Social contact patterns among school-going children in a rural location in Kenya
Abstract
Contact patterns drive the dynamics of transmission of respiratory viruses. Schools, as
places of very high contacts are presumably important for transmission. Contact data support
modelling the spread of respiratory infections, in particular, emerging viruses such as
pandemic influenza. There are no direct estimates on respiratory virus transmission or
contact patterns in schools in developing countries. We demonstrate the use of paper diaries
in defining contact patterns among school children in a developing country setting. 451
diaries (240 term, 211 vacation) obtained from 301 participants aged ≥10 years from 7
public schools (6 primary and 1 secondary) were analysed. Individuals were assigned
contact diaries on two separate occasions, school term and vacation period. Recorded details
of their contacts included age-group, sex, location and frequency of contact. A contact was
defined as either conversation (type 1) or physical touch (type 2). Multi-level linear models
were fit using STATA 11 to investigate source of variation in contacts at different levels
(individual, age-group, school). The final model explored gender, age-groups and diary
time..Students had significantly more contacts during the school term than the vacation
period (mean term =29.8 versus vacation = 23.5, multi -level analysis: p<0.001). 59% of
contacts during vacation occurred outside the home and 29% of these were in-school
contacts due to vacation additional tuition. The trend for unadjusted mean number of unique
contacts across age groups was significant (p=0.001) ranging from 30.5 in 11-13 year olds to
20 in those above 20 years. There was no significant difference in rates of type 1 (Mean
(Standard Deviation) =12.7 (12.9) and type 2(Mean (SD) = 12.4(11.9) contacts; t (2) =-
1.132, p=0.375. Students tended to mix with people of the same age groups . After
adjustment for various variables, new contacts were significantly associated with being
female Mean= 32.1 (p>0.02) and belonging to a lower age group (Mean = 26.8
p>0.001).This first study of school-child contact patterns revealed evidence for higher
mixing among girls and younger individuals, coupled to significant assortative (within
group) mixing. The demonstration of significantly greater contact rates in school term than
vacation has potential implications for school closures during outbreaks of emergent viruses
e.g. pandemic influenza. Diary studies in the developing country setting are feasible and
have use in defining data for modelling disease spread.