key: cord-0292656-jtfgxt8i authors: Ngwa, M. C.; Ihekweazu, C.; Okwor, T. J.; Williams, N.; Yennan, S.; Elimian, K.; Karaye, N. Y.; Oche, J. A.; Bello, I. W.; Sack, D. A. title: The micro-hotspots of cholera in Kano State, Nigeria, 2010-2019:analysis of patient characteristics, Spatio-temporal patterns and contextual determinants at the ward level. date: 2021-08-23 journal: nan DOI: 10.1101/2021.08.20.21262313 sha: 449de12bb133dfbde2cead7f2f0e537db91af547 doc_id: 292656 cord_uid: jtfgxt8i Cholera is endemic in Nigeria, and Kano State reports outbreaks yearly with a case fatality rate (CFR) of 3.3% from 2010 to 2019. The lack of data at ward level has enabled the disease to evade focused interventions. The goal of this study was to describe the geographic distributions, care-seeking behaviors, Spatio-temporal cluster patterns of the micro-hotspots (hotspots wards) linked with suspected and confirmed cases and deaths of cholera in Kano State. Suspected and confirmed cholera morbidity and mortality at the ward level from 2010-2019 were acquired from the Nigeria Centre for Disease Control. Population and waterbody data were obtained from the Nigeria Expanded Program on Immunization and online, respectively. Data analysis used SaTScan and methods recommended by the Global Task Force on Cholera Control. During these ten years, 18,483 suspected and confirmed cases (617 deaths) were reported with 67.7% of the cases and 72% of the deaths from rural wards. The ages of the cases ranged from 1 month to 100 years with a distribution skewed to the older years. CFRs were statistically higher in the <5-year olds compared to those >14 years (p-value = 0.0005). For 2010-2019, gender was statistically associated with cholera outcome (survived/died) (p-value = 0.0006), and women in the rural setting disproportionately died from cholera than women in the urban area (p-value = 0.003). Cholera severity, as measured by hospitalization and death, was higher in the urban (77.4%) compared with the rural (53.4%) setting with the highest severity (84.7%) registered among those >14 years. Rapid Diagnostic Tests (RDT) were performed in 1.3% (249) samples of all suspected cases and ranged from 0.7% among the 5-14 year-olds in the rural to 3.5 % among the < 5-year-olds in the urban areas. Of the stool samples collected, 62.7% tested positive for V. cholerae using RDT. The positivity rate was least in the urban setting amongst the <5 years (41.2%) while care-seeking-behavior ranged from 52.1% in the urban to 82.7% in the rural settings. Seasonal patterns of disease often differed between urban and rural settings with outbreaks occurring in both the dry and rainy seasons, but with more intense transmission occurring during the rainy season from week 22 (early June) to week 40 (late September). A Spatio-temporal clustering analysis detected 168 micro-hotspots out of 404 wards, with a population of 4,876,254, having a significantly higher risk (relative risk 1.01-18.73) compared to the State as a whole. While 79 micro-hotspots with a population of 2,119,974 had a RR [≥] 2. The micro-hotspots tended to cluster around waterbodies. SaTScan and GTFCC methods generally agreed in micro-hotspots detection. This study shows the epidemiology of cholera in Kano State differs between urban and rural settings and that hotspot maps at the ward level, not hotpots maps at the Local Government Area level, are best suited for targeting interventions including vaccines. Appropriate studies are needed to further delineate the urban and rural divide of outbreaks but targeting interventions to the identified high-priority micro-hotspots will facilitate cholera elimination from the state. During these ten years, 18,483 suspected and confirmed cases (617 deaths) were reported with 67.7% of the 48 cases and 72% of the deaths from rural wards. The ages of the cases ranged from 1 month to 100 years with a 49 distribution skewed to the older years. CFRs were statistically higher in the <5-year olds compared to those 50 >14 years (p-value = 0.0005). For 2010-2019, gender was statistically associated with cholera outcome 51 (survived/died) (p-value = 0.0006), and women in the rural setting disproportionately died from cholera than 52 women in the urban area (p-value = 0.003). Cholera severity, as measured by hospitalization and death, was 53 higher in the urban (77.4%) compared with the rural (53.4%) setting with the highest severity (84.7%) 54 registered among those >14 years. Rapid Diagnostic Tests (RDT) were performed in 1.3% (249) samples of all 55 suspected cases and ranged from 0.7% among the 5-14 year-olds in the rural to 3.5 % among