Figure S12 and table S12.
Figure S12: Estimated Adjusted Secondary Attack Rate stratified per variant when restricting the data to household with only one contact, stratified per variant dominance (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected before 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 2,783 index-contact relations for the EU variant, 2,016 for the alpha variant, 2,266 for the delta variant and 1,604 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.] when restricting the data to household with only one contact, providing the additional effect on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates, p values and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected before 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 2,783 index-contact relations for the EU variant, 2,016 for the alpha variant, 2,266 for the delta variant and 1,604 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Misclassification of tertiary case
In order to control for potential misclassification of tertiary case, we reproduced the multivariable analysis but considering only contacts that were set in quarantine at maximum de day after their last contact with the index case.This sensitivity analysis has for objective to test for a potential misclassification of community case: in the main analysis, we suppose that the contacts are infected by the index cases, whereas they could by infected by other persons.By considering contact set in quarantine right after their last contact with the index, we lower this risk.The obtained dataset is composed of 38,277 index-contact relations.The results are presented in Figure S13 and table S13.
m Figure S13: Adjusted Secondary Attack Rate (SAR) stratified per variant when restricting the data to contacts set in quarantine at maximum one day after their last encounter with the index case, stratified per variant dominance (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected before 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 31,429 index-contact relations for the EU variant, 16,699 for the alpha variant, 7,838 for the delta variant and 3,260 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.table S13: Estimated coefficients of the multivariate generalized estimating equation [Confidence Interval] when restricting the data to contacts set in quarantine at maximum one day after their last encounter with the index case, providing the additional effect on the reference secondary attack rate (first line), for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected before 10 days after the last encounter with its index case, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 31,429 index-contact relations for the EU variant, 16,699 for the alpha variant, 7,838 for the delta variant and 3,260 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Figure S1 :
Figure S1: Total number of tests performed by the contacts as a function of the delay in days since their last encounter with their index case

Figure S8 :
Figure S8: Adjusted Secondary Attack Rate (SAR) stratified per variant (EU1, alpha, delta and omicron), with the reference value indicated with a vertical line (reference of each covariate is indicated in bold in parenthesis).Estimates and confidence intervals are produced by a generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42,295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant.Estimates are adjusted for the index case gender, age, obesity, presence of symptoms, presence of cough, immunity status, neighbourhood socio-economic condition, vulnerability and type of living; the link between the index case and its contacts, and for the contact persons, their gender, age, number of tests performed the six months before the contact date with the index case, and their immunity status.The reference index case -contact relation of this multivariate analysis is the contact between two men of age below 65 living at the same place, the index being not vaccinated not infected (NVNI), not obese, living in a wealthy neighbourhood and being not a vulnerable person, living in a housing building, and the contact person being a NVNI adult men who performed one SARS-CoV-2 test during the last 6 month preceding the contact.

Table S4 :
Coefficients [Confidence Interval]of the univariable generalized estimating equations linear regression with robust standard errors predicting a binary outcome indicating if the contact was infected by the index or not, using the index cases as cluster and an exchangeable correlation structure.The estimates are based on 42295 index-contact relations for the EU variant, 20,311 for the alpha variant, 27,260 for the delta variant and 21,808 for the omicron variant..Each coefficient yields the additional effect of the variable on the secondary attack rate for the 4 periods of dominance of the variants EU1, alpha, delta, and omicron.The reference category for each categorical variable is indicated in bold in parenthesis.The left column indicates if the variable concerns the index case, the contact, or their relation.p values are indicated with *. *: 0.01<p<0.05,**:0.001<p<0.01,***: p<0.001

Table S7 :
descriptive statistics of the Number of tests performed by the contact during the last 6 months preceding their last encounter with the index case, stratified by variant.
table S8: Estimated coefficients of the multivariable generalized estimating equation [Confidence Interval] periods were:

table S12 :
Coefficients of the multivariate generalized estimating equation [Confidence Interval