Intensity of rainfall and severity of melioidosis, Australia - Research

Emerging Infectious Diseases, Dec, 2003 by Bart J. Currie, Susan P. Jacups

In a 12-year prospective study of 318 culture-confirmed cases of melioidosis from the Top End of the Northern Territory of Australia, rainfall data for individual patient locations were correlated with patient risk factors, clinical parameters, and outcomes. Median rainfall in the 14 days before admission was highest (211 mm) for those dying with melioidosis, in comparison to 110 mm for those surviving (p = 0.0002). Median 14-day rainfall was also significantly higher for those with pneumonia. On univariate analysis, a prior 14-day rainfall of >125 mm was significantly correlated with pneumonia (odds ratio [OR] 1.70 [confidence interval [CI] 1.09 to 2.65]), bacteremia (OR 1.93 [CI 1.24 to 3.02]), septic shock (OR 1.94 [CI 1.14 to 3.29]), and death (OR 2.50 [CI 1.36 to 4.57]). On multivariate analysis, rainfall in the 14 days before admission was an independent risk factor for pneumonia (p = 0.023), bacteremic pneumonia (p = 0.001), septic shock (p = 0.005), and death (p < 0.0001). Heavy monsoonal rains and winds may cause a shift towards inhalation of Burkholderia pseudomallei.

**********

Melioidosis, infection with Burkholderia pseudomallei, is endemic in Southeast Asia and northern Australia (1). Within the disease-endemic region, reported incidence has been increasing; melioidosis is now recognized as the most common cause of severe community-acquired sepsis in parts of northeast Thailand (2) and the most common cause of fatal community-acquired bacteremic pneumonia in the tropical "Top End" of the Northern Territory of Australia (3). The recognized endemic region for melioidosis has also been expanding, with recent reports from Taiwan (4), China, and India (1). Sporadic foci of melioidosis have occurred in temperate locations, possibly resulting from introduced infection (1,5). Melioidosis is also an important infection to consider in travelers returning from a disease-endemic region (6,7). While most cases are from recent infection with B. pseudomallei, latency is well recognized, and disease has occurred tip to 29 years after a person has left a melioidosis-endemic area (8).

The association between rainfall and melioidosis has long been recognized, with 75% and 85% of cases occurring in the wet season in northeast Thailand (9) and northern Australia (3), respectively. In both regions, the number of seasonal cases correlates with total rainfall.

B. pseudomallei is an environmental bacterium of soil and surface water in disease-endemic locations. We have previously documented the incubation period for melioidosis from defined inoculating events to be 1-21 (mean 9) days (10). While most cases are considered to be from percutaneous inoculation (10,11), inhalation is also well recognized as a mode of infection. We have noted that melioidosis patients are often more severely ill after heavy monsoonal rainfall. We now show that intensity of rainfall is an independent predictor of melioidosis in persons admitted to hospital with pneumonia and of death. We postulate that heavy rainfall results in a shift towards inhalation as the mode of infection with B. pseudomallei, which leads to more severe illness.

Methods

Patients

The Darwin prospective melioidosis study has documented 318 culture-confirmed cases of melioidosis that occurred in the Top End of the Northern Territory in the 12 years from October 1989 until October 2001. Patient data are stored using Oracle software, version 8.0.4 (Oracle, North Sydney, Australia). Patient variables, as defined previously (3), include age, sex, ethnicity (aboriginal, iron-aboriginal), location, and risk factors, including diabetes, alcohol excess, chronic lung disease, smoking, chronic renal disease, and kava use. Clinical parameters include nature of primary melioidosis signs and symptoms (pneumonia, other), presence of bacteremia, septic shock (presence of hypotension not responsive to fluid replacement together with hypoperfusion abnormalities manifest as end organ dysfunction) (12), and outcome (death, survival).

Rainfall Data

The Top End covers 516,945 [km.sup.2]. Daily rainfall data from 12 recording stations, located throughout the region and including major remote aboriginal communities, were provided by the Bureau of Meteorology in Darwin. From these data we calculated the rainfall at each patient's location for defined periods before date of admission. Given a mean incubation period of 9 days for acute melioidosis, we used rainfall in the 14 days before admission for each patient (14-day rainfall) to broadly reflect the rainfall exposure around the infecting event.

Statistical Analysis

Statistical analyses were performed by using Intercooled STATA 7.0 (Stata, College Station, TX). Initially, median 14-day rainfall was compared for patient variables and clinical parameters. Analysis by t tests was performed after the rainfall data were normalized by using square root transformation. Subsequently, univariate and multivariate analysis was performed with the outcomes being the various clinical parameters. Categorical variables included were age (<45 years, [greater than or equal to] 45 years), sex, ethnicity, diabetes, alcohol excess, chronic lung disease, smoking, chronic renal disease, kava use, absence of any risk factors (those listed above or age [greater than or equal to] 45 years or cardiac failure, malignancy, or immunosuppressive therapy) and 14-day rainfall (<125 mm, [greater than or equal to] 125 mm). Separate multivariate analysis was also performed with normalized 14-day rainfall data as a continuous variable. All logistic regressions were performed by using stepwise forwards technique to find the most parsimonious and significant model.