the < 5-year-olds 56 in the urban areas. Of the stool samples collected, 62.7% tested positive for V. cholerae using RDT. The 57 positivity rate was least in the urban setting amongst the <5 years (41.2%) while care-seeking-behavior ranged 58 from 52.1% in the urban to 82.7% in the rural settings. Seasonal patterns of disease often differed between 59 urban and rural settings with outbreaks occurring in both the dry and rainy seasons, but with more intense 60 Introduction 125 126 Cholera is a waterborne infection that makes over a million people sick yearly with an estimated 95,000 127 deaths worldwide [1] . In the last decade, Sub-Saharan Africa (SSA) reported the greatest cases and deaths with 128 the Democratic Republic of Congo and Nigeria contributing the greatest burden of cases in Africa. Nigeria Characteristics of cholera's epidemiological patterns have been described [9] [10] [11] [12] . The Spatio-temporal 136 characteristics of the 2010 outbreak that started in Borno State and grew in magnitude to spread to entire 137 Northern Nigeria in three waves appeared to be amplified by flooding [9] . Modeling of hospital case data from 138 Kano, Sokoto, and the Zamfara States between 1991 and 2011 demonstrate two peaks of disease transmission 139 between April and August [9] [10] [11] [12] . Increases in temperature, rainfall, poverty, and population density were 140 found to be associated with both cholera cases and deaths [9] . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint pre-emptively vaccinate populations in identified hotspots nationwide between 2018 and 2023 [16] . The 149 current hotspot maps in the NSPACC, as well as those recently published for Kano State [8] , are all based on 150 the Local Government Areas (LGA). However, the LGA is a large area, and hotspot maps at this level may be 151 too large to be feasible. Specifically, cholera risk does not appear to be uniform within an LGA; rather, higher 152 risk areas likely exist in smaller pockets within the LGAs. We hypothesize that cholera hotspot maps at the 153 ward level would be more helpful in targeting interventions including OCV, but for which we found no studies 154 delineating cholera hotspots at the ward level (hotspot wards). In this study, hotspot wards are termed micro- 155 hotspots. The ward is the fourth and smallest administrative level in Nigeria; thus, studying cholera at this 156 level will provide an improved understanding of micro-hotspots, the specific wards prone to initiating, 157 sustaining, and spreading yearly epidemics. This is urgently needed to provide decision-making vis-a-vis critical 158 interventions including OCV for cholera elimination by 2030. 159 160 To illustrate the difference between an intervention at the level of an LGA compared to a ward, the average 161 population of an LGA of Kano State is about 300,000 [17] , and there are 10 to 15 wards per LGA. 162 Administratively, Kano State is divided into 44 LGAs; In turn, the 44 LGAs are further divided into 484 wards 163 (Fig 1) . Of these, 88 are urban wards. Though we identified hotspots LGAs in Kano State [8] , the specific ward 164 where the interventions need to focus was not identified; in fact, very little is known about cholera 165 transmission at the ward scale. Studies by George et al. [18] and by Ali et al. [19, 20] illustrate the focal nature 166 of cholera transmission and show that cholera risk is much higher close to index cases than it is at a distance. 167 Attempts to define hotspots at the LGA level [8] miss the true facilitators of transmission that can be identified 168 by studies of micro-hotspots. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint and Ngwa et al. [7, 8] . The national plan for Uganda included hotspot mapping to identify the districts to be 173 targeted for OCV and water sanitation hygiene (WASH) [22] . However, within the districts, local officials 174 provided additional local, but still subjective, information as to the sub-districts, which were the "true 175 hotspots," and these sub-districts were then targeted. This study replaces the subjective impressions of local 176 officials with objective information related to rates and risks of cholera at ward level. analyzing 3) temporal, and 4) geographical patterns of spread incorporating attack/incidence rates, and 5) 194 describing cholera micro-hotspots. The study aims to compare all five points between the urban and rural 195 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Board (IRB) determined that the study is not human subjects' research and was exempt from IRB review. 