bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

1 Spatial and epidemiologic features of dengue in ,

2

3 4 Amanda Murphy1,2*, Giri Shan Rajahram3,4, Jenarun Jilip5, Marilyn Maluda5, Timothy

5 William4,6, Wenbiao Hu7, Simon Reid8, Gregor J. Devine1^, Francesca D. Frentiu2^.

6 7 8 9 1 Mosquito Control Laboratory, QIMR Berghofer Medical Research Institute, Brisbane, 10 Australia 11 2 School of Biomedical Sciences, and Institute for Health and Biomedical Innovation, 12 Queensland University of Technology, Brisbane, Australia 13 3 Queen Elizabeth Hospital, Ministry of Health Malaysia, , Malaysia 14 4 Infectious Disease Society of Kota Kinabalu-Menzies School of Health Research Clinical 15 Research Unit, Kota Kinabalu, Malaysia 16 5 Sabah Department of Health, Ministry of Health Malaysia, Kota Kinabalu, Malaysia 17 6 Gleneagles Kota Kinabalu Hospital Sabah, Kota Kinabalu, Malaysia 18 7 School of Public Health and Social Work, Queensland University of Technology, Brisbane, 19 Australia 20 8 School of Public Health, University of Queensland, Brisbane, Australia 21 22 23 * Corresponding author 24 E-mail: [email protected] 25 26 27 ^ These authors contributed equally to this work 28 29 30 Keywords: dengue, rural, Sabah, Aedes albopictus, Borneo, South East Asia 31 32

1 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

33 Abstract 34 35 In South East Asia, dengue epidemics have increased in size and geographical distribution in

36 recent years. Most studies investigating dengue transmission and control have had an urban

37 focus, while less consideration is currently given to rural settings, or where urban and rural

38 areas overlap. We examined the spatiotemporal distribution and epidemiological

39 characteristics of reported dengue cases in the predominantly rural state of Sabah, in

40 Malaysian Borneo – an area where sylvatic and urban circulation of pathogens are known to

41 intersect. We found that annual dengue incidence rates were spatially variable over the 7-

42 year study period from 2010-2016 (state-wide mean annual incidence of 21 cases/100,000

43 people; range 5-42/100,000), but were highest in rural localities in the western districts of

44 the state (, , and ). The eastern districts exhibited

45 lower overall dengue rates; however, we noted a concentration of severe (haemorrhagic)

46 dengue cases (44%) in and districts. Dengue incidence was slightly higher

47 for males than females, and was significantly higher for both genders aged between 10 and

48 29 years (24/100,000; p=0.029). The largest ever recorded outbreaks occurred during 2015-

49 2016, with the vector Aedes albopictus found to be most prevalent in both urban and rural

50 households (House Index of 64%), compared with Ae. Aegypti (15%). These findings suggest

51 that dengue outbreaks in Sabah are driven by the sporadic expansion of dengue virus in both

52 urban and rural settings. This may require tailoring of preventative strategies to suit

53 different transmission ecologies across Sabah. Further studies to better understand the

54 drivers of dengue in Sabah may aid dengue control efforts in Malaysia, and more broadly in

55 South East Asia.

2 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

56 Author summary 57 58 In order to combat the rising regional incidence of dengue in South East Asia, the drivers of

59 transmission must be better characterised across different environmental settings. We

60 conducted the first retrospective analysis of dengue epidemiology in the predominantly rural

61 state of Sabah, Malaysia, where both urban and sylvatic transmission cycles exist. Human

62 notification data over a 7-year period were reviewed and spatiotemporal and demographic

63 risk factors identified. We found:

64 1. Urban habitats and population density are not the only determinants mediating the

65 spread of epidemic dengue in Sabah. Case from both urban and rural localities

66 contributed equally to dengue outbreaks.

67 2. Human demographic risk factors included being aged between 10 and 29 years, and

68 being male.

69 3. High incidence areas for dengue do not predict the occurrence of severe dengue. Severe

70 dengue was largely localised to lower incidence districts in the east of the state.

71 4. The sole presence of Aedes albopictus in and around the majority of urban and rural

72 case households suggests that this vector may play a major role in facilitating outbreaks.

73 A complex interplay of risk factors likely mediates dengue transmission in Sabah, influenced

74 by both regional climate trends and localised human and ecological influences. This study

75 emphasises that the increasing spread of dengue in urban South East Asia is also mirrored in

76 more rural areas, and suggests a need for control strategies that address both urban and

77 rural dengue risk.

78

3 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

79 Introduction 80 81 Dengue is the most rapidly spreading vector-borne disease in the world, and the most

82 prevalent arboviral disease of humans (1). Now endemic in more than 100 countries, the

83 disease causes an enormous burden on communities and health care systems in tropical and

84 sub-tropical regions (2). The causative agent of dengue is dengue virus (DENV), transmitted

85 between humans by Aedes mosquitoes across a range of domestic and sylvatic

86 environments. Urban expansion, human migration, travel and trade have facilitated an

87 increasing number of infections, primarily in Asia, Africa and the Americas (3, 4). These areas

88 experience up to 70% of the estimated 390 million annual dengue infections worldwide (1,

89 4). Explosive outbreaks have become common in recent decades, and both classical and

90 severe (haemorrhagic) forms of dengue now occur in previously unaffected countries (1, 3,

91 5). South East Asia has one of the highest burdens of dengue, following marked increases in

92 the number, severity and geographical distribution of dengue epidemics since the 1950s.

93 During this dramatic expansion, the four virus serotypes (DENV 1-4) have become well-

94 established and commonly co-circulate within the region (6).

95 In Malaysia, dengue has been considered a major public health problem since 1973 (7), with

96 regular epidemics resulting in significant morbidity and economic burden (8, 9). The majority

97 of reported cases are concentrated in the large, urban cities of and ,

98 which are located on the Malaysian peninsula. The circulation of all DENV serotypes has

99 been documented across the country, as well as the presence of unique sylvatic strains (10-

100 12). As with many South East Asian countries, the characterisation and control of

101 transmission in Malaysia is primarily focused on highly populated urban areas (13). The

102 majority of spatial and eco-epidemiological studies to date have therefore focused on

4 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

103 peninsular Malaysia, and relatively few studies have explored the factors driving

104 transmission in rural parts of the country.

105 The Malaysian states of Sabah and , located on the island of Borneo, report lower

106 incidence rates than mainland Malaysia (14) and patterns of transmission in these states are

107 not well characterised. The island possesses rapidly developing urban areas in close

108 proximity to disturbed forest environments, with potential risk of spill-over of sylvatic

109 pathogens to human populations (15). Sabah state, positioned on the northern tip of

110 Borneo, reports the highest incidence of the sylvatic malaria parasite Plasmodium knowlesi,

111 with transmission risk linked to deforestation (16, 17). The emergence of other zoonotic

112 pathogens has also been documented in Sabah (18, 19), including Zika virus in 2015 (20).

