Transcript Social Factors

Integrating social sciences
into the equation?
Mark L. Wilson
Department of Epidemiology
School of Public Health
The University of Michigan
SEARCH: Impacts of Climate Change on Infectious Diseases
Embassy of France, Washington, DC
December 14-15, 2009
Integrating social sciences
into the equation?
YES!
Social factors fundamental to
describing patterns
Social factors play critical role
in causal pathways
BUT HOW DEVELOP ANALYSES?
Starting Points
• Infection ≠ Disease
- Transmission of infectious microbes necessary but not sufficient to appearance
of infectious disease
• Climate is important to microbe transmission
- Evidence suggests components of climate system function in diverse ways
• Climate parameters ≠ Climate Variability ≠ Climate Change
- Impacts on transmission operate differently on different variables, time scales
- All likely to be important in affecting epidemiologic patterns
• ID patterns associated with other environmental factors
- Enormous variety of biological, behavioral, social & economic variables
- Challenge: develop conceptual framework and analytic models that evaluate
multiple variables and complex interactions.
• “All models are wrong, but some are useful” (Box)
Some Goals
• Understand mechanisms
- Expand beyond statistical associations toward analysis of causal pathways
• Identify changing risk patterns (who, where, when)
- Analysis of associations, even lacking knowledge of why, can be useful
• Determine most important drivers in web of causation
- Not all factors contribute equally, some stronger or affect more of the system
• Develop forecasting capacity
- Determine models that permit recognition of future events
• Modeling trade-off: generality, precision, realism (Levins)
- Complexity may preclude detailed analysis to consider multiple variables
operating at different temporal or spatial scales
Social Influences on ID Risk
• Long history of recognition that “social” conditions
affect disease patterns
- Many examples of diseases associated with particular environments
• Shift of focus with development of “germ theory”
- Analyses directed toward microbe as cause of infectious disease
• Contemporary studies may consider “social” factors,
but rarely as part of the causal pathway
- Frequent attempts to adjust for…, control for…, remove the effects of….
• Increasing interest in applying approaches from “Social
Epidemiology” to Infectious Diseases
- Goals include explicit consideration of “upstream” and “contextual” factors
that influence impacts of biophysical environment on transmission and disease
CAN
“SOCIAL EPIDEMIOLOGY”
HELP RESEARCH on
and
UNDERSTANDING of
CLIMATE and INFECTIOUS
DISEASES?
Social Epidemiology
Disparities –
SES / SEP
Psycho-social
factors
Multilevel analyses
“Eco-Social” analyses
Life Course
approaches
Social Epidemiology and IDs
• Why so little application of social epidemiological approaches
and tools to IDs?
• Has such work aimed at understanding and prevention of IDs
lagged behind other studies of "chronic diseases?"
• Are studies that more carefully evaluate complex interactions
among social and environmental factors and risk of IDs likely
to provide useful insights?
• Might insights and methods from social epidemiology enhance
analyses of the role of climate in ID risk?
Systematic Literature Review
Methods:
• Evaluated published research on social determinants of various disease
categories.
• Quantified and assessed temporal trends.
• MEDLINE, PsycINFO and ISI Web of Science, 1966–2005.
Cohen JM, Wilson ML, Aiello AE. Analysis of social epidemiology research on infectious diseases: historical patterns and future opportunities.
J. Epid. Community Health. 2007. 61:1021–1027.
Results:
Average annual number of socially related citations increased by
- 180 for neuropsychiatric conditions
- 82 for chronic conditions
- 45 for sexually transmitted infectious diseases
- 19 for non-sexually transmitted infectious diseases (p = 0.0001).
Of 279 publications found to employ the term ‘‘social epidemiology’’
- 15 (5.4%) investigated infectious outcomes.
Cohen JM, Wilson ML, Aiello AE. Analysis of social epidemiology research on infectious diseases: historical patterns and future opportunities.
J. Epid. Community Health. 2007. 61:1021–1027.
◄Total Number of Citations
◄Total Number Cross-Referenced
with "Social Factors"
Cohen JM, Wilson ML, Aiello AE. Analysis of social epidemiology research on infectious diseases: historical patterns and future opportunities.
J. Epid. Community Health. 2007. 61:1021–1027.
Review Articles of Social Factors and Diseases
Show a Similar Pattern
"Chronic Diseases"
▼
"Infectious Diseases"
▼
Cohen JM, Wilson ML, Aiello AE. Analysis of social epidemiology research on infectious diseases: historical patterns and future opportunities.
J. Epid. Community Health. 2007. 61:1021–1027.
Social Epidemiology Frameworks
Are they relevant to understanding IDs?
