Temporal and spatial dynamics of Tuberculosis in Singapore

Sourav Das (Saw Swee Hock School of Public Health, National University of Singapore)

Frank Adams 2,

Tuberculosis (TB) has long been a major cause of death across the globe. In 1993 the
World Health Organization (WHO) declared TB a global health emergency. TB was also highlighted
in the United Nation’s millennium development goals. Much has been achieved since
then; especially the reduction of 45% in mortality rates. However challenges persist in terms
of prevalence of the disease. With 29% of the reported global TB incidence, South East Asia
remains a significant contributor to global TB burden and Singapore, a major economic hub
in region, has witnessed a sudden increase in TB incidence contrasting the trends among advanced
economies. In this two part analysis we investigate the temporal and spatial patterns
of TB risk in Singapore. We model the monthly temporal risk of TB data using Seasonal
Autoregressive Moving Average (SARIMA) and Generalized Linear Autoregressive Moving
Average (GLARMA) methodology. We observe that temporal risk of TB in Singapore has
several periodic effects and the non-resident TB risk has an overall decreasing trend. However
TB is a communicable disease driven by several socio-economic and environmental factors
leading to a random spread (as also the count) of the disease in a geographical area. Within a
fixed time period, assuming TB incidence to be a spatial point process (identified by residential
address of cases), we construct risk maps of TB in Singapore over a period of seventeen years
(1995-2011), using standard point process methodology. We classify Singapore’s administrative
districts into 9 clusters based on the interpolated spatial risk. We construct a Monte-Carlo
test to assess the statistical significance of change in overall risk of the disease. We observe
that TB risk in Singapore is inhomogenous and there is statistically significant inter-case interaction.
However resident and non-resident cases do not display significant residential clusters,
implying low inter-demographic group transmission in residential neighbourhood.

Import this event to your Outlook calendar
▲ Up to the top