Environmental Determinants of Infectious Disease Roads and diarrheal

Environmental Determinants of Infectious Disease Roads and diarrheal disease Joseph Eisenberg, Ph. D University of Michigan April 2016

How New Roads Affect the Transmission of Enteric Pathogens v World bank and health impact assessment § Roads provide access to health care § Report will emphasize hazards associated with the building of the road Downstream effects are ignored

How New Roads Affect the Transmission of Enteric Pathogens v Both ecological and social drivers of transmission are influenced by roads. § Past studies focused on § § Roads and STDs, driven largely by social processes Roads and vectorborne diseases, driven largely by ecological processes § Enteric pathogens affected by both types of drivers § § Waterborne, foodborne pathogens are mediated by environment Also transmitted by person-person contact

Roads: What Do They Bring? Primary Roads Lead to secondary road construction

Roads Facilitate Movement

Roads Facilitate Changes in Population Structure

Roads Cause Environmental Change Ecologic change Social change

§ Study Site § Coastal rainforest § Afro-Ecuadorian and Chachi Indians § 300 -year history of living independently § Riverine transport system adequate for gold mining, rubber tapping, and tagua § Road system needed for logging and oil palm plantation § 1996 begin road construction linking coast and Andes § Connects villages along three rivers

Interdisciplinary approach v Study design is interdisciplinary in scope § Public Health § § Epidemiology (Village-level cohort , within village case-control) Molecular Biology (Strain analysis of marker pathogens) § Environmental assessment § § § Water quality Sanitary assessments Food distribution patterns § Systems § § § Social Network Theory – mapping of contacts Spatial analysis – GPS (villages in relation to roads/rivers, climate) Mathematical modeling – Integration using GIS and models of disease transmission

Study Components v Study components § Mapping / Climate § Census § Active surveillance Weekly visits to each house by village health promoter § Case/control studies Rolling village visits, once in the dry season and once in the wet season § Social network surveys § Ethnography v Health outcomes § Diarrheal disease, antibiotic resistance, dengue § Nutrition (anthropometry, hemoglobin, diet)

Causal links between roads and diarrheal disease Ecologic and social drivers

Social drivers v From remote to more proximate road access § Increased reintroduction of pathogens from outside of regions § People in remote villages have less contact with the outside world § Decreased social cohesion § § People remain in remote villages for longer periods People have more interaction in remote villages

Comparison of Infection Prevalence v Overall infection prevalence (including subclinical cases) § Adjusted for age, village population, sanitation, rain v The remoteness metric compares farthest village with closest Eisenberg et al 2006

Comparison of Infection Prevalence v Differences in spatial trends among marker pathogens § Pathogens with higher Ro less impacted by remoteness § Ro function of shedding rates, environmental persistence, and infectious inoculum

Level and Type of Antibiotic Resistance v Overall 44% of isolates tested were resistant to one or more antibiotics Ab tested for Ampicillin Tetracycline Sulfamethoxazoletrimethoprim Chloramphenicol, Cefotaxime, Gentamicin, Ciprofloxin

Regional Patterns of Antibiotic Resistance v Communities aggregated into far, medium and close with respect to road access § Based on seven 15 -day case control studies (2003 -08) § Prevalence estimates are a weighted sum of cases (diarrhea) and controls Multivariate model controlling for age, population size, and Ab use Adjusting for correlation within the village Eisenberg et al 2011

Possible Explanations of Spatial Patterns v Hypotheses § Antibiotic use § Spread of antibiotics, antibiotic resistant bacteria, gene-gene transfer § Reintroduction of antibiotic resistant bacteria

Antibiotic use v Self-report antibiotic use § During the past week § N = 2532 § Aggregated across 21 villages No significant trend by remoteness Antibiotic Amoxicillin Ampicillin Benzipenicillin Sulfamethoxazoletrimethroprim Gentamycin Ciprofloxacin Garamycin Chloramphenicol Tetracycline Cefotaxime Other* Unknown Total % 20 1 4 2 (S) 6 (T) 7 8 1 0 28 9 267

Village Level Transmission Analysis v Transmission rates (b and h) § Use E. coli prevalence in road vs. remote villages (PNAS 2006) § Assume SIS model of transmission v Antibiotic use rates (r) l (ingestion rate) b. Z Exposed h. Y § Survey data (20% random sample of households) r Amplified Colonized - - Transmission r: Antibiotic use rate Eisenberg et al 2011