113 Given the marked environmental change occurring in Sabah, and the increase in dengue

114 cases noted in recent years (12, 14), it is essential from a public health perspective to

115 understand current transmission patterns and their drivers. This study examined the

116 epidemiology of dengue in the state of Sabah, in Malaysian Borneo, between 2010 and 2016.

117 We aimed to document recent spatial and temporal trends of dengue disease, and to

118 identify some potential risk factors driving DENV transmission and spread in this

119 understudied region of the country.

120

121 Methods 122 123 Ethics statement

124 This study was approved by the Medical Research and Ethics Committee (MREC), Ministry of

125 Health Malaysia; and the Human Research Ethics Committee (HREC) of the QIMR Berghofer

5 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

126 Medical Research Institute, Brisbane, Australia. All human case data analysed were

127 anonymized.

128 Study site

129 The Malaysian state of Sabah lies at the most north-eastern tip of the island of Borneo. It

130 borders the Malaysian state of Sarawak and the Indonesian province of Kalimantan (Fig. 1).

131 The climate is tropical, with high humidity and rainfall throughout the year. Sabah has a

132 geographical area of 73,904 km2 and is divided into 25 districts (21). The state’s population

133 density is second lowest in the country (44 people/km2), after Sarawak (20 people/km2), and

134 Sabah also has one of the lowest overall proportions of urban population (54%) in the

135 country (14). Within Sabah, has the highest population density (1,397

136 people/km2), where the capital city of the same name is located.

137 138 Fig 1. Map of Malaysia and Sabah state. 139 Peninsula Malaysia and Malaysian Borneo are shown, along with the 13 Malaysian states 140 and 2 territories. States are coloured according to their population density, expressed as 141 number of people per square km. The island of Borneo includes the Malaysian states Sabah 142 and Sarawak, and is also shared by the country of Brunei and the Indonesian province of 143 Kalimantan. Inset: Sabah state, showing its 25 districts. The three largest cities in the state 144 are indicated by black circles: the capital city Kota Kinabalu, Sandakan and Tawau. 145

146 Historically, Sabah was almost entirely covered by primary rainforest and still has the

147 second-highest proportion of forested areas in the country (60%) after Sarawak (64%). It also

148 has high rates of forest loss, with monocultures of rubber and palm plantations now

149 estimated to cover 36-56% of the land area (15, 21). Sabah has the second-highest

150 proportion of Indigenous people in the country (25%) after Sarawak (30%), as well as the

151 highest proportion of non-Malaysian residents (25%) of all the Malaysian states (14).

152 6 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

153 Epidemiological data

154 State-wide data from monthly notified cases of dengue between the years 2010 and 2016

155 were obtained from the Sabah State Department of Health (Jabatan Kesihatan Negeri

156 Sabah), Malaysian Ministry of Health. Prior to 2010, detailed data were not available in

157 disaggregated and electronic format. Variables analysed included age, sex, district and

158 locality of each case residence (based on home address), disease severity and outcome

159 (survival or death), and diagnostic tests performed (IgG, IgM and/or NS1). In our dataset,

160 cases from 2011-2016 were designated as residing in either urban or rural localities (the

161 smallest residential geographical unit) by the Sabah Ministry of Health (MoH). MoH

162 designation of locality status is based on the Malaysian Department of Statistics definitions,

163 where urban localities are gazetted census areas with 10,000 people or more, with ≥60% of

164 the working population (≥15 years) engaged in non-agricultural activities (14). Population

165 and demographic data were obtained from the Malaysian Department of Statistics, for the

166 year 2010. Incidence rates were calculated using population projections for each year, based

167 on census data from the year 2010, along with annual growth rate projections as per the

168 published growth rate in Sabah (22).

169 During the study period, clinical cases were identified using World Health Organization

170 (WHO) guidelines using clinical symptoms and/or positive NS1 or serology (presence of IgM

171 or IgG) (23). From 2014 onwards, Malaysian national notification guidelines were modified,

172 in line with WHO advice, to require a positive laboratory diagnostic test (either NS1 and/or

173 IgM/IgG serology) in addition to the presence of clinical symptoms, and case notification

174 within 24 hours of diagnosis (24, 25). Therefore, the majority of cases prior to 2014 were

175 clinically diagnosed (with 30-50% per year confirmed by laboratory tests in our dataset),

176 while cases from 2014-2016 were 100% laboratory confirmed.

7 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

177 Entomological data

178 Entomological surveillance data (number of larvae, mosquito species identified) were

179 generated from active surveillance of potential aquatic habitats, primarily water-holding

180 containers, in and around 719 case residences inspected during the 2015-16 outbreaks. Of

181 these, 255 (36%) of residences were in a locality designated as urban by local public health

182 authorities, 437 (61%) were considered rural, and 27 (4%) had no rural or urban designation

183 recorded. Where mosquito larvae were found in or around a case household, samples were

184 taken to local public health laboratories for species identification. The presence or absence

185 of one or more species per household was recorded, and the House Index (HI) was

186 calculated as the proportion of houses infested with larvae and/or pupae (26). HI was also

187 calculated for each mosquito species present in larvae-positive households.

188 Data analysis

189 We assessed seasonal characteristics of the temporal distribution of cases using a seasonal

190 trend decomposition procedure in SPSS software. The procedure is based on the Census

191 Method I, otherwise known as the ratio-to-moving-average method where time series data

192 are separated into a seasonal component, a combined trend and cycle component, and an

193 "error" or irregular component (27). The seasonal component is then isolated from the

194 overall and irregular trends through a multiplicative model. Seasonal decomposition analysis

195 was applied to monthly dengue case numbers across the 7-year period to examine the

196 seasonal trends of case notifications across Sabah.

197 Annual and cumulative incidence of dengue was calculated using the number of notifications

198 per month and Sabah population estimates based on the 2010 Malaysian census. Incidence

199 rates were standardized for age and sex using census data and plotted for each of the 25

200 administrative districts. Ages of cases were grouped into four categories to broadly separate 8 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

201 young children from older children and adults (0-9, 10-29, 30-49 and ≥50). The statistical

202 significance of observed differences between means was determined using the Kruskal

203 Wallis test. SPSS Statistics software (SPSS, IBM New York USA; version 23) was used for data

204 analyses, with statistical significance set at p<0.05. Spatial maps of Malaysia and Sabah

205 dengue cases and incidence were created using ArcGIS (Esri Redlands USA; version 10.5.1).

206 We assessed overall and annual trends of rural versus urban cases at the state-wide level for

207 a 6-year period where locality status was available (2011-2016). This included a total of

208 9,791 cases. Of these, 756 (7.7%) cases were missing a designated locality status (rural or

209 urban). We classified these cases with no locality status as having ‘unspecified’ localities, and

210 excluded these from rural-urban incidence calculations. For the remaining 9,035 cases, we

211 calculated the total proportions and incidence rates for urban and rural cases, using

212 population projections calculated from state-wide rural-urban population data published in

213 2010 (22). At district level, we calculated annual and overall proportions of rural and urban

214 cases per district. Where cases with unspecified localities were included in analyses (Tables

215 1, 2 and S1), the proportion of unspecified localities were indicated. Annual and overall

216 relative risks (RR) of dengue for each individual district were calculated using:

Observed incidence rate 217 RR = Expected incidence rate

218 where the expected incidence rate for each district is based on the mean rate for the state

219 multiplied by the population of each district. A RR value > 1 indicates increased incidence of

220 dengue in that location compared to the expected (mean) incidence, and a value < 1

221 indicates lower than expected dengue incidence.