207 Based on the anonymity of data, we did not obtain informed consent. of Dorayi (Gwale LGA) (Fig 1B) . The population is highest in the Center urban wards (Fig 1 C) and progressively 219 decreases outwards towards the rural wards. Among the states of Nigeria, we chose Kano State because of the 220 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Health. The state analyzes the data and sends these to the NCDC, the national public health institute that 235 coordinates surveillance activities, and from there to other government entities and international partners. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint Nigeria, and classification of wards as urban or rural was read directly from the ward Shapefile. Furthermore, 245 (inland water bodies) data on river lines and lakes, i.e., the contextual factors of cholera transmission, were 246 obtained online (http://www.diva-gis.org/Data). All spatial data in the geographic coordinate system were 247 projected into the Universal Transverse Mercator, Zone 32N, coordinate system. showing cholera in wards not known to report the disease, we proceeded with using the crude ARs. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint To describe the dynamics of yearly cholera outbreaks for 2010-2019, we stratified the cases and deaths by 293 location (urban/rural), gender (female/male), and age groups (<5, 5-14, >15 years). We then calculated the 294 overall attack (suspected and confirmed cases/population) and death (deaths/population) rates, CFRs 295 (deaths/suspected and confirmed cases), proportion of suspected cases and deaths, proportion of cases 296 classified as severe (hospitalized cases), proportion of female, proportion of stool samples taken and positive 297 rapid diagnostic tests, and proportion of rapid care-seeking. Severity was defined as hospitalization due to or 298 death from cholera. A table and histogram were produced to visualize the distribution of cholera suspected 299 cases and deaths as a function of location, and age group. We used the Chi-square test to analyze associations 300 between location, gender, and age group, and outcome (survived/died) of cholera. P-values of the test 301 statistic were reported with 5% (i.e., α = 0.05) significance level. . This is a common method, which ascertains whether the number of 314 reported cases of the disease in a ward exceeds the expected number compared to rates outside the ward. 315 We also used a method recently developed by the GTFCC but applied it to the geographic area of the ward 316 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint rather than the LGA (district) [29] . We chose SaTScan primarily because it was commonly used in detecting 317 clusters of disease incidence including cholera hotspots detection [7, 22, 43, 44] . Secondly, SaTScan 318 incorporates spatial and Spatio-temporal scan statistics with linkage to GIS for results visualization. We also 319 used the GTFCC tool [29] to assess how it compares with SaTScan in hotspot identification and classification. 320 The GTFCC tool, based on Microsoft Excel, can readily be used with ease in resource-poor settings at the 321 periphery level (ward and health facility level) by health personnel without statistical/GIS skill competencies 322 needed to use SaTScan and GIS. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint incidence by dividing the number of reported cholera cases of each ward by its population size for each year. 341 Next, we calculated the mean of the annual incidences for the ten-year study period for each ward. Types of micro-hotspot (T) were assigned to each ward based on the two indicators delineating priority levels 346 into high (T1), medium (T2), medium (T3), and low (T4). A 55 th and 71.6 th percentile values were defined as the 347 cut-off points for high mean incidence rate and persistence; however, the cut-off levels could be adjusted as 348 desired. T1 micro-hotspots corresponded to high priority areas and were wards with high mean annual 349 incidence and high persistence of cholera during the surveillance period. T2 micro-hotspots corresponded to 350 medium priority areas, which were wards with a high mean annual incidence rate and low persistence of 351 cholera. T3 micro-hotspots corresponded also to medium priority areas characterized by a low mean annual 352 incidence rate, but a high persistence of cholera. T4 micro-hotspots corresponded to low priority wards, which 353 had a low mean annual incidence rate and low persistence of cholera. The cut-off values for incidence and Descriptive person (patient) characteristics variation. In our study period, 18483 suspected and confirmed 364 cholera cases and 617 deaths (CFR = 3.3%) were reported in urban and rural settings from Kano State. Of 365 these, 67.7% (12580/18483) of the cases for the study period were from rural wards; however, overall AR was 366 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint slightly higher in the urban wards. Still, 72% (443/617) of the deaths were from rural wards, which saw a 367 slightly higher death rate (Table 1 ). CFR decreased with increasing age ranging from 7.4% among children 368 under five years old to 2.6% among the working-age group (≥15 years) in the urban area. Cholera severity 369 (hospitalization or death) was higher in the urban (77.4%) compared with the rural (53.4%) setting and the . