Chikungunya, Dengue, and Zika in Cali, Colombia: epidemiological and
geospatial analyses
Amy R Krystosik, BS, MPH, Ph.D. (candidate)
Department of: Biostatistics, Environmental Health Sciences, and EpidemiologyCollege of Public Health, Kent State University
March, 2016
ChikungunyaCountries and territories where cases have been reported* as of October 20, 2015 (CDC, 2016)
*Does not include countries or territories where only imported cases have been documented. This map is updated weekly if there are new countries or territories that report local chikungunya virus transmission.
Colombia
DengueCartogram of the annual number of infections for all ages as a proportion of national or subnational (China) geographical area (Bhatt et al., 2013)
Zika Countries and territories showing historical distribution of Zika virus (autochthonous transmission), 1947 – 2016 as of February 7, 2016; Colombia: >31,000 cases reported with no increase in incidence of microcephaly.
Dengue in ColombiaPattern of endemicity for dengue 2008-2013
Source: Republic of Colombia, Ministry of Health and Social Protection
Cali
Chikungunya in CaliConfirmed cases by municipality of transmission October 17, 2015.
The accumulated confirmed cases nationally up to Up to the cut-off date is 439,000 cases accumulated in 712 municipalities.
Source: Instituto Nacional de Salud -INS
Chikungunya in CaliConfirmed cases by municipality of transmission for the epidemiological weeks 1-6, 2016.
The accumulated confirmed cases nationally up to the cut-off date is 6,356 in 371 municipalities.
Source: Instituto Nacional de Salud -INS
Zika in CaliCases by territory up to epidemiological week 8, 2015-2016
Source: Sivigila
Chikungunya Dengue ZikaVector Aedes Aedes Aedes
Virus Alphavirus Flavivirus Flavivirus
Fever >80% >80% 20-80%
Relative bradycardia (pulse-temperature deficits) <20% 20-80% <20%
Neurological- headache, meningitis, etc. <20% >80% 20-80%
Myalgia (muscle pain); or muscular mass or swelling 20-80% >80% 20-80%
Pulmonary, thoracic, chest wall or cardiac <20% <20% 20-80%
Arthritis 20-80% <20% 20-80%
Hepatic dysfunction <20% 20-80% <20%
Neutropenia (An abnormally low count of a type of white blood cell)
20-80% 20-80% <20%
Macules and/or papules (rash) 20-80% 20-80% >80%
Musculoskeletal- muscle, bone, and joint >80% >80% 20-80%
Thrombocytopenia (deficiency of platelets in the blood causing bleeding into the tissues, bruising, and slow blood clotting after injury)
<20% 20-80% <20%
Disease
Gideon.
Abrupt fever, leukopenia, myalagia and prominent bilateral joint pain; maculopapular rash appears on 2nd to 5th days in greater than 50% of cases; fever resolves within 7 days, but the joint pain may persist for months.
Headache, myalgia, arthralgia, relative bradycardia, leukopenia and macular rash
A mild dengue-like illness with conjunctivitis and a pruritic maculopapular rash that starts on the face and spreads to the rest of the body; joint pain is common; myalgia; retro orbital pain and leg edema may occur; may be associated with congenital neurological defect.
Currently
Situation
• Hypoendemic dengue
• Recent outbreak of chikungunya
• Increasing concern over zika
• Current vector control methods implemented by mayor’s office insufficient
Gaps
• Maps lack local detail and community knowledge
• Geographical factor has not been considered in epidemiological risk analysis
• New local mapping methods are ready to be field tested
HypothesisDengue, chikungunya, and zika cases and risk factors have a spatial
component which can predict future disease patterns, and thus can be used to direct control efforts
Specific Goals
1. Describe hotspots and environmental risks
2. Describe geographical and social risk factors for dengue, chikungunya, and zika incidence
3. Describe the geographical and social risk factors for predicting severity of dengue
Design
• Mixed methods retrospective and prospective
• Incident case data collected from secretary of health
• Spatial video geonarratives: cross-sectional sample in hotspots
Data
• SIVIGILA: The National System of Vigilance in Public Health
• Available ~ 1 year after events occurred
• Cleaned and distributed by The Ministry of Health and Social Protection by department
• Publically available data
• October 2014 - October 2015
Parameter ValueChikungunya 44,877
Dengue 13,443Neighborhoods in Cali
303
Independent Variables
Age, sex,ethnicity, occupation,geographic location of cases
Factor Chikungunya Dengue W/O
warning signs
Dengue W/ warning
signs
Severe
dengue
Dengue death
N 2,255 10,630 2,716 84 13
Age 43.25 (24.2) 31.7 (18.8) 27.7 (21.6) 31.4 (24.5) 48.7 (30.8)
Sex (F) 1,473 (65.3%)5,085 (47.8%) 1,378 (50.7%) 37 (44.0%) 6 (46.2%)
Ethnic Group 782 (34.7%) 5,545 (52.2%) 1,338 (49.3%) 47 (56.0%) 7 (53.8%)
Indigenous8 (0.4%) 30 (0.3%) 15 (0.6%) - -
rom, gitano9 (0.4%) 18 (0.2%) 7 (0.3%) - -
raizal5 (0.2%) 11 (0.1%) 6 (0.2%) - -
negro mulato
afrocolombian
61 (2.7%) 235 (2.2%) 120 (4.4%) 3 (3.6%) -
other2,172 (96.3%)10,336 (97.2%) 2,568 (94.6%) 81 (96.4%) 13 (100.0%)
Factor Chikungunya Dengue W/O
warning signs
Dengue W/
warning signs
Severe
dengue
Dengue
death
N 2,255 10,630 2,716 84 13
Disabled 1 (<1%) 5 (0.1%) 5 (0.2%) - -
Displaced 4 (0.2%) 6 (0.1%) 6 (0.3%) 1 (1.3%)
Migrant - 1 (<1%) - - -
