Intraindividual variability in non-household contacts: a German longitudinal study, April 2020–December 2021

Background Day-to-day variability in social contacts can shape transmission dynamics, yet is rarely quantified. We aimed to quantify intraindividual variability (IIV) in non-household contacts during the COVID-19 pandemic in Germany and assessed its associations with sociodemographic characteristics, vaccination, and policy stringency. Methods We analysed contact survey data with 33 waves (April 2020–December 2021; 7,845 participants; 59,462 observations). Pearson residuals from a mixed-effects negative binomial model were used to calculate the within-person standard deviation (riSD) for participants with ≥ 2 observations, serving as a proxy for IIV. We fitted Gamma regression models with log link to estimate mean ratios (MR) in three analyses: (1) sociodemographic characteristics (n = 6,251), (2) vaccination effects in participants observed both before and after their first dose within ± 100 days (n = 1,203), and (3) policy stringency effects in participants observed under both strong (index ≥ 70) and weak (< 70) conditions (n = 2,446). Results Children and adolescents (0–18 years) showed higher riSD than other age groups (MR = 1.13, 95% CI 1.10–1.16). Households with ≥ 3 members had slightly higher riSD (1.04, 95% CI 1.02–1.06) compared to single-person households. Retired (0.94, 95% CI 0.92–0.96), homemakers (0.88, 95% CI 0.85–0.91), and unemployed individuals (0.91, 95% CI 0.88–0.94) had lower riSD than those who were employed. Vaccination showed no overall association with riSD (0.99, 95% CI 0.93–1.06), though heterogeneity emerged by age and sex. Weaker stringency was strongly associated with higher riSD (1.34, 95% CI 1.31–1.37). Conclusions IIV in non-household contacts was shaped by age, household composition, and employment status, but not by vaccination status. Children and adolescents, living in larger households, and assessments during periods of weaker policy stringency exhibited greater IIV, while retired, housemakers, and unemployed individuals showed lower IIV. Vaccination did not have a consistent effect. Analyses relying solely on average contacts may misrepresent risk when IIV is high. Both models and policies should account for IIV, not just mean contact levels.