Characteristics of human encounters and social mixing patterns relevant to infectious diseases spread by close contact: a survey in Southwest Uganda

Characteristics of social contacts by time, person and place

We analysed the frequency distribution of contacts for a set of covariates, including age, sex, and occupation, day of the week, distance travelled, and type of contact. Encounters reported with the same individual in different settings counted as one contact only. Straight-line distances between the centre point of all villages and towns in the dataset were calculated, and these were then used to evaluate how far people travelled, based on the reported names of villages and town where each reported encounter took place, and their own village or town of residence.

We used negative binomial regression to estimate the ratio of the mean contacts as a function of the different covariates of interest. Negative binomial was preferred over Poisson regression given evidence of over-dispersion (variance > mean, and likelihood ratio significant (P < 0.05) for the over-dispersion parameter). We considered variables associated with contact frequency at p < 0.10 for multivariable analysis, and retained them in multivariable models if they resulted in a reduction of the Bayesian Information Criterion (BIC).

Next, we explored whether people reporting a high frequency of casual contacts (≥10 casual contacts) differed from those reporting fewer contacts with regards to their socio-demographic characteristics. We did so using log-binomial regression to compute crude and adjusted relative risks (RRs) for having a high frequency. In all analyses we accounted for possible within-cluster correlation by using linearized based variance estimators [17]. Analyses were also weighted for the unequal probabilities of sampling selection by age group.

Age-specific social contact patterns

We analysed the age-specific contact patterns through matrices of the mean number of contacts between participants of age group j and individuals in age group i, adjusting for reciprocity, as in Melegaro et al. [6].

If x _ij denotes the total number of contacts in age group i reported by individuals in age groups j, the mean number of reported contacts (m _ij ) is calculated as x _ij /p _j , where p _j is the study population size of age group j. At the population level the frequency of contacts made between age groups should be equivalent such that m _ij P _j  = m _ji P _i . The expected number of contacts between the two groups is thereforeC _ij = (m _ij P _j  + m _ji P _i )/2. Hence, the mean number of contacts corrected for reciprocity \( \left({m}_{ij}^C\right) \) can be expressed as C _ij /P _j .

We tested the null hypothesis of proportionate mixing by computing the ratio of observed mixing patterns to that of expected mixing patterns if social contact occurred at random. Under the assumption of random mixing, the probability of encounter between age groups thus depends on the population distribution in each age group, and the contact matrix under this random mixing hypothesis was calculated based on the percentage of population in each age group. The ratio of observed over expected contacts was then computed, and confidence intervals were obtained through bootstrapping, with replacement, for a total of 1000 iterations. This approach is similar to that taken by others [11].

Epidemic simulations

Finally, in order to explore the infection transmission dynamics resulting from our contact pattern data, we simulated the spread of an immunizing respiratory infection transmitted through close contact in a totally susceptible population, thus assuming a Susceptible-Infected-Recovered (SIR) model. The model contained nine mixing age groups, with a transmission rate β _ij at which individuals in age group jcome into routine contact with individuals in age group i computed as \( {\beta}_{ij}={qm}_{ij}^C/{\omega}_i \), where ω _i is the proportion of individuals in age group i, and \( {qm}_{ij}^C \) is the next generation matrix, with qrepresenting the probability of successful transmission per contact event [18]. We assumed q to be homogeneous and constant across all age groups and conducted a set of simulations for fixed values of q between 25% and 40%, in line with what has been reported with influenza pandemic strains [18, 19]. The basic reproduction number (R₀) – which corresponds to the average number of people infected by one infectious individual in a totally susceptible population – was calculated as the dominant eigenvalue of the next generation matrix. We took uncertainty estimates in the contact matrices (and hence final size outputs) into account by iterating the model on bootstrapped matrices.

,where F represents the final epidemic size by age group (i.e. the proportion of individuals who are infected in each age group).

Estimates obtained using the contact data from Uganda were compared to that of Great Britain, using data from the POLYMOD study [4] for the latter and a similar approach to compute the mixing matrix. The model was parameterised with social contact data on physical contacts only, lasting ≥5 min, rather than all contacts, given that physical contacts generally seem to better capture contact structures relevant for the transmission of respiratory infections [6]. Data for Great Britain were available for the same age groups, and for physical contacts specifically, in the same way that physical contacts were defined in our study, which made the data for physical contacts only more comparable between studies than that of overall contacts.

Analyses were performed in either Stata13.1 IC or R version 3.2 [21].

Contacts (i.E. ≥5 min long)

A total of 3965 contacts with different individuals were reported, corresponding to an average of 7.2 contacts per person (median 7, range 0 - 25) (Fig. 1). The majority of contacts were physical, thus involving skin-to-skin contact (mean 5.1, median 5 (range 0 – 18)). The age and sex distribution of study participants can be found in Table S1 (Additional file 2 with supplementary figures and tables).

Over half of all contacts (n = 2060 (52%)) were with household members, 627 (16%) with other relatives, 873 (22%) with colleagues/friends/schoolmates and 402 (10%) with other individuals. The duration of routine contacts is shown in Figure S1 (Additional file 2).

Most contacts (82%) were with individuals who would be normally seen daily, 520 (13%) with people normally seen at least weekly, 4% with people met more rarely and 1% of the reported contacts were with people that the participants had never met before.

