Clinical Epidemiology of 7,126 Melioidosis Patients in Thailand and the Implications for a National Notifiable Diseases Surveillance System
Hantrakun V., Kongyu S., Klaytong P., Rongsumlee S., Day NPJ., Peacock SJ., Hinjoy S., Limmathurotsakul D.
Abstract Background National notifiable diseases surveillance system (NNDSS) data in developing countries is usually incomplete, yet the total number of fatal cases reported is commonly used for national priority setting. Melioidosis, an infectious disease caused by Burkholderia pseudomallei, is largely underrecognized by policy makers due to the underreporting of fatal cases via the NNDSS. Methods Collaborating with the Epidemiology Division (ED), Ministry of Public Health (MoPH), we conducted a retrospective study to determine incidence and mortality of melioidosis cases already identified by clinical microbiology laboratories nationwide. A case of melioidosis was defined as a patient with any clinical specimen culture positive for B. pseudomallei. Routinely available microbiology and hospital databases of secondary-care and tertiary-care hospitals, the national death registry, and NNDSS data were obtained for analysis. Results A total of 7,126 culture-confirmed melioidosis patients were identified from 2012 to 2015 in 60 hospitals countrywide. The total number of cases diagnosed in Northeast, Central, South, East, North, and West Thailand were 5,475, 536, 374, 364, 358 and 19 cases, respectively. Overall 30-day mortality was 39% (2,805/7,126). Only 126 (4%) of deaths were reported to the NNDSS. Age, presentation with bacteremia and pneumonia, prevalence of diabetes, and 30-day mortality were different by geographical regions (all p values<0.001). ED, MoPH has agreed to include the findings of our study in the next annual report of the NNDSS. Conclusions Melioidosis is an important cause of death in Thailand nationwide, and its clinical epidemiology may be different by region. In developing countries, NNDSS data can be supplemented by integrating information from readily available routine datasets.