Integrating the effects of space, environment, and social networks - - PowerPoint PPT Presentation

Integrating the effects of space, environment, and social networks in cholera vaccines

Michael Emch Carolina Perez-Heydrich

University of North Carolina at Chapel Hill

Cholera Background

Vibrio cholerae: curved gram-negative rods with a polar flagellum Cholera toxin Watery diarrhea Dehydration If untreated ~50% case fatality rate, if treated ~1%

ICDDR,B field site: Matlab, Bangladesh 142 villages Longitudinal DSS since 1966 120 CHWs Health surveillance Diarrhea hospital & laboratory

Study Area

ICDDR,B Community Health and Demographic Surveillance

Does the oral cholera vaccine provide an indirect effect (i.e., is there herd protection)? Does neighborhood vaccination proportion affect disease incidence? Is the answer different when defining neighborhoods by Euclidean distance, environmental connectivity, and social connectivity?

Preventing Cholera

Protective efficacy is the proportionate reduction of the incidence of the target infection by vaccination

2 3 4 1 5 6 7 11 12 1000 meters 9 8 10 1000 meters

Identification Number Vaccinee Population Placebo Population Vaccinee Cholera Cases Placebo Cholera Cases 1 12 7 1 2 2 6 3 23 25 4 24 22 1 2 5 25 32 6 12 25 1 1 7 25 45 8 22 23 9 34 25 1 10 25 20 Total

204 230 2 5

Vaccinee Incidence

0.0098

Placebo Incidence

0.022

Efficacy

0.55

αi= protective efficacy in neighborhood i ϑi=vaccinee incidence rate in neighborhood i λi= nonvaccinee incidence rate in neighborhood i

100 ) 1 ( x

i i i

λ ϑ α − =

 Herd protection is protection of an individual from a

disease because others are immune to the disease.

 This is called herd protection because non-immunized

people in the population are protected since most people in the population, i.e., the herd, are protected.

Herd Protection

Background and Data: Cholera Vaccine Trial



Clinical trial conducted through the International Centre for Diarrhoel Disease Research, Bangladesh (ICDDR,B) between



All children (2-15 yrs old) and women (> 15 yrs old) randomly assigned to one of three treatment assignments: 2 different vaccine treatments and 1 placebo group



Two vaccine doses were given to 50,499 people and two placebo doses were given to 25,252 in the target group in six-week intervals.



The vaccine trial used a passive surveillance system to identify cholera cases from the study area.

Oral cholera vaccine coverage

Level of vaccine coverage# All recipients of >2 doses Vaccinees Placebo recipients PE

N % of total N Cases Incidence rate/1000/ Year‡ N Cases Incidence rate/1000/ year§ <28% 8,479 11.5 5,627 15 2.66 2,852 20 7.01 61%* 28-35% 13,312 18.0 8,883 22 2.47 4,429 26 5.87 57%* 36-40% 16,275 22.0 10,772 17 1.57 5,503 26 4.72 66%* 41-50% 17,314 23.4 11,513 26 2.25 5,801 27 4.65 51%* 51%+ 18,623 25.1 12,541 16 1.27 6,082 9 1.47 13% Total 74,003 100 49,336 96 1.94 24,667 108 4.37 55%† ‡Spearman’s rank P=0.02 §Spearman’s rank P=0.08 * P<.01 † P<.001

Note 1: A multivariate model (generalized estimating equations with logit link function) controlled for the potential confounding variables (age, gender, river distance, treatment center distance, dysentery). Note 2: There was not an inverse relationship between dysentery incidence and coverage (Spearman’s rank -.3-, p=0.62).

Cholera incidence rate and protective efficacy (PE) among >2 dose recipients by the level of cholera vaccine coverage

• Lancet. (2005) 366 (9479):44-9; Health & Place (2007) 13: 238-248.



Killed oral cholera vaccines are not licensed for infants and young children.



Cholera is known to be an important problem in these younger age groups



We investigated whether older children and adults can confer herd protection to children too young to be vaccinated by determining whether the incidence was lower with higher coverage during the first year of surveillance.

Are young children who can’t be vaccinated also protected?

Level of vaccine coverage† Total no. of children<24 mos Cases Risk/1,000‡ <28% 2,378 45 18.92 28-35% 2,371 27 11.38 36-40% 2,297 36 15.67 41-50% 2,207 29 13.14 51%+ 2,205 19 8.61 Total 11,458 156 13.61

• Defined as children <24 months of age at the time of dosing in the trial

† Within 500 meters of bari ‡ P= .004 for trend.