222

223 9 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

224 Results 225 226 Temporal trends across the state

227 A total of 11,882 dengue cases were reported in Sabah during the 7-year study period, with

228 25 deaths. Cases were reported year-round, with outbreaks commonly occurring in the

229 second half of the year between July and December, sometimes continuing into January and

230 February (Fig 2). Seasonal decomposition analysis showed that, on average, notifications

231 peaked each January, with the highest risk period being between November and March.

232 Smaller peak periods were also observed occasionally in July and October (S1 Fig).

233 234 Fig 2. Temporal pattern of dengue in Sabah, 2010-2016. 235 The monthly number of reported dengue cases per year are shown (primary vertical axis), 236 and the corresponding monthly incidence rate (secondary vertical axis). The change in case 237 definition during the study period is indicated by different colour bars: grey bars during the 238 years 2010-2013 where case diagnoses were predominantly clinically-based (with or without 239 laboratory confirmation), and blue bars for the period 2014-2016 where all cases were 240 laboratory confirmed. 241 242 Outbreaks varied in magnitude between years, with the largest outbreaks in 2010 and from

243 2015-2016 (Fig 2). During these large outbreak years, state-wide annual incidence peaked at

244 between 35 and 43 cases per 100,000, respectively. Conversely, incidence rates dropped to

245 5-9 per 100,000 during the smaller outbreak years between 2011 and 2013. The mean state-

246 wide annual incidence rate across the 7 years was 21 cases per 100,000 people.

247 For the period 2011-2016, state-wide mean annual incidence of dengue in urban localities

248 was 44/100,000 versus 47/100,000 for rural localities, and annual rates of dengue in urban

249 and rural localities often contributed similarly to the overall burden (Fig 3). However, there

250 was a notable difference during the large outbreaks of 2015 and 2016, when the highest

10 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

251 incidence localities appeared to switch between being predominantly urban in 2015 to

252 predominantly rural in 2016.

253 254 Fig 3. State-wide annual incidence of dengue in rural and urban localities, 2011-2016. 255 Annual incidence rates across the state are shown for cases residing in either urban or rural 256 localities, over a 6-year period. 257 258 Demographic trends

259 Analyses of demographic trends across Sabah indicated a slightly higher proportion of male

260 dengue cases (60%) than females (40%). After adjusting for differences in population

261 proportions, incidence rates were not significantly different between the two (29/100,000

262 for males and 20/100,000 for females, p=0.32; Fig 4). This was relatively consistent across all

263 Sabah districts; however, there were some districts with above-average proportions of male

264 cases – in particular, in and (75% and 65% male cases, respectively).

265 266 Fig 4. Incidence of dengue in Sabah by age group and gender, 2010-2016. 267 Age- and gender-adjusted incidence rates across Sabah during the 7-year period are shown 268 for males, females and for both genders. * Age-group 10-29 for both genders had a 269 statistically significantly higher mean rate, p=0.029. 270 271 Older children and young adults were the dominant age groups affected by dengue (Fig 4),

272 with the majority of cases occurring between 10 and 29 years (mean annual incidence of 24

273 cases/100,000; 47% of the burden across age groups), followed by 30-49 years (mean 13

274 cases/100,000/year; 26% of total cases). The median age of all notifications was 25. After

275 adjusting for differences in population proportions, the 10-29 age group had significantly

276 higher incidence than the other age groups (p=0.029). The lowest proportion of notifications

11 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

277 occurred below 10 years of age (mean 6/100,000/year), followed by those 50 years of age

278 and above (8/100,000/year).

279 Spatial trends across districts

280 District-level incidence rates were highly variable each year, with a mean annual rate of 50

281 cases/100,000 (range 19-161 cases/100,000) across the 7 years (Table 1, S2 Fig). High annual

282 variability meant that there was here was no significant difference in mean incidence rates

283 between the districts overall (p=0.462); however, the highest mean incidence rates were

284 found in districts in the west of the state with relatively low human population density,

285 including Kuala Penyu, Nabawan, Tenom and Kota Marudu. These high incidence districts

286 also displayed the greatest extremes in annual dengue rates, ranging between 15 and 942

287 cases/100,000 each year (S2 Fig). The overall relative risks were highest in Kuala Penyu, Kota

288 Marudu and districts (RR=3.5, 2.1 and 1.8, respectively; Table 1). The 4 highest-

289 incidence districts reported a low proportion of cases residing in urban localities (0-12%;

290 Table 1). Lower, less variable incidence rates were recorded from some of the central and

291 eastern districts including Kinabatangan, Tongod, and Tawau (annual incidence range

292 of 3-62 cases/100,000 each year). These districts also had some of the lowest relative risks

293 (RR=0.3, 0.4, 1, and 0.8, respectively; Table 1, S2 Fig), along with a wide range in their

294 proportions of urban cases (4-73%).

12 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

295 Table 1. Summary of population and dengue burden across Sabah, 2010-2016.

Proportion of Severe dengue Human Population Mean annual cases from mean annual Overall relative Number of District population density incidence urban incidence risk dengue deaths (2010) (people/km2) (per 100,000) localities* (per 100,000) Beaufort 66,406 38 0.06 49 0.2 0.9 0 # 106,632 14 0.14 34 0.9 0.8 1 177,735 50 0.11 57 0.2 0.9 0 Kinabatangan 150,327 23 0.04 19 0.0 0.3 0 93,180 67 0.02 52 0.7 1.4 0 Kota Kinabalu 462,963 1,315 0.85 57 0.2 1 5 Kota Marudu 68,289 36 0.02 79 0.4 2.1 1 Kuala Penyu 19,426 43 0.06 161 0.0 3.5 0 Kudat 85,404 66 0.50 52 0.5 1.8 1 Kunak 62,851 55 0.62 33 1.3 1 0 206,861 28 0.42 45 0.8 1.3 2 Nabawan^ 32,309 5 0.00 88 0.0 1.6 0 Papar 128,434 103 0.20 31 0.0 0.6 0 125,913 270 0.51 74 0.6 1.3 1 Pitas 38,764 27 0.29 36 0.0 0.8 0 ** 55,864 1,397 0.32 73 0.0 1 2 95,800 26 0.09 37 0.1 0.7 1 Sandakan 409,056 180 0.80 57 1.0 1.1 4 137,868 120 0.53 46 1.1 1 2 35,764 13 0.14 56 0.0 1.2 0 36,297 27 0.00 37 0.0 0.8 0 Tawau 412,375 67 0.73 33 0.8 0.8 5 Tenom 56,597 13 0.12 84 0.7 1.5 0 Tongod** 36,192 4 0.05 22 1.0 0.4 0 105,435 90 0.26 56 0.4 1.3 0 Total 3,206,742 43 0.48 50 0.5 25 296 *Rural urban locality data was included between 2011-2016; overall proportions were 0.48 urban, 0.44 rural and 0.08 unspecified. 297 # Beluran was formerly known as Labuk Sugut. ^ Nabawan was formerly known as . 298 ** Putatan and Tongod districts only commenced notifications in 2012.

