Migration of persons between households in rural Alaska: considerations for study design

INTRODUCTION: Recent epidemiologic research studies in rural Alaska have examined risk factors for infectious diseases collected at the household level. Examples include the health effects of in-home piped water and household air quality. Because the exposure is measured at the household level, it is necessary to determine if participants remained in the same house throughout the course of follow-up. METHODS: We used data from a pneumococcal nasopharyngeal carriage study in 8 rural Alaska villages [3 regions; average number of persons: 642 (min 210, max 720 per village) to quantify changes in household membership and individual movements from 2008 to 2010. We define a household as a group of individuals living in a home together. Because the same households participated in carriage surveys over several years, we could determine changes on an annual basis. We calculated the percentage of households with a ≥1 person change in household members from year to year. Additionally, we present the percentage of individuals that changed households during consecutive years. RESULTS: In 3 regions of Alaska, the average household size was 5 persons. Between 2008 and 2009, 50% (250/497) of households had a change in their membership (≥1 person in-migrated or out-migrated). Fifty-three percent of households experienced some migration of their members between 2009 and 2010. A total of 27 and 15% of households had a change of ≥2 and ≥3 persons, respectively. The percentage of households with movement was similar among the 3 rural regions and varied from 42 to 63% between villages. At the individual level, an average of 11% of persons changed households between years. The group with the most movement between houses was persons 18–29 years of age (19%), and least movement was in 5–10 and 50–64 years of age (6%). There was no difference in movement by gender. CONCLUSIONS: In rural Alaska, 52% of households experienced movement of members between years and 11% of individuals change households. These are important demographic figures to consider when planning and designing studies that measure an epidemiological exposure at the household level. Power and sample size calculations should account for the loss to follow-up associated with in- and out-migration of individuals from households.