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. From 2010-2019, the ages of the cases ranged from 1 month to 100 years. In this period overall, we found 383 statistically significant evidence that cholera outcome (survived/died) were gender (p-value = 0.0006), and age 384 group (<1-4, 5-14, and >15 years) (p-value = 0.0005) dependent. Although we found that location was not 385 associated with overall outcome (p-value = 0.2465), yet, amongst women, cholera outcome was statistically 386 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint and significantly associated with location (p-value = 0.003). In contrast, amongst men, cholera outcome was 387 not linked with location (p-value = 0.4527). Excluding 2011 and 2013, there were more cases and deaths 388 among women in the rural setting than those in the urban setting (Fig 2) . Apart from 2010 and 2011 in the 389 urban setting, the annual age distribution of suspected and confirmed cholera cases and deaths showed a 390 right skew shape (Poisson distribution); i.e., there were more cases of cholera among people in younger age 391 brackets (<5 and 5-10 years old) compared to any other age bracket (Fig 2) . Descriptive Spatio-temporal variation of cholera. Weekly reported cholera cases in our study period started in the 393 rural setting during week 31 of 2010 during the rainy season (Fig 3, S1 Fig A) . Weekly CFRs ranged from 80% (late September) with peaks in August (Fig 3) . The spatial distribution of wards with high cholera crude AR per 403 100,000 population revealed considerable spatial heterogeneity between 2010 and 2019 with cholera 404 occurring in the central, north, south, east, and west portions (Fig 4 CR) . We found marginal differences 405 between spatial distribution in the cholera crude ARs and EBS rates (Fig 4 CR, Fig 4 EBS) . The widest spatial 406 spread occurred during 2014 and 2018, but in no year were all wards affected even after smoothing (Fig 4) . In Fig 6 and is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint cluster (Fig 5) , 168 yielded relative risk greater than 1 (greater observed than expected cases) (Fig 6) . A 430 breakdown of the 168 micro-hotspots of RR > 1 is in order (Table 3, Fig 6) . The (1) . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint *Note: Table is arranged in descending order of relative risk (RR), which represents the cholera incidence rate among the population in the scan 440 window (exposed group) divided by the cholera incidence rate among the population out of the scan window (unexposed group). The ward 441 population reflects the year 2019 estimates. In total, SatScan identified 168 wards with RR > 1 with a population of ~ 4.9 million; and thus, the so-442 called micro-hotspots wards. Still, 79 micro-hotspots had a RR ≥ 2 with a population of ~ 2.2 million compared to the State as a whole. 71.6th percentiles, the cut-off for the mean annual incidence rate and persistence were 10/100,000 persons 450 and 2.50%, respectively. Application of these criteria yielded 115 wards as T1 micro-hotspot, 105 as T2 micro-451 hotspot, 23 as T3 micro-hotspot, and 241 as T4 micro-hotspot (Table 4, S1 Table) . In Fig 7, we present the 452 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint cholera micro-hotspot classification chart ( Fig 7A) and map (Fig 7B) while Table 4 and S1 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. In keeping with our objective of comparing SaTScan and GTFCC methods in analyzing micro-hotspot patterns, 464 both showed points of similarities and dissimilarities. Remarkably, for 2010-2019, both methods agreed on 465 eight out of the top ten ranked micro-hotspots. Yet, SaTScan identified Kunchi and Tumbau within its top ten 466 ranked micro-hotspots while GTFCC classified Kabo and Kumbotso within its top ten ranked micro-hotspots 467 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint (Tables 3 and 4) . Similarly, considering the T1 and some T2 wards, of the 168 micro-hotspots detected using 468 SaTScan, 57.74% (97/168) and 42.26% (71/168) were classified as T1 and T2 by GTFCC method, respectively. 