13 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

299 The changing annual spatial trend is shown in Fig 5, which shows high annual and mean

300 incidence rates often occurring in the western districts of Sabah. A shift in dynamics

301 occurred during the large 2015 outbreak, when incidence increased markedly in the more

302 densely populated western districts of Kota Kinabalu, Penampan, Putatan, and in Sandakan

303 and Semporna in the east (Fig 5). The overall urban case proportions in these districts ranged

304 from 53-85%. During 2016, cases from both urban and rural localities contributed to the

305 outbreak, but the greatest overall incidence was in Tenom, Nabawan and Keningau districts,

306 where the majority of cases were from rural localities.

307 308 Fig 5. Annual spatial incidence of dengue in Sabah, 2010-2016. 309 Incidence rates across districts are shown for each year, as well as the overall mean annual 310 incidence during the 7-year period. The 3 major cities of Sabah (Kota Kinabalu, Sandakan and 311 Tawau) are indicated by black circles. 312 313 Severe dengue

314 Of all dengue cases reported over the 7 years, 1.1% were severe (haemorrhagic) dengue

315 cases. The average annual state-wide number of severe cases was 18, although this

316 increased to 28 and 25 cases during 2011 and 2013, respectively, despite these being

317 relatively low incidence years (Figs 2, 6). The greatest proportion of severe cases were

318 concentrated in Sandakan (24%) and Tawau (20%) districts on the eastern side of the state,

319 with the highest severe dengue incidence found in Kunak, Sandakan and Tongod (Fig 6; Table

320 1). The lowest proportion and incidence of severe dengue was observed in the western

321 districts, several of which recorded zero severe cases, despite recording high overall dengue

322 incidence (Figs 5, 6; Table 1). Severe dengue occurred evenly across both genders and age

323 groups, although the burden was highest for age groups under 30 years, with the largest

324 proportion (35%) reported within the 10-19 years age group. There were 9 severe dengue

14 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

325 deaths during the study period, 4 of which were in . Deaths from severe

326 dengue occurred consistently across years, genders and age groups.

327 328 Fig 6. Total severe dengue notifications by district, 2010-2016. 329 Total number of severe dengue cases reported for each district during the 7-year period. The 330 3 major cities of Sabah (Kota Kinabalu, Sandakan and Tawau) are indicated. 331 332 Entomological factors

333 Entomological data collected within the 2015-2016 outbreaks indicated that the partially

334 sylvatic vector Aedes albopictus was the predominant species detected in larval collections

335 from both rural and urban case residences (Table 2). Of 719 dengue case residences that

336 were inspected as part of active surveillance in 2015-2016, 618 were found to contain

337 mosquito larvae (HI=86%). Of those, Ae. albopictus larvae were identified from 394

338 residences (HI=64%), either alone (383 residences) or with Ae. aegypti (11 residences).

339 Conversely, 94 residences were positive for Ae. aegypti (83 alone, 11 with Ae. Albopictus;

340 HI=15%), 33 were positive for Culex species (HI=5%), and 108 larval samples could not be

341 identified (17%).

15 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

342 Table 2. Mosquito larvae species collected from case residences in Sabah during 2015-2016. Total No. of larvae- No. of case Number of larvae-positive residences with specific species present (HI) number of positive case House Index Locality residences cases in residences (HI) Ae. Ae. Ae. aegypti & inspected Culex spp. Undetermined 2015-2016 (all species) aegypti albopictus Ae. albopictus* Urban residences 3,157 255 206 0.81 47 (0.23) 142 (0.69) 7 (0.03) 6 (0.03) 4 (0.02) Rural residences 2,824 437 388 0.89 33 (0.09) 225 (0.58) 4 (0.01) 26 (0.07) 100 (0.26) Locality unspecified 565 27 24 0.89 3 (0.13) 16 (0.67) 0 (0.0) 1 (0.04) 4 (0.17) Total 6,546 719 618 0.86 83 (0.13) 383 (0.62) 11 (0.02) 33 (0.05) 108 (0.17) 343 HI = proportion of residences positive for mosquito larvae, calculated as number of residences with larvae/number of residences inspected. 344 * Both species found breeding together in one household. 345 Species were undetermined if the larvae failed to survive to adults to be identified, or if identification was pending/incomplete.

16 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

346 The specific districts where case residences were inspected are detailed in S1 Table.

347 Inspections were conducted in 21/25 districts, though the majority were conducted in

348 Tawau (158 inspections; with 96 in urban localities) and Nabawan (107 inspections; all in

349 rural localities). The majority of residences positive for Ae. aegypti were in the east coast

350 districts of Tawau (larvae found in 60/158 residences) and Lahad Datu (18/43 larvae-positive

351 residences). Residences in Tawau and Lahad Datu districts together comprised 80% of all

352 Sabah residences where Ae. aegypti was identified. Ae. albopictus was prevalent across both

353 urban and rural residences of most of districts surveyed. The majority of residences positive

354 for Ae. albopictus larvae were also in Tawau (101/158 residences), followed by Penampang

355 (52/58 larvae-positive residences) and Nabawan and Keningau (46 residences each; S1

356 Table).

357

358 Discussion 359 360 Spatial and temporal trends

361 In recent years, the scale of epidemics in South East Asia has increased in both urban and

362 rural areas (28-32). We found an overall increasing incidence trend in the Malaysian state of

363 Sabah, with the highest risk period occurring annually between November and March. While

364 spatial trends in dengue incidence varied from year to year, the most intense transmission

365 across all years occurred in districts along the western coast of Sabah. The timing of large

366 epidemic years in Sabah (2010, 2015 and 2016) was consistent with patterns observed at

367 national and regional levels during the same period (wider Malaysia, Indonesia, )

368 (33, 34). This suggests shared seasonal influences on outbreak occurrence although,

17 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

369 especially in the tropics, the causative relationship between temperature, rainfall and

370 transmission remains poorly understood (35, 36).

371 We also noted consistently high incidence in rural localities across the study period, as well

372 as a shift in the spatial dominance of urban versus rural localities during the large outbreaks

373 of 2015-2016. Our findings suggest that high density, urbanised areas are not necessarily the

374 primary drivers of ongoing epidemics in Sabah, and that factors other than population size

375 may drive the risk in rural areas. Recent studies in other areas of Malaysia have also shown

376 that dengue infection can occur at equivalent rates in both rural and urban areas (37, 38). It

377 could be that the threshold human density required to maintain transmission may be lower

378 than previously thought, although human movement (which may relate to population

379 density) between urban and rural areas is also likely to have influenced the patterns we

380 observed (39, 40).

