Cairns Hospital is the sole public microbiological service provider for FNQ, a region of >380,000 km² (7). We reviewed all culture-confirmed cases of B. pseudomallei infection identified in the hospital’s laboratory during January 1, 1998–December 31, 2019. Clinical details of each case were recorded as described by J.D. Stewart et al. (4); predisposing conditions were diabetes mellitus (glycated hemoglobin ≥6.5%), hazardous alcohol use, chronic lung disease, chronic renal disease, and immunosuppression. We used data from the Australian Bureau of Statistics to calculate disease incidence and from the Australian Bureau of Meteorology to record local climatic factors including rainfall, temperatures, cloud cover, dew points, and cyclones. For our analyses we used Stata version 14.2 statistical software (https://www.stata.com) and determined trends over time by using an extension of the Wilcoxon rank-sum test and using year of presentation as a continuous variable (8). We constructed maps by using MapInfo Pro 2019 Geographic Information System software (https://support.precisely.com); in the absence of a clear occupational or recreational exposure, we used participants’ residential addresses as the site of B. pseudomallei exposure. The study was approved by the Far North Queensland Human Research Ethics Committee (HREC/15/QCH/46-977).

A total of 297 cases of melioidosis were diagnosed during the study period, of which 284 were acquired from FNQ and included in our analysis. The mean annual incidence in FNQ increased from 4.0 (95% CI 2.7–5.2) cases/100,000 population during 1998–2002 to 9.9 (95% CI 4.9–14.9) cases/100,000 population during 2016–2019 (p<0.001) (Table 1). In the Cairns area, incidence rose from 0.6 (95% CI 0.1–1.1) cases/100,000 population during 1998–2002 to 6.6 (95% CI 3.0–10.2) cases/100,000 population during 2016–2019 (p<0.001) (Table 2; Figure 1).

During the study period, the proportion of patients in FNQ with different predisposing conditions for melioidosis did not change. The proportion of bacteremic patients also remained stable (Table 1). The case-fatality rate declined during the study period (Table 1).

In the Cairns area, where increased incidence was more marked, the small increases in mean temperature, cloud cover, and dew points in the final period of the study did not reach statistical significance. During the study period, 14 cyclones came within 200 km of Cairns, but only 1 occurred during 2016–2019 (p = 0.86) (Appendix).

Figure 2

Figure 2. Cases of melioidosis occurring near to a motorway upgrade in southern Cairns, Far North Queensland, Queensland, Australia, 1998–2019.

Of the 284 cases included in the study, 111 (39%) were in the Cairns area; only 3 of these patients reported having an occupation in construction. Before commencement of the southern motorway expansion in the Cairns area in 2011, only 1/19 (5%) cases in the Cairns area were within 1,000 m of the existing road and 2/19 (11%) were within 2,000 m. However, after January 2012, a total of 92/168 (55%) cases occurred in the Cairns area, of which 15/92 (16%) were within 1,000 m of the highway construction and 27/92 (29%) within 2,000 m (Figure 2).

The incidence of melioidosis in FNQ is rising, increasing during the study period by ≈10-fold in the Cairns area. The proportion of bacteremic patients has not changed, suggesting improved diagnosis. Similarly, we found no statistically significant change in climate or frequency of cyclones. The proportion of patients who had the common predisposing conditions remained similar. However, urban expansion may be contributing because almost a third of cases in the past 8 years of the study period occurred within 2,000 m of development of a large motorway. Of note, the motorway is built predominantly through alluvial plain soils with moderate clay content and poor drainage, which favor B. pseudomallei growth (9).

Increased rainfall, dew points, cloud cover, and temperatures have been associated with increased melioidosis cases; however, these climatic factors were stable over our study period (5). Cyclones have been linked to increased melioidosis cases; however, we did not observe that association in FNQ (10). Indeed, since 2015 when melioidosis incidence in the Cairns area sharply increased, there has been only 1 cyclone within 200 km of the area.

Why the rates of bacteremic melioidosis remain higher in FNQ than in other parts of Australia remains unclear (11). The higher rates may be partly explained by fewer diagnoses of cutaneous disease in rural and remote communities; however, skin swab samples are frequently taken to identify other pathogens, and skin and soft tissue B. pseudomallei infections are uncommon in urban areas, where most new cases have been identified (12). Virulence factors in local B. pseudomallei strains may contribute (13). Despite the increasing incidence, the overall case-fatality rate from melioidosis in FNQ decreased significantly during the study period, which can probably be explained by early recognition and prompt access to multimodal intensive care unit support.

Among the limitations of our study, data collection was predominantly retrospective; in addition, in the absence of clear inoculation with B. pseudomallei, we assumed residential addresses to be the sites of exposure. Domestic gardens are a source of melioidosis in Australia, but it is possible that unrecorded patient movements may have resulted in exposure to B. pseudomallei elsewhere (14). Additional confounding factors that increase the risk for melioidosis (e.g., socioeconomic disadvantage) may help explain regional variations in incidence, although the local geographic distribution of this socioeconomic disadvantage has not substantially changed in the past 20 years (15). In conclusion, although host factors and climate continue to influence the risk of acquiring melioidosis, we hypothesize that urban expansion and construction in soils harboring B. pseudomallei may explain the recent rapid increase in Far North Queensland, Australia.