¹Martin-Luther-Universität Halle-Wittenberg, Halle (Saale), Germany
²Universität Münster, Münster, Germany

Text

Introduction: Public health responses to epidemics rely heavily on non-pharmaceutical interventions (NPIs). While essential for reducing disease transmission, measures such as household quarantines can impose significant socio-economic burdens. Existing research has established that household characteristics significantly influence transmission dynamics. For instance, studies by Dönges et al. [1] and Liu et al. [2] highlight the importance of household size. The role of household composition has been explored by Endo et al. [3] and Møgelmose et al. [4], while work from Tseng et al. [5] discusses the specific impact of children in households. This study explores the potential of targeted intervention strategies to achieve a better balance between public health benefits and the resulting societal costs. In particular, we examine whether taking into account household characteristics, such as size and composition, allows for more efficient interventions compared to non-selective approaches.

Methods: We developed an “Infection Contribution” (IC) metric, which quantifies the relative importance of different household types in driving overall infection dynamics. This metric traces the involvement of particular household types over entire infection chains and shows how they are involved in spreading infections. Using the agent-based German Epidemic Microsimulation System (GEMS), we simulate a wild-type COVID-19-like epidemic in the Saarland region of Germany. We evaluate 72 different targeted 14-day household quarantine strategies. These strategies specifically target households based on their size (e.g., 2+, 3+, 4+, 5+, or 6+ persons) and composition (e.g., with/without schoolchildren, with/without workers, number of schoolchildren/workers). We compare their effectiveness in reducing the basic reproduction number (R0) with their associated societal costs, measured in total quarantine days, lost school days, and lost workdays.

Results: Simulations demonstrate that targeted interventions based on household characteristics can provide more efficient alternatives to non-selective strategies. We identify Pareto-optimal quarantine strategies that achieve significant R0 reduction while minimizing societal costs. Specifically, strategies targeting households with specific compositions, such as the presence of schoolchildren or workers who increase the household's out-of-household contacts, show improved efficiency. For instance, quarantining households without schoolchildren reduces R0 by 0.74 without incurring any lost school days. Eleven out of the 25 Pareto-optimal scenarios focus on households of various sizes that explicitly exclude schoolchildren. Conversely, targeting households with two or more workers was the most effective strategy short of quarantining everyone, achieving an R0 reduction by 1.12. Our analysis also confirms that larger households have a disproportionately higher contribution to infection chains relative to their population share.

Conclusion: Household structure, including size and the presence of schoolchildren or working adults, significantly influences infection dynamics, suggesting that targeting these characteristics can improve intervention efficiency and reduce societal costs. However, our results highlight important limitations. The optimal strategy is highly context-dependent; repeating the analysis for a metropolitan area such as Berlin yielded significantly different efficiency frontiers. Therefore, while targeted interventions are a promising approach to improve epidemic response efficiency and reduce socioeconomic costs compared to non-selective interventions, their efficient design requires careful consideration of the relevant epidemiological and demographic context. Their practical implementation, furthermore, presents additional challenges.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

This work has been submitted to the Winter Simulation Conference 2025 and is currently under review.

References

[1] Dönges P, Götz T, Kruchinina N, Krüger T, Niedzielewski K, Priesemann V, Schäfer M. SIR Model for Households. SIAM Journal on Applied Mathematics. 2024;84(4):1460–1481.
[2] Endo A, Uchida M, Kucharski AJ, Funk S. Fine-scale family structure shapes influenza transmission risk in households: Insights from primary schools in Matsumoto city, 2014/15. PLoS Comput Biol. 2019 Dec 26;15(12):e1007589. DOI: 10.1371/journal.pcbi.1007589
[3] Liu P, McQuarrie L, Song Y, Colijn C. Modelling the impact of household size distribution on the transmission dynamics of COVID-19. J R Soc Interface. 2021 Apr;18(177):20210036. DOI: 10.1098/rsif.2021.0036
[4] Møgelmose S, Vijnck L, Neven F, Neels K, Beutels P, Hens N. Population age and household structures shape transmission dynamics of emerging infectious diseases: a longitudinal microsimulation approach. J R Soc Interface. 2023 Dec;20(209):20230087. DOI: 10.1098/rsif.2023.0087
[5] Tseng YJ, Olson KL, Bloch D, Mandl KD. Smart Thermometer-Based Participatory Surveillance to Discern the Role of Children in Household Viral Transmission During the COVID-19 Pandemic. JAMA Netw Open. 2023 Jun 1;6(6):e2316190. DOI: 10.1001/jamanetworkopen.2023.16190