Explaining Patterns of E. coli Resistance in Communities v Comparing road and remote villages § Antibiotic use determines the importance of introduction of resistance vs. transmission Low antibiotic use High antibiotic use

Ecological Perspective: Do risks come from neighboring villages? Markov chain model: state of village k (high, medium, low diarrheal rates) at t depends on state of 21 villages at t-1. § 4 yrs. active surveillance data across 21 villages § Villages weighted using a gravity model (distance and size) Goldstick et al 2014

Ecological Perspective: Regional Transmission v Risk factors often characterized as static But may vary by environmental and biological contexts v Regional spread: Environmental transport vs. human movement

Ecological Perspective: Climate Outcome: Diarrhea § Weekly visits to households over 4 years

Ecological Perspective: Climate Exposure § Extreme rainfall: 90 th percentile over 4 year period Contextual variable § 8 -week total rainfall

Ecological Perspective: Climate Total 8 -week rainfall IRR (95% CI) Low (78 - 425 mm) 1. 39 (1. 03, 1. 87) Medium (426 - 604 mm) 0. 70 (0. 44, 1. 11) High (605 - 1356 mm) 0. 74 Risk (diarrhea) associated with a 2 week lagged extreme rain event (0. 59, 0. 92) Water treatment can counteract risk associated with extreme rain events Low total rainfall Water treatment is required to achieve protective effect associated with extreme rain events High total rainfall

Ecological Perspective: Social Networks v Background § Social networks typically seen as conduits of transmission § But social relationships can also be protective v Disease spreads more slowly to and in rural villages that are more remote due to § Reduced contact § Greater density of social ties between individuals in remote communities facilitates spread of individual and collective protective practices Zelner et al 2012

Ecological Perspective: Social Networks v Cross sectional survey (2007): N > 4000; 24 villages § Self-report diarrheal disease § Sociality networks § Who do you talk to for important matters? § Contact networks § Who have you spent time with during the last week (outside your household)? – For anyone or for infectious individuals

Ecological Perspective: Social Networks v Heterogeneous social landscape across villages § Networks for similar size villages Remote village Close village Isolates not shown

Ecological Perspective: Social Networks v Risks and protective effects are mediated through a number of social processes OR = 0. 49 (0. 29, 0. 84) OR = 1. 12 + Within-village Infectious contacts OR = 0. 89 (0. 81, 0. 98) (1. 00 , 1. 2 5)

An Ecological Perspective Ø The presence of road causes environmental changes (social and ecological) § These changes occur differentially across the landscape of villages § § § Affects social structure Spread of microorganisms differentially through water sanitation and hygiene pathways Affects movement and migration patterns at multiple scales Affects climate and hydrological processes Regional patterns of environmental change will vary over time.

Acknowledgments United States Ecuador v v v v William Cevallos (Project director) Gabriel Trueba (PI: Microbiologist) Diana Lopez (Microbiologist) Eugenia Meja (Microbiologist) Maria Ines Baquero (Microbiologist) Andres Acevedo (Field anthropologist) Vilma Requene (Field assistant) Mariuxi Ayovi (Field assistant) Deni Tenorio (Field assistant) Mauricio Ayovi (Field assistant) Maritza Renteria (Field assistant) Jose Ortiz (Transportation coordinator) Emel Bustamante (Data entry) • • James Trostle Betsy Foxman, Carl Marrs, Lixin Zhang, Karen Levy James Fuller Ian Spicknell, Jason Goldstick, Jon Zelner, Robert Wood Health Promoters Deni Orobio, Pastor Mercado, Cecilia Mercado, Carmen Nazareno, Ludis Castillo, Mirtha Campaz, Estela Arroyo, Ramona Sabando, Maria Ayovi, Blanca Vega, Jorge Peralta, Santos Mina, Amelia Preciado, Marco B. , Ereccni Cuero, Julio Valdez, Lucrecio Palacio, Heroina Arboleda, Juliana Mina, Adalin Valencia, Mariuxci C. , Dominga A. , Maria Arroyo, Gonzolo M. , Gabriel Ayovi, Maria Corozo