469 Conversely, the GTFCC tool classified 20 wards from Rogo Ruma to Gwangwan as T1 (Table 4) when SaTScan 470 did not detect these wards as micro-hotspots; i.e., the 20 wards had a RR<1. Finally, 52.98% (89/168) of micro-471 hotspots detected by SaTScan had a RR < 2 (Table T3) ; nonetheless, GTFCC classified most of these wards as T1 472 (Table 4 ). For instance, according to SaTScan, the Kaura Goje micro-hotspot has a RR = 1.02 (Table 3) , but was 473 a T1 micro-hotspot according to GTFCC with a very high cholera persistence of 10.26% (Table 4) . is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint younger (Fig 2) . The CFR was higher in rural settings (3.4%) but decreased with increasing age in both settings. 492 We observed a 77.4% case severity in the urban compared to 53.4% in the rural setting. 2) could be attributed to rural practices regarding food (eating habits), hygiene, and gatherings such as 506 funerals, weddings, markets, and the use of surface water. Further, children might be finishing adults' plates, 507 women preparing food using surface water, and attending and eating at funerals/weddings without proper 508 hand hygiene, all factors that have been associated with cholera outbreaks in Cameroon [55] For 2010 to 2019, we confirmed that in Kano State, though the disease occurs in the dry and rainy seasons 512 alike, higher rates of cholera occurred during the rainy season (June to September) with a peak in August (Fig. 513 3) [9-12]. Most importantly, cases in the rural and urban settings peak asynchronously, i.e., when cases in the 514 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint rural setting are at a peak, those in the urban location are at a trough (Fig 3) . Control efforts would benefit 515 from studies of the seasonality of cholera in Kano State that correlate rainfall data with cases of cholera. 516 517 We observed that CFRs are high (3.5% rural vs 3.0% urban), far exceeds the <1% recommended by the WHO 518 and decreasing with increasing age (Table 1) , as has been documented elsewhere [57] . High CFRs call for 519 better hospital management of cases in general, but with particular attention on children <5 years old, and 520 improvements in surveillance to trigger a timely response. The decreasing rates of disease with age are 521 thought to be related to acquired immunity, and this is supported by the increasingly elevated vibriocidal The present study also shows that between 2010 and 2019, sample collection and testing to ascertain V. 534 cholerae in stool was only 1.3% overall. Likely, the 2.2% to 1.0% (reflected in the RDT positivity rates 61.7% to 535 63.8%) disparity in urban and rural stool sample testing indicates the difficulty in reaching rural locations with 536 RDTs. Inadequate testing occurred among <5 years and 5-14 years olds in the rural setting, age groups in 537 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint which suspected cases have a higher chance of not being cholera. Unfortunately, data on culture confirmation 538 was not available for analysis. This insufficient testing calls for urgent research to understand the limiting 539 factors to laboratory surveillance including, but not limited to, personnel and availability of RDT kits in the 540 urban and rural settings alike. 541 542 There was a clear tendency for wards reporting cholera to be clustered around inland water bodies of the 543 northern, central, and southeastern portions of the state (Fig 5) . We detected 168 wards that were at high risk 544 of cholera and were considered to be micro-hotspots, compared to other wards, and both methods used were 545 generally in agreement. 546 However, SaTScan did not detect the bottom 20 T1 micro-hotspots of Table 4 (from Rogo Ruma to Gwangwan) 547 as micro-hotspots. SaTScan is dependent primarily on incidence whereas the GTFCC method includes both 548 incidence and persistence. Thus, areas with higher incidence may be better identified when using the GTFCC 549 method. Yet, as both methods generally agreed (Figs 6 and 7) , our finding corroborates our hypothesis that 550 although an LGA may be detected as a hotspot LGA not all wards within it are micro-hotspots [8] . For example, 551 in our earlier hotspot detection at the LGA level, Rogo LGA was not classified as hotspot LGA [8], but at the 552 fine-scale of ward, both methods detected 3-4 micro-hotspots in Rogo (Figs 6 and 7) . Also, Kunchi LGA was 553 classified as hotspot LGA, but only one ward in Kunchi LGA is a micro-hotspot, and so forth. Therefore, we 554 conclude that hotspot maps at the ward level, not hotspot maps at the LGA level, are best suited to target 555 cholera interventions including vaccines. 