Incidence of cholera among children too young to be vaccinated* by level of vaccine coverage

• * Defined as children <24 months of age at the time of dosing in the trial
• † Multivariate odds ratio for the cited variable, adjusted for all other variables in the table, in a
• model using Generalized Estimating Equations (GEE) with the logit link function.
• ‡ In Model 1, the level of coverage is based on all persons eligible for vaccination
• § In Model 2, the level of coverage is expressed separately for adult women (>15y)
• and for children (2-15y)

Predictors of the risk of cholera among children too young to be vaccinated* during one year of follow-up in the Bangladesh cholera vaccine trial

Model 1‡ Model 2§ Factors OR† 95%CI P-value OR† 95%CI P-value Age (in years) 1.46 1.10-1.92 .01 1.46 1.10-1.92 .01 Male 1.11 0.81-1.53 .50 1.11 0.81-1.52 .52 Muslim 1.10 0.67-1.82 .70 1.03 0.62-1.71 .90 Distance of the child’s residence to nearest river (km) 1.06 0.89-1.26 .46 1.07 0.90-1.27 .42 Distance of the child’s residence to nearest treatment center (km) 1.02 0.93-1.13 .62 0.99 0.89-1.10 .88 Experienced dysentery during follow-up 4.19 2.00-8.74 <.001 4.11 1.97-8.54 <.001 Overall vaccine coverage of the child’s bari¶ 0.98* 0.96-0.99 <.001

• Vaccine coverage of women

>15yrs in the child’s bari

• 0.95

0.92-0.99 <.01 Vaccine coverage of children aged 2-15 yrs in the child’s bari

• 1.02

0.98-1.06 .24



Incidence ranged from 18.9 in clusters in the lowest quintile of vaccine coverage to 8.6 in clusters in the highest quintile (P= .004).



Vaccine coverage of adult women (P< .01), but not of older children, was independently associated with a lower risk of cholera in under two year-olds.

Young Children Herd Immunity Findings

The Pediatric Infectious Disease Journal. (2008) 27(1): 33-37.

Spatial and Environmental Risk Factors



Emch, M., et al. (2009). Spatial and environmental connectivity analysis in a cholera vaccine

• trial. Social Science and
• Medicine. 68: 631-637.
• Addressed how environmental

networks could influence risk of cholera infection among placebo recipients.

• Digitized ponds using Quickbird

satellite imagery to define pond networks

• Defined vaccine coverage at the

bari-level as proportion of vaccinated individuals that were connected by shared ponds

• Found that the risk of cholera

among placebo recipients declined as vaccine coverage within pond networks increased

Kinship Networks

 Connections are clustered spatially

and socially.

 Sons live near fathers  Purdah: limits social interactions

• f women, expected to stay at

home or only leave out of necessity

 Restricts social interactions to be

mostly kin-based

 Through DSS, individuals can be

traced through time, and migrations between baris are captured

• Individual-level migrations link

baris

• Bari connections are mutual and

non-directional

 Main assumption: when an

individual moves, he/she maintains contact with the previous bari of residence

Spatial and Social Network in disease tranmission

 Emch, M., et al. (2012).

Integration of spatial and social network analysis in disease transmission

• studies. Annals of the

American Geographers Association.

• Compared the spatial

and social clustering of cholera and shigellosis through time (1983- 2003) within rural Bangladesh

• Social networks

intermittently important to disease incidence, but spatial clustering more important

Significance of spatial clustering Significance of social clustering

Social Network analysis in oral cholera vaccine evaluation



Root, E.D., et al. (2011). “The role of vaccine coverage within social networks in cholera vaccine efficacy.” PLoS One. 6(7): 1-8.

• Addressed how vaccine

coverage within bari-level and household-level kinship networks were associated with cholera incidence among placebo recipients

• Risk of cholera inversely

related to level of vaccine coverage in bari-level social networks

Ongoing model building

Summary of variables included in analysis Demography

SES Mean Age

Environmental

Tube well density Pond density Distance to River

Network- related (Pond and Kinship)

Ties to cases Ties to vaccinees Betweenness centrality

Spatial

1000m neighborhood adjacency matrices

Zero-inflation

Distance to hospital

Pond and Kinship Networks

 Considered environmental

connectivity via ponds

 Bari linkages defined according

to shared ponds within 200 m of baris

 Pond ties were defined as a

binary relationship (tie/no tie)

 Betweenness centrality –

location within the network that would predispose individuals within a bari to greater risk of infectious contacts

14 Models assessed

 Dem  Dem + Env  Dem + Env + SN  Dem + Env + SN +

Space

 Dem + SN  Dem + SN + Space  Dem + Space  Env  Env + SN  Env + SN + Space  Env + Space  SN  SN + Space  Space * * 75,000 iterations with a burn-in of 30,000 iterations

Excess risk as a measure of efficacy

• Excess risk: change in risk associated with exposure

relative to the unexposed, referent group

Results: Model Comparison

Risk Map

Efficacy of vaccination

I ndirect effects Direct effects

The oral cholera vaccine confers indirect herd

protection.

It works better than we thought so we need to

STOCKPILE THE VACCINE and we are now.

We wouldn’t know this if we hadn’t incorporated

space into the study.

Conclusions and Implications

MORE INFO www.cpc.unc.edu/projects/spatialhealthgroup