Disparities – SES / SEP
Psycho-social factors
Life Course approaches
“Eco-Social” analyses
Multilevel analyses
SEP - Psychosocial
Is this framework
useful to analyses of
infectious diseases?
Kaplan 2004
Life Course
Is this framework
useful to analyses of
infectious diseases?
Ben-Shlomo & Kuh 2002
Life Course
Is this framework useful
to analyses of infectious
diseases? HOW TO
IMPLEMENT?
Ben-Shlomo & Kuh 2002
Life Course – infectious
Lifediseases
Course
Ben-Shlomo & Kuh 2002
Hall et al., 2002
Life Course – infectious disease
Aiello et al. 2007
Multilevel Influences
Is this framework useful
to analyses of infectious
diseases? HOW TO
IMPLEMENT?
Kaplan 2004
Multilevel Influences
Is this framework useful
to analyses of infectious
diseases? HOW TO
IMPLEMENT?
Kaplan 2004
Multilevel Influences
Poundstone et al. 2004
Multilevel Influences
Poundstone et al. 2004
Multilevel Influences
Poundstone et al. 2004
Multilevel Influences
Poundstone et al. 2004
Multilevel Influences
Poundstone et al. 2004
Multilevel and Life Course
Kaplan 2004
Multilevel and Life Course
Kaplan 2004
Multilevel analysis - malaria
Robert et al 2003
Environmental Determinants of
Human Disease
Multilevel Influences
Social and Economic Policies
Institutions (including medical care)
Living Conditions
Social Relationships
Individual Risk Factors
Genetic/Constitutional
Factors
Infectious and
Pathophysiologic
pathways
Individual/Population
Health
Modified from Kaplan, 2002
Environmental Determinants of
Human Disease
Multilevel Influences
Social and Economic Policies
Institutions (including medical care)
Living Conditions
Social Relationships
Individual Risk Factors
Genetic/Constitutional
Factors
Infectious and
Pathophysiologic
pathways
Individual/Population
Health
Modified from Kaplan, 2002
An Example of this Approach
Background:
• American Cutaneous Leishmaniasis (ACL) is a sand fly-borne
(Leishmania spp.) infection, with various reservoirs throughout the world.
• ACL associated with changes in relationships between people and
forests, suggesting forest ecosystems increase transmission risk.
•
Analyzed county-level
incidence rates of ACL in
Costa Rica as a function of
social and environmental
variables relevant to
transmission ecology with
statistical models that
incorporate breakpoints.
Chaves LF, Cohen JM, Pascual M, Wilson ML. Social exclusion modifies climate and deforestation impacts on a vector-borne disease.
PLoS Neglected Trop Dis. 2008. 2(2): e176. doi:10.1371/journal.pntd.0000176.
Methods:
• Monthly number of ACL by county (N=81) for all Costa Rica (1996 – 2000)
• County-level data on % people living ≤5 km from forest; percent forest cover
• Social marginality index (access to resources ensuring quality of life)
• Social exclusion variables (income, literacy, education, health insurance, etc.)
• Monthly rainfall (14 weather stations across the country) and ENSO signal
• Ordinary kriging to interpolate mean, min, max rainfall; elevation data layer
• Sand fly species and locations from systematic reviews
• Kuldorff’s Scan Statistic to find spatio-temporal clusters
• Local Indicators of Spatial Autocorrelation (LISA)
• Negative Binomial Generalized Linear Models (NB-GLM) with breakpoints
• Linear Models and Analysis of Covariance (ANCOVA)
Chaves LF, Cohen JM, Pascual M, Wilson ML. Social exclusion modifies climate and deforestation impacts on a vector-borne disease.
PLoS Neglected Trop Dis. 2008. 2(2): e176. doi:10.1371/journal.pntd.0000176.
Results:
•
Disease incidence and social marginalization highest in same counties
•
Not found for other ecological variables (precip, elevation, forest, dist. to
forest edge)
•
Pattern confirmed by spatial statistical analyses (overlapping
geographical clusters)
•
No clear landscape effects on sand fly species (PCA )
•
Marginalization, % near forest, rainfall, and elevation identified in nonlinear relationships with ACL incidence (78% of variance, GAM model
selection, backward elimination).
•
Non-linear pattern found (72% of variability - Negative binomial GLMs
incorporating breakpoints)
Chaves LF, Cohen JM, Pascual M, Wilson ML. Social exclusion modifies climate and deforestation impacts on a vector-borne disease.
PLoS Neglected Trop Dis. 2008. 2(2): e176. doi:10.1371/journal.pntd.0000176.
Chaves LF, Cohen JM, Pascual M, Wilson ML. Social exclusion modifies climate and deforestation impacts on a vector-borne disease.