556 557 As our main objective was to provide NSPACC with data to guide interventions, the micro-hotpots in Table 3 558 would benefit most from priority vaccination by the National Primary Health Care Development Agency, Kano 559 State Ministry of Health, and NCDC while longer-term WASH infrastructure is being put in place in line with the 560 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) This study has important limitations as well as strengths. Testing of suspected cases was very low. As such, we 571 cannot roll out misclassification of cases based on clinical criteria leading to underestimation or 572 overestimation of the true rates of cholera. Although unlikely, retrospective determination of ward of cases 573 could have led to assigning cases to the wrong ward. Nonetheless, the use of the health facility-based line-list 574 data spanning ten years afforded the prospect to describe demographic, testing, and behavioral 575 characteristics of cases as well as patterns of spatial distribution, Spatio-temporal clusters, and micro-hotspots 576 needed to inform intervention including vaccine; and thus, constitute major strengths of the study. Other 577 strengths include illuminating urban/rural disparity in cholera cases and deaths besides comparing two 578 methods to micro-hotspots detection, which generally agreed. Although SaTScan is widely used, its use 579 demands a strong statistical knowledge compared to the GTFCC method, which requires only entering data 580 into Microsoft Excel. As such, the later method can readily be applied in resource-poor settings without 581 statisticians for hotspots identification. Finally, policymakers have been served with analysis stratified by 582 gender, age, and location leading to uncovering the rural-urban disparity in cholera transmission. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The insert shows a better view of the urban area clusters. 819 820 821 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. we identified 168 (37.71%) with RR > 1; and thus, the so-called micro-hotspots. The ward specific RR was calculated 826 within each cluster using space-time scan statistics incorporating reported cholera cases for each week and the ward 827 population for each year. Here cholera micro-hotspot means a geographically limited location where environmental, 828 cultural and/or socioeconomic conditions facilitate the transmission of the disease and where cholera persists or 829 reappears regularly. The insert shows a better view of the micro-hotspots at the Centre location. Note that the labels in 830 the map represents Local Government Areas (LGAs) in which the wards are found. 831 832 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint population. T represents types of micro-hotspots. T1 (high priority) reflects high persistence and mean annual incidence; 836 T2 (medium priority) shows high persistence and low mean annual incidence; T3 (medium priority) indicates low 837 persistence and high mean annual incidence; and T4 (low priority) depicts low persistence and mean annual incidence, 838 respectively. The dots denote wards in the four types of micro-hotspot quadrants. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted August 23, 2021. ; https://doi.org/10.1101/2021.08.20.21262313 doi: medRxiv preprint Updated global burden of cholera in endemic countries ):e0003832. 625 2. World Health Organization. Cholera: Number of reported cases Cholera Plateform. Cholera Outbreaks in Central and West Africa : 2019 Regional Update -Week 13: 628 Cholera Platform Cholera in Cameroon, 2000-635 2012: Spatial and Temporal Analysis at the Operational (Health District) and Sub Climate Levels The cholera risk assessment in Nigeria: A historical review, mapping of hotspots and evaluation of contextual factors. PLoS 639 neglected tropical diseases Climate and socioeconomic influences on interannual variability of 641 cholera in Nigeria Potential future risk of cholera due to climate change in northern Nigeria. African 643 Research Review Perception and trends of associated health risks with seasonal climate 645 variation in Oke-Ogun region Unique Clones 647 of Vibrio cholerae O1 El Tor with Haitian Type ctxB Allele Implicated in the Recent Cholera Epidemics from 648 Cholera epidemiology in Nigeria: 650 an overview Nigeria Centre for Disease Control (NCDC). National Strategic Plan of Action on Cholera Control 657 (NSPACC) Randomized 