University of Pittsburgh
Preventative public health measures should be focused in the south of Viet Nam where incidence is highest. Vitamin D is unlikely to be a strong driver of seasonality but supplementation may play a role in a package of interventions.
There were 610,676 cases of TB notified between 2010 and 2015 in Vietnam. Heat maps of TB incidence per quarter per province showed substantial temporal and geospatial variation. Time series analysis demonstrated seasonality throughout the country, with peaks in spring/summer and troughs in autumn/winter. Incidence was consistently higher in the south, the three provinces with the highest incidence per 100,000 population were Tay Ninh, An Giang and Ho Chi Minh City. However, relative seasonal amplitude was more pronounced in the north. Mixed-effect linear model confirmed that TB incidence was associated with time and latitude. Of the demographic, socio-economic and health related variables, population density, percentage of those under 15 years of age, and HIV infection prevalence per province were associated with TB incidence. Of the climate variables, absolute humidity, average temperature and sunlight were associated with TB incidence.
Tuberculosis (TB) is a major global health burden, with an estimated quarter of the world's population being infected. The World Health Organization (WHO) launched the "End TB Strategy" in 2014 emphasising knowing the epidemic. WHO ranks Vietnam 12th in the world of high burden countries. TB spatial and temporal patterns have been observed globally with evidence of Vitamin D playing a role in seasonality. We explored the presence of temporal and spatial clustering of TB in Vietnam and their determinants to aid public health measures.
Data were collected by the National TB program of Vietnam from 2010 to 2015 and linked to the following datasets: socio-demographic characteristics; climatic variables; influenza-like-illness (ILI) incidence; geospatial data. The TB dataset was aggregated by province and quarter. Descriptive time series analyses using LOESS regression were completed per province to determine seasonality and trend. Harmonic regression was used to determine the amplitude of seasonality by province. A mixed-effect linear model was used with province and year as random effects and all other variables as fixed effects.