Spatial and temporal patterns of dengue incidence in Bhutan: a Bayesian analysis

Figures & data

Figure 1. Administrative map of Bhutan and districts with documented dengue.

Figure 2. Dengue incidence rates by sub-districts, Bhutan, 2016–2018

Table 1. Distribution of monthly means of dengue fever cases and climate and environmental variables in Bhutan, January 2016–June 2019.

Mean (Minimum–Maximum)
Districts	No. of sub-districts	Cases	Rainfall (mm)	Maximum temperature °C	Relative humidity (%)	NDVI
Chukha	11	0.69 (0.00–72.00)	11.52 (0.00–50.23)	29.29 (23.61–32.73)	71.30 (31.56–88.67)	0.46 (−0.01–0.98)
Dagana	14	0.01(0.00–3.00)	2.55 (0.00–15.37)	21.24 (11.32–26.50)	66.71 (44.43–92.80)	0.53 (0.01–0.95)
Pemagatshel	11	0.01(0.00–1.00)	3.52 (0.00–23.72)	21.81 (14.48–26.56)	82.60 (73.67–90.25)	0.61 (0.02–0.91)
Samdrup Jongkhar	11	0.43 (0.00–76.00)	7.79 (0.00–30.02)	25.41 (20.11–29.98)	74.38 (47.74–93.51)	0.47 (−0.01–0.88)
Samtse	15	0.18 (0.00–40.00)	12.62 (0.00–47.18)	28.47 (21.40–32.08)	71.00 (47.22–90.54)	0.41 (−0.06–0.98)
Sarpang	12	0.12 (0.00–13.00)	12.58 (0.00–79.73)	27.41 (21.60–33.60)	80.20 (60.25–93.03)	0.48 (−0.01–0.97)
Zhemgang	8	0.02 (0.00–2.00)	3.19 (0.00–19.01)	20.29 (12.51–26.84)	73.88 (54.74–89.64)	0.56 (0.02–0.92)
Overall	82	0.21(0.00–76.00)	7.95 (0–79.73)	25.08 (11.32–33.60)	73.96 (31.56–93.51)	0.51 (−0.06–0.98)

Figure 3. Raw standardized morbidity ratio of dengue by sub-districts in Bhutan, January 2016–June 2019

Figure 4. Temporal decomposition of numbers of dengue cases of Bhutan, January 2016–June 2019.

Figure 5. Spatial distribution of posterior means of structured (a) and unstructured random effects (b) in Bhutan, January 2016–June 2019 based on a Bayesian spatiotemporal model.

Table 2. Regression coefficients, relative risk and 95% credible interval from Bayesian spatial and non-spatial models of dengue cases in Bhutan, January 2016-June 2019.

Model/Variable	Coeff, posterior mean (95% CrI)	RR, posterior mean (95% CrI)
Model I (Unstructured)
α (Intercept)^a	−4.936 (−7.052, −3.481)
Age (Above 14 years as base)	−0.752 (−969, −0.539)	0.471 (0.379, 0.582)
Mean monthly trend	1.023 (0.764, 1.325)	2.782 (2.146, 3.762)
Rainfall lagged 2 months (mm)	0.044 (−0.043, 0.132)	1.036 (0.965, 1.113)
Relative humidity lagged 2 months (%)	0.053 (−0.144, 0.252)	1.004 (0.988, 1.021)
Maximum temperature without lag (°C)	2.188 (1.824, 2.564)	1.619 (1.495, 1.759)
NDVI lagged 1 month	−0.666 (−1.036, −0.305)	0.514 (0.355, 0.737)
Probability of extra zero	0.702 (0.641, 0.758)	2.017 (1.898, 2.133)
Heterogeneity^a
Unstructured	0.112 (0.043, 0.221)
Structured (trend)	23.88 (1.101, 152.600)
DIC	1421.200
Model II (Structured)
α (Intercept)^a	−4.699 (−6.413, −3.368)
Age (Above 14 years as base)	−0.752 (−1.208, −0.319)	0.472 (0.298, 0.726)
Mean monthly trend	1.041 (0.541, 1.546)	2.832 (1.717, 4.693)
Rainfall lagged 2 months (mm)	0.046 (−0.134, 0.224)	1.038 (0.897, 1.199)
Relative humidity lagged 2 months (%)	0.045 (−0.351, 0.439)	1.004 (0.972, 1.036)
Maximum temperature without lag (°C)	2.254 (1.270, 3.345)	1.644 (1.323, 2.091)
NDVI lagged 1 month	−0.636 (−614, 0.283)	0.529 (0.199, 1.328)
Probability of extra zero	0.704 (0.629, 0.770)	2.022 (1.878, 2.159)
Heterogeneity^a
Structured (spatial)	0.044 (0.735, 39.700)
Structured (trend)	7.048 (0.018, 0.086)
DIC	1529.61
Model III (Mixed)
α (Intercept)^a	−4.779 (−6.764, −3.468)
Age (Above 14 years as base)	−0.753 (−0.972, −0.540)	0.471 (0.378, 0.582)
Mean monthly trend	1.023 (0.758, 1.327)	2.782 (2.133, 3.769)
Rainfall lagged 2 months (mm)	0.045 (−0.043, 0.133)	1.037 (0.966, 1.114)
Relative humidity lagged 2 months (%)	0.050 (−0.151, 0.249)	1.004 (0.988, 1.020)
Maximum temperature without lag (°C)	2.199 (1.818, 2.596)	1.625 (1.493, 1.772)
NDVI lagged 1 month	−0.663 (−1.040, −0.290)	0.515 (0.353, 0.748)
Probability of extra zero	0.703 (0.642, 0.759)	2.019 (1.900, 2.135)
Heterogeneity^a
Unstructured	0.126 (0.047, 0.259)
Structured (spatial)	275.700 (0.593, 1843)
Structured (trend)	38.740 (1.142, 349.500)
DIC	1423.63

Abbreviations: coeff: coefficients; CrI: credible interval; RR-relative risk; DIC-deviance information criterion.

^aCoefficient.

Figure 6. Trend analysis of dengue incidence in Bhutan, January 2016–June 2019, based on the spatiotemporal random effects of a Bayesian model.