PLoS Neglected Trop Dis. 2008. 2(2): e176. doi:10.1371/journal.pntd.0000176.
Conclusions:
• Social marginality significantly reduced the "effect" of
living close to a forest on ACL risk
• Association was non-linear and better explained by
identifying a "breakpoint."
• Forest cover associated with modulation of ENSO effects
adding complexity to environmental forces and disease
patterns.
• Social factors (previously not evaluated) and
environmental/climatic factors, appear critical to
determining disease risk.
Chaves LF, Cohen JM, Pascual M, Wilson ML. Social exclusion modifies climate and deforestation impacts on a vector-borne disease.
PLoS Neglected Trop Dis. 2008. 2(2): e176. doi:10.1371/journal.pntd.0000176.
Another Example: Malaria in Vanuatu
• Pacific Archipelago
– Ethnically diverse
– Border for Anopheles spp
distribution
• Anopheles farauti s.s.
• P. falciparum and P. vivax
• Different degrees of endemicity
• Human traits: α-Thalassemias,
G6PDH
• Bednets introduced in 1988
• Malaria elimination in Aneytium
using MDA and bednets
Chaves LF, Kaneko A, Taleo G, Pascual M, Wilson ML. Malaria transmission pattern resilience to climatic variability is mediated by
insecticide-treated nets Malaria Journal. 2008, 7:100 doi:10.1186/1475-2875-7-100.
Hypothesis: Role of climate on malaria incidence affected by
introduction of insecticide treated bednets
Data sources:
Cases: Monthly 1983 - 1999 confirmed P falciparum and P vivax infections
Demographic, climatic data from Vanuatu government
Statistics:
Chaves LF, Kaneko A, Taleo G, Pascual M, Wilson ML. Malaria transmission pattern resilience to climatic variability is mediated by
insecticide-treated nets Malaria Journal. 2008, 7:100 doi:10.1186/1475-2875-7-100.
Chaves LF, Kaneko A, Taleo G, Pascual M, Wilson ML. Malaria transmission pattern resilience to climatic variability is mediated by
insecticide-treated nets Malaria Journal. 2008, 7:100 doi:10.1186/1475-2875-7-100.
Chaves LF, Kaneko A, Taleo G, Pascual M, Wilson ML. Malaria transmission pattern resilience to climatic variability is mediated by
insecticide-treated nets Malaria Journal. 2008, 7:100 doi:10.1186/1475-2875-7-100.
Conclusions:
• Seasonal cycles strongly correlated with temp.;
inter-annual with precipitation.
• Malaria dynamics underwent regime shift (1991)
following introduction ITN
• Incidence (both parasites) reduced by >50% when
~20% of population covered
• Influence of environmental drivers reduced by 30–80%
for climatic forces, and 33–54% for other factors
• Results emphasize need to implement control
programs focused on most vulnerable populations.
Chaves LF, Kaneko A, Taleo G, Pascual M, Wilson ML. Malaria transmission pattern resilience to climatic variability is mediated by
insecticide-treated nets Malaria Journal. 2008, 7:100 doi:10.1186/1475-2875-7-100.
Exposure
Disease
Exp A
Exp B
Exp C
Disease
Exp A
Exp B
Exp B
Exp C
Disease
Exp C
Exp A
Disease
Confounder? Effect
Modifier?
Causal Inference: Early Concepts
Factors necessary for causation of disease
•
•
•
•
•
•
•
•
•
(Hill)
Strength of association
Consistency of association
Specificity
Temporal Relationship
Dose response effect / biologic gradient
Plausibility
Congruence / coherence
Experimental evidence
Analogy
ARE THESE CRITERIA USEFUL OR IMPORTANT
TO ANALYSING CLIMATE – ID ASSOCIATIONS?
Causal Inference: Modern View
More nuanced consideration (Rothman)
• Outcomes are multi-causal
• Components to causality having differing
levels of importance
• Components are not always independent
ARE THESE CRITERIA USEFUL OR IMPORTANT
TO ANALYSING CLIMATE – ID ASSOCIATIONS?
Social Factors as part of the Ensemble of
Environmental Influences on IDs
Some Challenges
• How can we advance analyses of social drivers as they
interact with climate factors?
• Which data are needed to expand analyses in this direction?
• How can scientists with different conceptual frameworks, skill
sets, perspectives be encouraged to work together?
• What models (statistical, conceptual, qualitative, dynamical)
can be applied to address these many variables?
• What do we believe will be the “value added” of analyzing
such complex systems?
Knowing is not enough; we must apply.
Willing is not enough; we must do.
(Goethe)