THE GEOGRAPHY OF DISEASE SEASONALITY IN TEXAS: A FOURIER TRANSFORM APPROACH

Our research goal is to identify and describe the seasonality—disease relationship for different specific causes of hospitalization in various parts of Texas. Subsequent interpretation of the disease driver, the underlying causal relation, could inform geographic- and population-specific recommendations to minimize disease incidence and societal cost.

We examined asthma, acute myocardial infarction, and fractures in 3 widely separated geographic regions of Texas (El Paso, Houston, and Brownsville). A time series of 10 years of computerized daily hospital admission records was subjected to Fourier analysis. This technique reproduces a signal that varies in time, here daily hospital admissions, as the sum of a set of sine waves of specific frequencies, amplitudes, and phases. In the Fourier “frequency domain,” intense spectral peaks at frequencies that are integral cycles per year, e.g., 1 cpy, 2 cpy…, likely reveal a seasonal effect. The Fourier analysis establishes the existence of periodicity, reveals the frequencies of periodic components, and determines their relative influences.

All 3 cities exhibited similar annual cycles of asthma despite significant climate differences. El Paso is high (1200 m) desert with winter freezes; Houston and Brownsville are sea-level, humid, and warm year-round. Wider valleys in the Brownsville time domain represent a longer summer reprieve from asthma. We are now trying to explain what drives the inter-city similarities and differences in this well-established seasonal disease. Brownsville exhibited an as-yet unexplained robust late-year annual cycle in myocardial infarction admits, not evident in Houston or El Paso. All types of fractures were seasonal just in Houston, with a pronounced mid-winter lull. The annual fracture lull in Houston in February remains a challenge to interpret.

The Fourier approach to geographic variation in disease seasonality has uncovered unexpected patterns that thus far defy simple interpretation. Once the underlying drivers have been discovered, these patterns may prove valuable in preventing disease or assisting in deployment of health-care resources. As a perhaps whimsical example of the latter, the unexplained fracture data nonetheless suggest that Houston orthopedists should take their vacations in February.