We found marked differences in the number of contacts by age group, but not by sex. School-aged children reported the highest daily number of contacts, while the elderly had the fewest (Table 1). There was no difference in the mean number of contacts for individuals living in the district towns of Kabwohe and Itendero (n = 43) and the 523 others living in surrounding villages (chi-square test P-value = 0.79). Table 1 provides further details about the population characteristics, the mean number of contacts by socio-demographic and other covariates, as well as the ratio of mean contacts by covariate. Age was the only confounding factor.

Overall, contacts tended to be assortative, as shown by the strong diagonal feature on Fig. 2, with most of the intergenerational mixing occurring within households (Fig. 3). Only teenagers and adults reported non-physical contacts (Fig. 3). The quantification of assortativity can be seen in Figures S2a-c (Additional file 2), which show the ratios of observed contacts, as obtained in the survey but corrected for reciprocity, to that of expected contacts under the proportionality assumption, for all contacts and physical contacts only. The results show age-assortativity of contacts, for all age groups (other than < 2 year olds) for all contacts, and primarily for school-aged children when considering physical contacts only.

Reciprocity correction accounted for differences in reporting of contacts by age groups, particularly a proportionally higher frequency of contacts reported by young children with older age groups than older age groups reported with young children, as shown in Figure S3 (Additional file 2).

There was no statistical difference in the average number of contacts between weekend (Sunday) and weekdays (Monday, Tuesday, Thursday and Friday) (Table 1). As shown in Table 1, mean number of reported contacts on Sundays was 7.50 (95%CI 6.56; 8.44), slightly higher on average than on weekdays, where the average was 7.14 (6.79; 7.49), which was not statistically significant (P = 0.229). The balance of respondents reporting contacts from weekdays and weekends reflected the normal proportion of week vs. weekend days in a normal week.

About a quarter (n = 136 (24%)) of participants reported social encounters outside their village of residence, and about 12% of contacts occurred outside participants’ village of residence. The majority (56%) of people who travelled outside their village went to places located within a 5 km radius from the centre point of their village of residence, and 90% stayed within 12 km (Fig. 4). Adult males tended to travel more than females (Fig. 4). Overall, 29% of males had contact with someone outside their village, compared to 20% females (χ², P = 0.0406). Most (87%) children under 5 years of age stayed in their village, whereas about a quarter or more individuals travelled outside their village among 5 – 14 years old (25%), 15-44 years old (32%) and ≥ 45 years of age (25%), a difference by age group which was significantly different (P = 0.0081).

Overall, 30% of males travelled outside their village compared to 20% of females. When stratifying by age, the difference between sex were more marked, with no statistical difference between males and females < 5 years of age (P = 0.296) or 5 – 14 year olds (P = 0.272), but marked differences among adults (≥15 years old), with 42% of males travelling outside of their village compared to 24% of females (P = 0.0037).

The proportion of individuals who travelled outside their village differed by occupation (although not statistically significantly, with shop keepers (38%), those working in agriculture (31%) and office workers (50%) travelling outside their village more than others.

Most contacts made outside the household as well as those with individuals outside participants’ village were mostly assortative (Fig. 3), and the proportion of contacts outside the village was different by age group (P < 0.001); higher among adults, increasingly so as distance from home increased (Fig. 4).

‘Casual’ contacts (< 5 min long)

Information on the number of casual contacts was reported by 490 (87%) participants. Among those, 64% (n = 315) estimated they had fewer than 10 different contacts, 24% reported between 10 and 19 casual contacts, 6% reported between 20 and 29 contacts and 6% reported an estimated 30 contacts or more.

Individuals who reported high levels (i.e. ≥10 contacts) of casual contacts also tended to report more contacts (Table 1). We found no difference between those reporting high number of casual contacts (≥10) and others, by age, sex or day of the week (Additional file 2: Table S2). However, people whose primary activity was at home tended to report fewer casual contacts than others, and there were about 60% more individuals reporting high levels of casual contacts among those who travelled outside their village.

The 76 (13%) study participants for whom the number of casual contacts was not known known reported more non-casual contacts than others, with a mean number of contacts of 8.7, compared to 7.0 among the 490 for whom the number of casual contacts was estimated (Ratio of means 1.24 (95%CI 1.11 – 1.39). This may be due to the age distribution of individuals for whom estimates of casual contacts was missing, which was proportionally and significantly higher among school aged 5-9 year olds (29% with no information on number of casual contacts), who also report more contacts overall, and lower in all older age groups (Chi-square P-value< 0.001).

Epidemic simulations

Finally, we compared patterns of reported physical contacts in Uganda and Great Britain, and explored differences in the relative and absolute epidemic size by age group, as well as the corresponding R₀, for a hypothetical respiratory infection in an immune-naive population.

The number of reported physical contacts was similar between Uganda and Great Britain, with the average number of contacts by age group ranging from 3.2 (≥65 year olds) to 7.3 (2 – 4 year olds) in Uganda and from 3.3 (55 – 64 year olds) to 7.3 (10 – 14 year olds) in Great Britain. However contacts were more assortative in Britain than in Uganda (Fig. 5a & b), some of which might be related to differences in household structures and number of household contacts, as contacts outside the household were mostly assortative (Fig. 3).

The computed mean values of R₀ for a per contact infectivity value (q) ranging from 0.25 to 0.40 was slightly higher in Great Britain than in Uganda (1.51 to 2.41 vs. 1.40 to 2.24). Figure 5f shows the values for an infectivity parameter of 0.33. The proportion of people infected in younger age groups was also higher in Great Britain, and there were proportionally more adults infected in Uganda. However, given the differences in population structure, the total number of infections in the population was higher in Uganda than in Great Britain (Fig. 5c – e).