381 Rural dominance of dengue has also been observed elsewhere in the region, including in

382 Cambodia, Thailand, Vietnam and Sri Lanka (41-44). These countries have all reported

383 epidemics spreading between rural and urban areas – in both directions – via human or

384 mosquito movement, facilitated by favourable climatic conditions. Other potential

385 influences on both urban and rural dengue transmission in Sabah might include water

386 storage practices, mosquito vector ecology and sociocultural factors (32, 45-48). The relative

387 importance of some of these risk factors in mediating dengue transmission is still not well

388 understood, even in urban areas (49, 50). Understanding the relationships between these

389 risk factors may be challenging to disentangle in different environments, and especially

390 where urban and rural areas are highly interconnected; however, knowledge of these

391 dynamics may be important to optimise the design and targeting of dengue control

18 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

392 strategies. This is especially important given the potential for ongoing outbreaks in both

393 urban and rural areas of Sabah.

394 Demographic factors

395 Our findings indicated that the age-related dengue risk in Sabah was in line with regional

396 trends indicating a transition from children to adults being disproportionately affected by

397 dengue (51). Incidence was higher for males than for females across all districts of the state,

398 and was significantly higher for both genders in the 10-29 age group. This higher risk may

399 suggest that a larger proportion of people in this age-group (and possibly males in particular)

400 were either engaged in outdoor activities and/or being occupationally exposed. The

401 agriculture sector is the major employment sector in Sabah, and this type of work may

402 increase exposure to mosquitoes (52, 53). The particularly high proportion of males affected

403 in Tongod and Kinabatangan districts may reflect the fact that both are very rural, and may

404 have a larger proportion of men engaged in agricultural or recreational outdoor activities.

405 Outdoor activities, especially those in close proximity to forests or forest edges, are thought

406 to increase the risk of being bitten by the abundant exophilic vector, Ae. albopictus (54, 55).

407 However, studies in peninsula Malaysia have shown that Ae. albopictus can also adapt to

408 indoor urban environments (56, 57); hence, the high proportion of cases we observed in 10-

409 29 year olds could also suggest infection in indoor environments at home or at school.

410 Further investigation to identify the specific factors associated with infection risk in Sabah

411 may be useful to inform prevention strategies for this high-risk group.

412 Severe dengue

413 Changing demographic or immunological factors may also explain the observed pattern of

414 severe dengue in our study. Severe dengue showed a decreasing trend over time (despite

415 the overall increasing trend in incidence rates), but with the highest risk localised to two

19 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

416 main regions of the state: the eastern districts of Tawau and Sandakan. These districts

417 include major urbanised cities as well as rural surrounding areas, and comprised relatively

418 low dengue rates compared to the west of the state. The reasons for this spatial

419 concentration of cases in these eastern districts is unknown, though it’s possible that a

420 serotype switch from DENV 4 to DENV 1 reported to have occurred in Sandakan between

421 2013 and 2016 may have contributed (12). It might also be possible that different areas of

422 Sabah experience serotype changes more or less frequently depending on levels and

423 direction of population movement (33, 58). Surveillance information regarding which virus

424 serotypes and genotypes were circulating in Sabah was not available in this study, so we

425 were unable to assess the potential contribution of virus circulation patterns to the trends

426 we observed. Assessing serological surveillance data alongside epidemiological data in future

427 studies in Sabah could aid predictions of severe disease risk (59, 60).

428 Entomological factors

429 During the large outbreak period between 2015 and 2016, our entomological surveillance

430 data indicated a striking association between the presence of the mosquito vector Ae.

431 albopictus relative to Ae. aegypti, in both urban and rural case residences in the majority of

432 the state. Interestingly, the eastern districts of Sabah state appeared to have a higher

433 proportion of Ae. aegypti compared to the rest of the state, although overall dengue

434 incidence was lower on the east coast. Interestingly, this finding was consistent with those of

435 early entomological surveys of Sabah in the 1970’s, which reported higher numbers of Ae.

436 aegypti on the east coast and lower abundance on the west coast (61, 62). In those studies,

437 the greater presence of Ae. aegypti in the east was thought to be due to more frequent

438 travel by boat between east coast settlements for fishing and trade.

20 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

439 Although Ae. aegypti is generally considered responsible for most dengue transmission in

440 South East Asia (63, 64), Ae. albopictus is more common than Ae. aegypti across Malaysia

441 and Borneo (52, 54, 55, 65). Its competence for specific dengue genotypes, its abundance in

442 both rural and urban areas, its biting behaviour and its diverse aquatic habitat may all

443 account for patterns of mosquito-human contact and subsequent transmission in Sabah (48,

444 66). The presence of natural and artificial larval habitats for Ae. albopictus have previously

445 been associated with epidemic disease in both urban and rural areas of Malaysia (56, 57, 67)

446 despite the fact that globally, Ae. aegypti is undoubtedly the predominant vector driving

447 epidemics (68, 69). Urban dominance of Ae. albopictus has also been observed, at least

448 seasonally, in parts of Thailand, southern China and other South East Asian countries (70-72).

449 Given the likely role of Ae. albopictus in mediating dengue epidemics in Sabah, vector

450 control strategies may have to be expanded to include both Ae. aegypti and Ae. albopictus.

451 Because Ae. albopictus is commonly characterized as more exophagic and exophilic than Ae.

452 aegypti, and exploiting a wider range of hosts and habitats in peri-urban and rural

453 environments, targeting outdoor resting sites of adult Ae. albopictus may be a useful control

454 strategy in Sabah (73-75).

455 Limitations

456 The main caveat to our findings is that the changes in dengue case definition in Malaysia in

457 2014 may have influenced the trends reported here, in terms of either under- or over-

458 reporting of cases. Reduced reliance on clinical symptoms for case notification from 2014

459 onwards would be expected to reduce notifications dramatically but, in fact, a dramatic

460 increase in cases were recorded. It is possible that prior to that date, the lack of resources

461 for testing or notifying dengue, or other socioeconomic factors, may have resulted in under-

462 reporting (76). It is also possible that increases in diagnostic testing from 2014 were not

21 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

463 uniform across all districts, and/or that additional reporting inconsistencies may have

464 impacted our observations.

465 Conclusions

466 The rising magnitude of dengue in Sabah in both rural and urban areas suggests that a better

467 understanding of dengue transmission across different environments is needed. Our findings

468 support the notion that dengue epidemics can be both urban and rural environment-driven,

469 and suggest risk factors that may be of use for clinicians, public health practitioners and

470 vector control teams. The trends observed in Sabah indicate that localized ecological,

471 human, virus and vector dynamics may be predictive of dengue epidemics irrespective of

472 urban and rural environment. In Sabah, as with many countries of South East Asia, there is

473 likely a complex interplay of these factors operating in both rural and urban areas, and these

474 probably overlap (10, 29). Considering the ongoing expansion of dengue endemicity and

475 burden in the region, proactive strategies to increase understanding of the complex and

476 evolving epidemiological factors underlying dengue risk across varied environments are

477 critical.