661 controlled trial of hospital-based hygiene and water treatment intervention (CHoBI7) to reduce cholera Emerging infectious diseases Potential for controlling cholera using a 664 ring vaccination strategy: re-analysis of data from a cluster-randomized clinical trial Cholera cases cluster in time and space in Matlab, 667 Bangladesh: implications for targeted preventive interventions Oral cholera vaccine in 669 cholera prevention and control Identifying cholera" hotspots" in 671 Uganda: An analysis of cholera surveillance data from Interactive spatial data analysis: Longman Scientific & Technical Essex The Knox method and other tests for space-time interaction The detection of disease clustering and a generalized regression approach. Cancer research Density estimation for statistics and data analysis Visualising space and time in crime patterns: A comparison of 680 methods. Computers, environment and urban systems ):e59. 683 29. GTFCC. Guidance and tool for countries to identify priority areas for intervention 685 countries-to-identify-priority-areas-for-intervention/. 686 30. Nigeria: Administrative Devisions. States and Local Government Areas International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity Nigeria Cholera Surveillance Structure Abuja: Nigeria Centre 691 for Disease Control (NCDC) USA: World Health Organization and Centers for Disease Control and Prevention NSIPSS] NSfIaPSS. Nigeria Strategy for Immunisation and PHC System Strengthening Strengthening_3rd_Version-Final.pdf Current trends of immunization in Nigeria: prospect 702 and challenges R: A language and environment for statistical computing. Vienna, Austria R 704 Foundation for Statistical Computing Tango T. A class of tests for detecting 'general'and 'focused'clustering of rare diseases Adjusting Moran's I for population density A new proposal to adjust Moran's I for population density. Statistics in 711 medicine Centers for Disease Control and Prevention. Laboratory Methods for the Diagnosis of Vibrio cholerae: 715 Isolation of Vibrio Cholerae from Fecal Specimens Vibrio-cholerae-chapter Dynamics of cholera 719 outbreaks in Great Lakes region of Africa Spatial dependency of V. cholera prevalence on open space refuse dumps in Global Task Force on Cholera Control. Guidance and tool for countries to identify priority areas for 723 intervention: Global Task Force on Cholera Control Studies on interventions to prevent eltor 727 cholera transmission in urban slums Coverage and cost of a large oral 729 cholera vaccination program in a high-risk cholera endemic urban population in Dhaka International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity Clinical, epidemiological, and spatial 732 characteristics of Vibrio parahaemolyticus diarrhea and cholera in the urban slums of Kolkata, India. BMC 733 public health Studies on interventions to prevent eltor cholera transmission in Cholera Outbreak Guidelines -Preparedness, Prevention and Control: OXFAM Medecins Sans Frontieres Cholera outbreak guidelines: Preparedness, prevention and control: UNICEF Epidemic cholera in Guinea-Bissau: the 748 challenge of preventing deaths in rural West Africa Community mortality from 751 cholera: urban and rural districts in Zimbabwe. The American journal of tropical medicine and hygiene International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity Cultural influences behind cholera 754 transmission in the Far North Region, Republic of Cameroon: a field experience and implications for 755 operational level planning of interventions Antibody responses after 763 immunization with killed oral cholera vaccines during the 1985 vaccine field trial in Bangladesh A serological survey for cholera antibodies in rural east Pakistan The distribution of antibody in the control population of a cholera-vaccine field-trial area and the relation of 767 antibody titre to the pattern of endemic cholera Urban-rural 769 differences in health-care-seeking pattern of residents of Abia state, Nigeria, and the implication in the control 770 of NCDS. Health services insights Apart from the years 2010 and 787 2011 in urban setting, the distribution of cholera cases by age has a right skew (Poisson) shape. This implies that, irrespective of sex 788 and location, the cases of cholera were higher among people of younger ages (<10 years) than older ones. Excepting the years 2011 789 and 2013, amongst women, cholera survival significantly depended on setting (p-value = 0.002); Women in rural areas were more 790 likely to die from cholera than women in urban areas International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity