Social Mixing Patterns and 171 Empirical Contact Matrices for Infectious Disease Modeling in Sub-saharan Africa: A Systematic Review and Meta-analysis
Faculty Mentor
Professor Isaac Chun Hai Fung
Location
Russell Union Room 2080
Type of Research
Completed
Session Format
Oral Presentation
College
Jiann-Ping Hsu College of Public Health
Department
Biostatistics, Epidemiology & Environmental Health Sciences
Abstract
Background: Empirical social contact matrices are essential for modeling the transmission of respiratory pathogens, yet evidence from Sub‑Saharan Africa (SSA) remains fragmented.
Methods: We conducted a systematic review and meta‑analysis of empirical social contact surveys in SSA following PRISMA and MOOSE guidelines. Twenty‑two studies conducted between 2010 and 2022 were included, yielding 171 empirical contact matrices from 18 countries. Individual‑level data from 15 surveys were harmonized and analyzed using Bayesian negative binomial and logistic regression models to estimate determinants of total, physical, and long‑duration contacts.
Results: Children < 15 years reported the highest daily contacts (M =10, IQR = 7–14), while adults aged 15–44 years had modestly higher adjusted contact rates (unweighted adjCRR = 1.11, 95% CrI [1.07, 1.14]). Household size showed the strongest gradient: compared with 1–2‑person households, contact rates increased by 17% in 3–5‑person households (adjCRR = 1.17, 95% CrI [1.13, 1.21]), by 41% in 6–10‑person households (weighted adjCRR = 1.41, 95% CrI [1.36, 1.47]), and by 49% in 11-plus‑person households (weighted adjCRR = 1.49, 95% CrI [1.4, 1.58]). Rural participants reported more contacts (M = 11, IQR = 8–15) than urban participants (M = 8, IQR = 5–12). Methodological variables showed weaker and more uncertain effects, but retrospective recall tended to underestimate contacts (prospective M = 10, IQR = 7–14 vs. retrospective M = 5, IQR = 3–8). Weekend contacts were more physical and longer in duration, though total contact rates remained similar (weighted adjCRR = 0.97, 95% CrI [0.94, 1.00]).
Conclusions: Empirical variations, particularly in age, household size, and rural–urban context, dominate social mixing patterns in SSA, most methodological differences exert limited influence except recall type and weekend effects. These findings provide an evidence‑based foundation for selecting contact matrices in infectious disease modeling for SSA.
Program Description
.
Start Date
4-23-2026 10:00 AM
End Date
4-23-2026 10:15 AM
Recommended Citation
Kengne, Francis B., "Social Mixing Patterns and 171 Empirical Contact Matrices for Infectious Disease Modeling in Sub-saharan Africa: A Systematic Review and Meta-analysis" (2026). GS4 Student Scholars Symposium. 10.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/10
Social Mixing Patterns and 171 Empirical Contact Matrices for Infectious Disease Modeling in Sub-saharan Africa: A Systematic Review and Meta-analysis
Russell Union Room 2080
Background: Empirical social contact matrices are essential for modeling the transmission of respiratory pathogens, yet evidence from Sub‑Saharan Africa (SSA) remains fragmented.
Methods: We conducted a systematic review and meta‑analysis of empirical social contact surveys in SSA following PRISMA and MOOSE guidelines. Twenty‑two studies conducted between 2010 and 2022 were included, yielding 171 empirical contact matrices from 18 countries. Individual‑level data from 15 surveys were harmonized and analyzed using Bayesian negative binomial and logistic regression models to estimate determinants of total, physical, and long‑duration contacts.
Results: Children < 15 years reported the highest daily contacts (M =10, IQR = 7–14), while adults aged 15–44 years had modestly higher adjusted contact rates (unweighted adjCRR = 1.11, 95% CrI [1.07, 1.14]). Household size showed the strongest gradient: compared with 1–2‑person households, contact rates increased by 17% in 3–5‑person households (adjCRR = 1.17, 95% CrI [1.13, 1.21]), by 41% in 6–10‑person households (weighted adjCRR = 1.41, 95% CrI [1.36, 1.47]), and by 49% in 11-plus‑person households (weighted adjCRR = 1.49, 95% CrI [1.4, 1.58]). Rural participants reported more contacts (M = 11, IQR = 8–15) than urban participants (M = 8, IQR = 5–12). Methodological variables showed weaker and more uncertain effects, but retrospective recall tended to underestimate contacts (prospective M = 10, IQR = 7–14 vs. retrospective M = 5, IQR = 3–8). Weekend contacts were more physical and longer in duration, though total contact rates remained similar (weighted adjCRR = 0.97, 95% CrI [0.94, 1.00]).
Conclusions: Empirical variations, particularly in age, household size, and rural–urban context, dominate social mixing patterns in SSA, most methodological differences exert limited influence except recall type and weekend effects. These findings provide an evidence‑based foundation for selecting contact matrices in infectious disease modeling for SSA.