Pregnant 274 (19.8%) 101 (2.3%) 34 (3.0%) - -
youth under
government care
- 5 (0.1%) 2 (0.1%) - -
community mother 1 (0.1%) 2 (<1%) - - -
Ex-paramilitary
/Ex-guerrilla
1 (<1%) - - - -
under psych care 1 (<1%) 1 (<1%) 2 (0.1%) - -
other group 1,982 (88.6%) 10,518 (99.3%)2,668 (98.9%) 83 (98.8%) 13 (100.0%)
violence victims 1 (<1%) 4 (<1%) 2 (0.1%) - -
Selection Criteria
Inclusion
• Cases• Confirmed or suspect case of
chikungunya or dengue or zika*
• Community Health Worker• From the hotspot of interest; informed
consent to recorded interview; Spanish or English speaking
• Hotspots• Statistically significant at p ≤ 0.05
Exclusion
• Cases• Incarcerated, unable to be georeferenced
• Community Health Worker• Under 18; Speaks neither Spanish nor
English
• Hotspots• Too dangerous to collect data; no CHW
available
*if data become available
Spatial Video Geonarratives
Mechanics• Contour2+ Cameras with video and GPS
coordinates
• Mounted inside polarized windows of car
• Collect video of ground, buildings, animals, greenery, water, and people
• Collect GPS coordinate trail of video
• Likely to code: road condition (cement/paved in good condition, cement/paved with potholes, unpaved dirt), abandoned tires, planters, trash, puddles, lakes, canals, unkempt parks or green areas, and individuals who appear to be living on the street
• Interviews will be translated, transcribed, and coded as a map layer of incidence.
Previous examples• Described by Curtis et al 2013 and
Curtis et al 2015• collected and mapped fine street-level data for
environmental risks related to cholera transmission including standing water, trash accumulation, presence of dogs, cohort specific population characteristics, and other cultural phenomena in a challenging urban environment in Haiti
• Used previously to capture environmental risks for vector-borne diseases • Lewis, Fotheringham, & Winstanley,
2011; Mills, Curtis, Kennedy, Kennedy, & Edwards, 2010; Schultz, 2015
Spatial Video Geonarratives
Benefits
• Allows us to capture data in hard to reach areas at low cost and create time capsules for time-series data
• Applies to other diseases
• First time field testing this method for Aedes
Conditions
• Collected during drought October 14 –December 1, 2015 and January 25 –February 9, 2016
• Interviewed: vector specialist, TB community health nurse, and local community members
Mapping Cali
• Projection: EPSG Projection 3115 - MAGNA-SIRGAS / Colombia West zone
• Layers include: cases, environmental risks, schools, clinics, bodies of water, and potential vector breeding sites
Mapping Cali
Analysis
• Spatial: ArcGIS 10.3 and spatial analyst extension• Hotspot analysis• Kernel density analysis • Geographically weighted regressions
• Epidemiological: Stata 12; R 3.1.2• Univariate analysis- normality (Shapiro Wilk), mean and standard deviation for
normal and median and interquartile ranges for non-normal • Proportions and strata (categorical variables)(chi-squared tests and Fisher’s exact
test)• Cumulative Incidence and severity by social risk groups (displaced, immigrant,
pregnant, children), sex, age, ethnicity, occupation, and location• Incidence trend line estimated over time• Disease specific mortality rate • Case-fatality rate
Analysis
• Incidence: Geographically weighted Poisson regression
• Severity: Geographically weighted mixed multinomial ordered logit regression
• Variables with more than 30 subjects in each cell will be included if found to be significant at p < 0.05 in the bivariate analysis
• The final GWRs will likely include the independent variables of race, age, gender, ethnicity, occupation, social strata, and pregnant women.
• The parametric, semiparametric, and non-parametric models will be compared and the most parsimonious model chosen
Sample size
Epidemiological analysis
• All eligible dengue, chikungunya, and possible zika cases with will be included.
• Due to the descriptive nature of the analysis and the high number of cases a power calculation was not performed.
Geographical analysis• Moran’s I: n ≥ 25
• Nearest Neighbor Index: 30 events per cluster
• Kernel Density Analysis: the minimal sample size is set by the user and can be set during analysis. We will test various bandwidths and select that which minimizes error.
• Hotspots analysis: all safe and statistically significant hotspots as confirmed by nearest neighbor index and Moran’s I.
Anticipated limitations and solutions
• Severity of chikungunya
• 5% sample of chikungunya cases
• Cross-sectional nature of environmental data
• Semi-structured interviews
• CHW knowledge level varies
• Cases not able to be geocoded excluded from analysis
• Reliable data for zika in the region are not yet available
• Safety and social acceptance of field work
Chronogram
*Data collection, cleaning, and analysis can be reopened if zika data becomes available.
*
*
Chronogram
*Data collection, cleaning, and analysis can be reopened if zika data becomes available.
Collaborators
• KSU• Dr Madhav Bhatta• Dr Mark James• Dr Andrew Curtis
• ICESI• Dr Diana Dávalos
• Secretary of Health• Paola Buritica
• SIVIGILA• Jorge Humberto Rojas Palacios
• Valle de Lilli• Robinson Pacheco• Dr Sarita Rodriguez
• Cañaveralejo Hospital• Dr Javier Colorado
Thank you! Questions?