478

479 Acknowledgements 480 481 The authors would like to acknowledge the contribution of the Sabah Department of Health,

482 Ministry of Health, Malaysia for making dengue notification data available. We also thank

483 the Director General of Health Malaysia for the permission to publish this paper. We are

484 grateful for assistance and input provided by Nicholas Anstey, Matthew Grigg, Kimberley

485 Fornace, Christopher Wilkes, Eloise Stephenson and Andrea Rabellino. The author(s)

22 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

486 received no specific funding for this work. The authors declare no conflict of interest in

487 conducting this study.

488

23 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

489 References 490 491 1. Bhatt S, Gething PW, Brady OJ, Messina JP, Farlow AW, Moyes CL, et al. The global distribution 492 and burden of dengue. Nature. 2013;496(7446):504-7. 493 2. WHO. Dengue and severe dengue - fact sheet 2014 [Available from: 494 http://www.who.int/mediacentre/factsheets/fs117/en/. 495 3. Kyle JL, Harris E. Global spread and persistence of dengue. Annu Rev Microbiol. 2008;62:71- 496 92. 497 4. Messina JP, Brady OJ, Scott TW, Zou C, Pigott DM, Duda KA, et al. Global spread of dengue 498 virus types: mapping the 70 year history. Trends Microbiol. 2014;22(3):138-46. 499 5. Rico-Hesse R, Harrison LM, Salas RA, Tovar D, Nisalak A, Ramos C, et al. Origins of dengue type 500 2 viruses associated with increased pathogenicity in the Americas. Virology. 1997;230(2):244- 501 51. 502 6. Vasilakis N, Weaver SC. The history and evolution of human dengue emergence. Adv Virus 503 Res. 2008;72:1-76. 504 7. Hussin N, Jaafar J, Naing NN, Mat HA, Muhamad AH, Mamat MN. A review of dengue fever 505 incidence in Kota Bharu, , Malaysia during the years 1998-2003. Southeast Asian J 506 Trop Med Public Health. 2005;36(5):1179-86. 507 8. Shekhar KC, Huat OL. Epidemiology of dengue/dengue hemorrhagic fever in Malaysia--a 508 retrospective epidemiological study. 1973-1987. Part II: Dengue fever (DF). Asia Pac J Public 509 Health. 1992;6(3):126-33. 510 9. Mohd-Zaki AH, Brett J, Ismail E, L'Azou M. Epidemiology of dengue disease in Malaysia (2000- 511 2012): a systematic literature review. PLoS Negl Trop Dis. 2014;8(11):e3159. 512 10. Vasilakis N, Cardosa J, Hanley KA, Holmes EC, Weaver SC. Fever from the forest: prospects for 513 the continued emergence of sylvatic dengue virus and its impact on public health. Nat Rev 514 Microbiol. 2011;9(7):532-41. 515 11. Liu W, Pickering P, Duchene S, Holmes EC, Aaskov JG. Highly Divergent Dengue Virus Type 2 in 516 Traveler Returning from Borneo to Australia. Emerg Infect Dis. 2016;22(12):2146-8. 517 12. Gintarong TJ, Emran A, Sherin A, Thein TT, Aung TS. Circulation of all dengue virus serotypes 518 during dengue outbreak in Sandakan, Sabah, Malaysia (2016). J Vector Borne Dis. 519 2018;55(2):168-71. 520 13. Ahmad R, Suzilah I, Wan Najdah WMA, Topek O, Mustafakamal I, Lee HL. Factors determining 521 dengue outbreak in Malaysia. PLoS One. 2018;13(2):e0193326. 522 14. Statistics MDo. Labour Force and Social Statistics 2018 [Available from: 523 https://www.dosm.gov.my/v1/index.php?r=column/cthemeByCat&cat=152&bul_id=aktTTjhh 524 RHd1aHBCZGF1N01aaTl3dz09&menu_id=U3VPMldoYUxzVzFaYmNkWXZteGduZz09. 525 15. Gaveau DL, Sloan S, Molidena E, Yaen H, Sheil D, Abram NK, et al. Four decades of forest 526 persistence, clearance and logging on Borneo. PLoS One. 2014;9(7):e101654. 527 16. Fornace KM, Abidin TR, Alexander N, Brock P, Grigg MJ, Murphy A, et al. Association between 528 Landscape Factors and Spatial Patterns of Plasmodium knowlesi Infections in Sabah, Malaysia. 529 Emerg Infect Dis. 2016;22(2):201-8. 530 17. Cooper DJ, Rajahram GS, William T, Jelip J, Mohammad R, Benedict J, et al. Plasmodium 531 knowlesi malaria in Sabah, Malaysia, 2015-2017: ongoing increase in incidence despite near- 532 elimination of the human-only Plasmodium species. Clin Infect Dis. 2019. 533 18. Rajahram GS, Hameed AA, Menon J, William T, Tambyah PA, Yeo TW. Case report: two human 534 Streptococcus suis infections in Borneo, Sabah, Malaysia. BMC Infect Dis. 2017;17(1):188.

