The role of frailty in shaping social contact patterns in Belgium, 2022-2023

Abstract

Social contact data are essential for understanding the spread of respiratory infectious diseases and designing effective prevention strategies. However, many studies often overlook the heterogeneity in mixing patterns among older age groups and individual frailty levels, assuming homogeneity across these sub-populations. This shortcoming may undermine non-pharmaceutical interventions by not targeting specific contact behaviours, potentially reducing their effectiveness in controlling disease. To address this gap, we conducted a contact survey in Flanders, Belgium (June 2022-June 2023). We collected data from 5995 participants (overall response rates of 19.34%) who recorded 31,375 contacts with distinct individuals. Within this cohort, 14.50% were classified as frail, and 46.85% were classified as non-frail. On average, participants report 5.48 contacts daily, with a median of 4 contacts (IQR: 2-7). These contacts vary based on participants' age and frailty levels, influenced by the locations of their interactions. Using the collected data, we reconstructed frailty-dependent contact matrices and developed a contact-based mathematical model that integrates participants' and contactees' frailty levels to investigate how frailty levels affect transmission dynamics. Incorporating frailty levels into the mathematical model substantially alters the shape of epidemic curves and peak incidences. Such insights might provide useful insights for informing non-pharmaceutical interventions, indicating the potential benefit of similar data collection in different countries.

Keywords: Disease transmission; Frailty; Infectious diseases; Mathematical modeling; Older adults; Social contact.

Fig. 1

Number of contacts reported outside the home by participants’ age group and frailty level. The plots show the distribution of contacts, the average number of contacts (dots), model results (mean as solid blue lines with 95% confidence intervals as shaded area), and 95% prediction intervals (red dashed lines). Outlying points are excluded from the figure.

Fig. 2

Contact matrices showing the mixing patterns of participants based on their frailty level for contacts reported inside and outside the household, together with 95% confidence intervals obtained from non-parametric bootstrap.

Fig. 3

The relative number of contacts (dot) for contacts at home and not at home, along with their 95% confidence intervals based on an NBI regression model. The brackets ”(NA)” indicate that no answer was provided by the participants.

Fig. 4

Comparison of COVID-19-like epidemic curves together with 95% confidence intervals obtained from non-parametric bootstrap for the Belgian population with various frailty-based mixing patterns.