@article {Lam100222, author = {Ha Minh Lam and Amy Wesolowski and Ngyuyen Thanh Hung and Tran Dang Nguyen and Nguyen Thi Duy Nhat and Stacy Todd and Dao Nguyen Vinh and Nguyen Ha Thao Vy and Tran Thi Nhu Thao and Nguyen Thi Le Thanh and Phan Tri Tin and Ngo Ngoc Quang Minh and Juliet E Bryant and Caroline O Buckee and Tran Van Ngoc and Nguyen Van Vinh Chau and Guy E Thwaites and Jeremy Farrar and Dong Thi Hoai Tam and Ha Vinh and Maciej F Boni}, title = {Non-annual seasonality of influenza-like illness in a tropical urban setting}, elocation-id = {100222}, year = {2017}, doi = {10.1101/100222}, publisher = {Cold Spring Harbor Laboratory}, abstract = {In temperate countries, influenza and other viral respiratory diseases often have distinct seasonal peaks occurring during colder, wintertime months. However, little is known about the dynamics of influenza and viral respiratory disease dynamics in the tropics, despite high morbidity and a clear epidemiological link between tropical and temperate countries. In temperate countries, the dynamics of influenza and other respiratory diseases are often analyzed using syndromic surveillance data describing influenza-like illness (ILI) as ILI is highly correlated with virological surveillance for influenza. To obtain a detailed picture of respiratory disease incidence patterns in a large tropical city, we established an mHealth study in community outpatient clinics in Ho Chi Minh City, Vietnam (11N latitude). From August 2009 through December 2015, clinics reported daily case numbers of ILI using standard mobile-phone SMS messaging. A subset of these clinics performed molecular diagnostics for influenza A and B viruses. Unlike the annual patterns seen in temperate countries, ILI activity in Ho Chi Minh City exhibited strong non-annual periodicity and was not correlated with PCR-confirmed influenza. The dominant periodicity in the data was approximately 200 days. This was confirmed by a time series decomposition, a step-wise regression analysis on annual and non-annual covariates, and a forecasting exercise showing that forecasting was 30\% to 40\% more accurate when a 200-day non-annual cycle was included in the forecast. This suggests, for the first-time, that a non-annual cycle may be an essential driver of ILI dynamics in the tropics. This raises new questions about the seasonality and drivers of respiratory disease transmission in tropical countries.}, URL = {https://www.biorxiv.org/content/early/2017/01/13/100222}, eprint = {https://www.biorxiv.org/content/early/2017/01/13/100222.full.pdf}, journal = {bioRxiv} }