24 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

535 19. Centers for Disease C, Prevention. Update: outbreak of acute febrile illness among athletes 536 participating in Eco-Challenge-Sabah 2000--Borneo, Malaysia, 2000. MMWR Morb Mortal 537 Wkly Rep. 2001;50(2):21-4. 538 20. Tappe D, Nachtigall S, Kapaun A, Schnitzler P, Gunther S, Schmidt-Chanasit J. Acute Zika virus 539 infection after travel to Malaysian Borneo, September 2014. Emerg Infect Dis. 2015;21(5):911- 540 3. 541 21. Statistics MDo. Compendium of Environment Statistics 2017 [Available from: 542 https://newss.statistics.gov.my/newss- 543 portalx/ep/epFreeDownloadContentSearch.seam?cid=181040. 544 22. Statistics MDo. Population Distribution and Basic Demographic Characteristics, Population and 545 Housing Census of Malaysia 2010. Kuala Lumpur; 2011. 546 23. Organization WH. Dengue guidelines for diagnosis, treatment, prevention and control. In: 547 (TDR) SPfRaTiTD, editor. France: WHO, Geneva, Switzerland; 2009. 548 24. Organization WH. Handbook for clinical management of dengue. Geneva, Switzerland: World 549 Health Organization; 2012. 550 25. Health MMo. Guidelines for Health Management and Disease Control. , 551 Malaysia2013. 552 26. (WHO) WHO. Dengue control, vector surveillance: WHO; 2019 [Available from: 553 https://www.who.int/denguecontrol/monitoring/vector_surveillance/en/. 554 27. Macaulay FR. The Smoothing of Time Series. New York: The National Bureau of Economic 555 Research; 1931. 556 28. Vong S, Khieu V, Glass O, Ly S, Duong V, Huy R, et al. Dengue incidence in urban and rural 557 Cambodia: results from population-based active fever surveillance, 2006-2008. PLoS Negl Trop 558 Dis. 2010;4(11):e903. 559 29. Muhammad Azami NA, Salleh SA, Neoh HM, Syed Zakaria SZ, Jamal R. Dengue epidemic in 560 Malaysia: Not a predominantly urban disease anymore. BMC Res Notes. 2011;4:216. 561 30. Barde PV, Kori BK, Shukla MK, Bharti PK, Chand G, Kumar G, et al. Maiden outbreaks of 562 dengue virus 1 genotype III in rural central India. Epidemiol Infect. 2015;143(2):412-8. 563 31. Dubot-Peres A, Vongphrachanh P, Denny J, Phetsouvanh R, Linthavong S, Sengkeopraseuth B, 564 et al. An epidemic of dengue-1 in a remote village in rural Laos. PLoS Negl Trop Dis. 565 2013;7(8):e2360. 566 32. Schmidt WP, Suzuki M, Thiem VD, White RG, Tsuzuki A, Yoshida LM, et al. Population density, 567 water supply, and the risk of dengue fever in Vietnam: cohort study and spatial analysis. PLoS 568 Med. 2011;8(8):e1001082. 569 33. Mia MS, Begum RA, Er AC, Abidin RD, Pereira JJ. Trends of dengue infections in Malaysia, 570 2000-2010. Asian Pac J Trop Med. 2013;6(6):462-6. 571 34. Arima Y, Edelstein ZR, Han HK, Matsui T. Epidemiologic update on the dengue situation in the 572 Western Pacific Region, 2011. Western Pac Surveill Response J. 2013;4(2):47-54. 573 35. Servadio JL, Rosenthal SR, Carlson L, Bauer C. Climate patterns and mosquito-borne disease 574 outbreaks in South and Southeast Asia. J Infect Public Health. 2017. 575 36. de Castro DB, Sampaio VS, de Albuquerque BC, Pinto RC, Sadahiro M, Dos Passos RA, et al. 576 Dengue epidemic typology and risk factors for extensive epidemic in Amazonas state, Brazil, 577 2010-2011. BMC Public Health. 2018;18(1):356. 578 37. Chew CH, Woon YL, Amin F, Adnan TH, Abdul Wahab AH, Ahmad ZE, et al. Rural-urban 579 comparisons of dengue seroprevalence in Malaysia. BMC Public Health. 2016;16(1):824.

25 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

580 38. Dhanoa A, Hassan SS, Jahan NK, Reidpath DD, Fatt QK, Ahmad MP, et al. Seroprevalence of 581 dengue among healthy adults in a rural community in Southern Malaysia: a pilot study. Infect 582 Dis Poverty. 2018;7(1):1. 583 39. Stoddard ST, Forshey BM, Morrison AC, Paz-Soldan VA, Vazquez-Prokopec GM, Astete H, et al. 584 House-to-house human movement drives dengue virus transmission. Proc Natl Acad Sci U S A. 585 2013;110(3):994-9. 586 40. Vazquez-Prokopec GM, Montgomery BL, Horne P, Clennon JA, Ritchie SA. Combining contact 587 tracing with targeted indoor residual spraying significantly reduces dengue transmission. Sci 588 Adv. 2017;3(2):e1602024. 589 41. Rabaa MA, Ty Hang VT, Wills B, Farrar J, Simmons CP, Holmes EC. Phylogeography of recently 590 emerged DENV-2 in southern Viet Nam. PLoS Negl Trop Dis. 2010;4(7):e766. 591 42. Teurlai M, Huy R, Cazelles B, Duboz R, Baehr C, Vong S. Can human movements explain 592 heterogeneous propagation of dengue fever in Cambodia? PLoS Negl Trop Dis. 593 2012;6(12):e1957. 594 43. Thongsripong P, Green A, Kittayapong P, Kapan D, Wilcox B, Bennett S. Mosquito vector 595 diversity across habitats in central Thailand endemic for dengue and other arthropod-borne 596 diseases. PLoS Negl Trop Dis. 2013;7(10):e2507. 597 44. Bodinayake CK, Tillekeratne LG, Nagahawatte A, Devasiri V, Kodikara Arachichi W, Strouse JJ, 598 et al. Emergence of Epidemic Dengue-1 Virus in the Southern Province of Sri Lanka. PLoS Negl 599 Trop Dis. 2016;10(10):e0004995. 600 45. Arunachalam N, Tana S, Espino F, Kittayapong P, Abeyewickreme W, Wai KT, et al. Eco-bio- 601 social determinants of dengue vector breeding: a multicountry study in urban and periurban 602 Asia. Bull World Health Organ. 2010;88(3):173-84. 603 46. Banu S, Hu W, Hurst C, Tong S. Dengue transmission in the Asia-Pacific region: impact of 604 climate change and socio-environmental factors. Trop Med Int Health. 2011;16(5):598-607. 605 47. Cheong YL, Leitao PJ, Lakes T. Assessment of land use factors associated with dengue cases in 606 Malaysia using Boosted Regression Trees. Spat Spatiotemporal Epidemiol. 2014;10:75-84. 607 48. Rohani A, Suzilah I, Malinda M, Anuar I, Mohd Mazlan I, Salmah Maszaitun M, et al. Aedes 608 larval population dynamics and risk for dengue epidemics in Malaysia. Trop Biomed. 609 2011;28(2):237-48. 610 49. Anders KL, Nga le H, Thuy NT, Ngoc TV, Tam CT, Tai LT, et al. Households as foci for dengue 611 transmission in highly urban Vietnam. PLoS Negl Trop Dis. 2015;9(2):e0003528. 612 50. Ali M, Wagatsuma Y, Emch M, Breiman RF. Use of a geographic information system for 613 defining spatial risk for dengue transmission in Bangladesh: role for Aedes albopictus in an 614 urban outbreak. Am J Trop Med Hyg. 2003;69(6):634-40. 615 51. Cummings DA, Iamsirithaworn S, Lessler JT, McDermott A, Prasanthong R, Nisalak A, et al. The 616 impact of the demographic transition on dengue in Thailand: insights from a statistical analysis 617 and mathematical modeling. PLoS Med. 2009;6(9):e1000139. 618 52. Young KI, Mundis S, Widen SG, Wood TG, Tesh RB, Cardosa J, et al. Abundance and 619 distribution of sylvatic dengue virus vectors in three different land cover types in Sarawak, 620 Malaysian Borneo. Parasit Vectors. 2017;10(1):406. 621 53. Jeffree SM, Ahmed K, Safian N, Hassan R, Mihat O, Lukman KA, et al. Falciparum Malaria 622 Outbreak in Sabah Linked to an Immigrant Rubber Tapper. Am J Trop Med Hyg. 2018;98(1):45- 623 50. 624 54. Brown R, Hing CT, Fornace K, Ferguson HM. Evaluation of resting traps to examine the 625 behaviour and ecology of mosquito vectors in an area of rapidly changing land use in Sabah, 626 Malaysian Borneo. Parasit Vectors. 2018;11(1):346.

