Indigenous Australian household structure: a simple data collection tool and implications for close contact transmission of communicable diseases

Households are an important location for the transmission of communicable diseases. Social contact between household members is typically more frequent, of greater intensity, and is more likely to involve people of different age groups than contact occurring in the general community. Understanding household structure in different populations is therefore fundamental to explaining patterns of disease transmission in these populations. Indigenous populations in Australia tend to live in larger households than non-Indigenous populations, but limited data are available on the structure of these households, and how they differ between remote and urban communities. We have developed a novel approach to the collection of household structure data, suitable for use in a variety of contexts, which provides a detailed view of age, gender, and room occupancy patterns in remote and urban Australian Indigenous households. Here we report analysis of data collected using this tool, which quantifies the extent of crowding in Indigenous households, particularly in remote areas. We use these data to generate matrices of age-specific contact rates, as used by mathematical models of infectious disease transmission. To demonstrate the impact of household structure, we use a mathematical model to simulate an influenza-like illness in different populations. Our simulations suggest that outbreaks in remote populations are likely to spread more rapidly and to a greater extent than outbreaks in non-Indigenous populations.

Households are an important location for the transmission of communicable diseases. Social contact between household members is typically more frequent, of greater intensity, and is more likely to involve people of different age groups than contact occurring in the general community. Understanding household structure in different populations is therefore fundamental to explaining patterns of disease transmission in these populations. Indigenous populations in Australia tend to live in larger households than non-Indigenous populations, but limited data is available on the structure of these households, and how they differ between remote and urban communities. We have developed a novel approach to the collection of household structure data, suitable for use in a variety of contexts, which provides a detailed view of age, gender, and room occupancy patterns in remote and urban Australian Indigenous households. Here we report analysis of data collected using this tool, which quantifies the extent of crowding in Indigenous households, particularly in remote areas. We use this data to generate matrices of age-specific contact rates, as used by mathematical models of infectious disease transmission. To demonstrate the impact of household structure, we use a mathematical model to simulate an influenza-like illness in different populations. Our simulations suggest that outbreaks in remote populations are likely to spread more rapidly and to a greater extent than outbreaks in non-Indigenous populations. 45 Households are an important location for the transmission of communicable diseases due to the frequency, 46 duration and strength of the interactions that occur there. Patterns of household structure in a population 47 can influence how a disease will spread, and potentially inform how it may best be controlled. Data 48 on household structure is therefore a valuable input into mathematical models of disease transmission 49 used for decision making on control measures. Especially, due to the different household structures in 50 remote and isolated communities, it is important to take them into consideration in disease surveillance  Detailed household-level information is often not publicly available in most demographic data collec-58 tion surveys including the national census. This is particularly the case in resource-limited settings where 59 literacy levels may be low and household structures may differ markedly from the nuclear household in Australia tend to be larger than non-Indigenous households, contain more extended family members, 62 and may change in composition more rapidly (Morphy, 2006(Morphy, , 2007. Furthermore, national censuses are 63 resource intensive and conducted relatively infrequently. There is therefore a need for more lightweight 64 methods that allow for rapid, repeated measurement in specific populations where literacy levels may be 65 low. These methods would contribute in understanding the differences of household structures among 66 Indigenous communities with more accurate data, better models for prediction of outbreaks and support 67 decisions regarding control measures.

69
Here we describe a novel visually-based method for collecting data on the structure of Indigenous 70 households and provide a descriptive analysis of data collected as part of the Aboriginal Birth Cohort 71 (ABC) study. We compare the age-specific patterns of contact within these households to those occurring 72 in a non-Indigenous population. Finally, we explore potential implications of observed differences in 73 household composition for the transmission of a respiratory infection such as influenza.

Study Design and Sample
76 Study design and sample information for the ABC study has been described in Sayers et al. (2003). In          The contact matrix for an individual household, which is symmetric, is therefore given by Table 1. tributable to sleeping in close proximity. In the analyses that follow, the room factor reflects this weighting.

149
A room factor of 1 indicates that no additional weighting was attributed to sharing a room, a room factor 150 of 2 indicates that sharing a room counted twice when determining the level of contact, and so on.  transmission process as susceptible (S), who can acquire infection; exposed (E), who have been exposed 162 to infection and are in a latent incubation stage; infectious (I), who are infectious; and recovered (R), who 163 are immune to the infection from natural immunity (Fig 2). from susceptible to exposed, σ is the rate of change from exposed to infectious and γ is the rate of change 168 from infected to recovered.
In order to calculate the transmission rate of the population, Equation 5 was used.
Contact matrices for household structure (C h ) were calculated based on the data (Fig 5 and 6) and the contact matrices, the same parameters were used for each simulation. We assumed a latent period of 178 1.5 days, an infectious period of 1.5 days, and that probability of transmission within households (q 1 ) 179 was twice that of transmission within community (q 2 ). We calibrated these probabilities to produce a

188
Descriptive analysis 189 Household size data was collected using the magnetic board method as well as the questionnaire method. shows that more than one third of the population has a household size of 7 or more in the remote towns 195 where ABC studies were conducted ( Supp Fig 1). Therefore, data from the magnetic board is considered  The median proportion of household members who were adult in remote areas (67%, IQR 55-83%) 215 was less than urban areas (78%, IQR 50-100%). In contrast, the median proportion of school-aged 216 children in a household in remote areas was higher (20%, IQR 0-38%) than urban areas (0%, IQR 0-29%).

217
However, the median proportions of pre-school aged children were almost equal in both remote and urban 218 which are 0%(IQR 0-14%) and 0%(IQR 0-18%) respectively. The median proportion of male were equal 219 (50%) in both remote and urban areas.   For comparison, we also generated a household contact matrix derived from data collected in two 240 local government areas (LGAs) of Melbourne, Boroondara and Hume. Figure 6 shows the household in an Indigenous remote community the peak occurs more quickly around the 30th day with a peak 254 prevalence of 14%. In an Indigenous urban community, time taken for the peak infectious period is 255 also higher (around 50 days) compared to non-Indigenous population with a peak prevalence of 6%.

256
The total population affected by this influenza like illness for Indigenous remote, Indigenous urban and and higher room occupancy rates. Remote Indigenous communities have much higher household sizes 269 compared to urban Indigenous communities (Fig 4). In this study, we show that differences in household The methodology described is able to capture detailed data on household occupancy in a simple and 280 robust fashion. The data collected represents a "middle way" between the extensive but comparatively  The analysis of this data is subject to some limitations. Data collected may represent a somewhat 285 biased sample due to the nature of recruitment. All households sampled will, as a consequence of the 286 ABC study design, contain at least one member who is approximately 25 years old.

288
The simplicity of the data collection method imposed some limitations on the granularity of the 289 collected data. In particular, the allocation of household members to only three age categories limits the 290 resolution of the age-structured contact matrices that can be derived. It is worth noting however, that 291 the age categories chosen are typically taken to be epidemiologically significant, due to the different 292 opportunities for mixing that these groups tend to have.  In future, when conducting similar studies, a more fine-grained age structure will be useful in further 307 understanding the contact patterns among different age groups. Currently we classified household 308 members as only adult, school aged child and pre-school aged child. Categorizing household members 309 into 5-year age groups would provide a more detailed picture of contact patterns and disease transmission.

310
Also, combining the simple methodology described above with the use of mobile digital technology such 311 as a smartphone or iPad application may enable richer data to be collected without compromising the 312 intuitive nature of the method, and also remove the need for subsequent manual entry of data. Such