26 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

627 55. Rozilawati H, Tanaselvi K, Nazni WA, Mohd Masri S, Zairi J, Adanan CR, et al. Surveillance of 628 Aedes albopictus Skuse breeding preference in selected dengue outbreak localities, peninsular 629 Malaysia. Trop Biomed. 2015;32(1):49-64. 630 56. Dieng H, Saifur RG, Hassan AA, Salmah MR, Boots M, Satho T, et al. Indoor-breeding of Aedes 631 albopictus in northern peninsular Malaysia and its potential epidemiological implications. 632 PLoS One. 2010;5(7):e11790. 633 57. Roslan MA, Shafie A, Ngui R, Lim YA, Sulaiman WY. Vertical infestation of the dengue vectors 634 Aedes aegypti and Aedes albopictus in apartments in Kuala Lumpur, Malaysia. J Am Mosq 635 Control Assoc. 2013;29(4):328-36. 636 58. Ng LC, Chem YK, Koo C, Mudin RNB, Amin FM, Lee KS, et al. 2013 dengue outbreaks in 637 Singapore and Malaysia caused by different viral strains. Am J Trop Med Hyg. 638 2015;92(6):1150-5. 639 59. Fried JR, Gibbons RV, Kalayanarooj S, Thomas SJ, Srikiatkhachorn A, Yoon IK, et al. Serotype- 640 specific differences in the risk of dengue hemorrhagic fever: an analysis of data collected in 641 Bangkok, Thailand from 1994 to 2006. PLoS Negl Trop Dis. 2010;4(3):e617. 642 60. Tan KK, Zulkifle NI, Abd-Jamil J, Sulaiman S, Yaacob CN, Azizan NS, et al. Disruption of 643 predicted dengue virus type 3 major outbreak cycle coincided with switching of the dominant 644 circulating virus genotype. Infect Genet Evol. 2017;54:271-5. 645 61. Macdonald WW, Rajapaksa N. A survey of the distribution and relative prevalence of Aedes 646 aegypti in Sabah, Brunei, and Sarawak. Bull World Health Organ. 1972;46(2):203-9. 647 62. King JHL. A re-survey of Aedes aegypti and Aedes albopictus in Sabah, Malaysia. Health 648 Studies in Indonesia. 1978;11(1):45-51. 649 63. Whitehorn J, Kien DT, Nguyen NM, Nguyen HL, Kyrylos PP, Carrington LB, et al. Comparative 650 Susceptibility of Aedes albopictus and Aedes aegypti to Dengue Virus Infection After Feeding 651 on Blood of Viremic Humans: Implications for Public Health. J Infect Dis. 2015;212(8):1182-90. 652 64. Brady OJ, Golding N, Pigott DM, Kraemer MU, Messina JP, Reiner RC, Jr., et al. Global 653 temperature constraints on Aedes aegypti and Ae. albopictus persistence and competence for 654 dengue virus transmission. Parasit Vectors. 2014;7:338. 655 65. Black WCt, Hawley WA, Rai KS, Craig GB, Jr. Breeding structure of a colonizing species: Aedes 656 albopictus (Skuse) in peninsular Malaysia and Borneo. Heredity (Edinb). 1988;61 ( Pt 3):439- 657 46. 658 66. Joanne S, Vythilingam I, Teoh BT, Leong CS, Tan KK, Wong ML, et al. Vector competence of 659 Malaysian Aedes albopictus with and without Wolbachia to four dengue virus serotypes. Trop 660 Med Int Health. 2017;22(9):1154-65. 661 67. Chen CD, Nazni WA, Lee HL, Seleena B, Mohd Masri S, Chiang YF, et al. Mixed breeding of 662 Aedes aegypti (L.) and Aedes albopictus Skuse in four dengue endemic areas in Kuala Lumpur 663 and , Malaysia. Trop Biomed. 2006;23(2):224-7. 664 68. Lambrechts L, Scott TW, Gubler DJ. Consequences of the expanding global distribution of 665 Aedes albopictus for dengue virus transmission. PLoS Negl Trop Dis. 2010;4(5):e646. 666 69. Gratz NG. Critical review of the vector status of Aedes albopictus. Med Vet Entomol. 667 2004;18(3):215-27. 668 70. Li Y, Kamara F, Zhou G, Puthiyakunnon S, Li C, Liu Y, et al. Urbanization increases Aedes 669 albopictus larval habitats and accelerates mosquito development and survivorship. PLoS Negl 670 Trop Dis. 2014;8(11):e3301. 671 71. Peng HJ, Lai HB, Zhang QL, Xu BY, Zhang H, Liu WH, et al. A local outbreak of dengue caused by 672 an imported case in Dongguan China. BMC Public Health. 2012;12:83.

27 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

673 72. Xu G, Dong H, Shi N, Liu S, Zhou A, Cheng Z, et al. An outbreak of dengue virus serotype 1 674 infection in Cixi, Ningbo, People's Republic of China, 2004, associated with a traveler from 675 Thailand and high density of Aedes albopictus. Am J Trop Med Hyg. 2007;76(6):1182-8. 676 73. Muzari MO, Devine G, Davis J, Crunkhorn B, van den Hurk A, Whelan P, et al. Holding back the 677 tiger: Successful control program protects Australia from Aedes albopictus expansion. PLoS 678 Negl Trop Dis. 2017;11(2):e0005286. 679 74. Delatte H, Paupy C, Dehecq JS, Thiria J, Failloux AB, Fontenille D. [Aedes albopictus, vector of 680 chikungunya and dengue viruses in Reunion Island: biology and control]. Parasite. 681 2008;15(1):3-13. 682 75. Valerio L, Marini F, Bongiorno G, Facchinelli L, Pombi M, Caputo B, et al. Host-feeding patterns 683 of Aedes albopictus (Diptera: Culicidae) in urban and rural contexts within Rome province, 684 Italy. Vector Borne Zoonotic Dis. 2010;10(3):291-4. 685 76. Beatty ME, Stone A, Fitzsimons DW, Hanna JN, Lam SK, Vong S, et al. Best practices in dengue 686 surveillance: a report from the Asia-Pacific and Americas Dengue Prevention Boards. PLoS 687 Negl Trop Dis. 2010;4(11):e890. 688

28 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.

689 Supporting Information 690 691 S1 Fig. Seasonal decomposition of incidence rates in Sabah, 2010-2016. 692 The seasonal trend of dengue is shown in panel A, with the largest seasonal peak occurring on 693 average between Nov and May each year (indicated by vertical black lines). The additional 694 components separated from the seasonal trend during the decomposition procedure are also 695 indicated in panels B-D (cyclical component (B), irregular component (C) and overall smoothed 696 trend (D)). 697 698 S2 Fig. Variation in dengue incidence across Sabah districts, 2010-2016. 699 Dengue mean annual incidence rates across all years are plotted, in order of highest-lowest 700 mean annual incidence rate, showing the mean (x), median (line), and the range of rates (upper 701 and lower whiskers) across the years. 702 703 S1 Table. Entomological surveillance of case residences by district, 2015-2016. 704 705 S1 Checklist: STROBE Checklist for observational studies.

29 bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/657031; this version posted May 31, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.