Assessing the Online Social Environment for Surveillance of Obesity Prevalence

Rumi Chunara, Lindsay Bouton, John W. Ayers, John S. Brownstein

Research output: Contribution to journalArticle

Abstract

Background:Understanding the social environmental around obesity has been limited by available data. One promising approach used to bridge similar gaps elsewhere is to use passively generated digital data.Purpose:This article explores the relationship between online social environment via web-based social networks and population obesity prevalence.Methods:We performed a cross-sectional study using linear regression and cross validation to measure the relationship and predictive performance of user interests on the online social network Facebook to obesity prevalence in metros across the United States of America (USA) and neighborhoods within New York City (NYC). The outcomes, proportion of obese and/or overweight population in USA metros and NYC neighborhoods, were obtained via the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance and NYC EpiQuery systems. Predictors were geographically specific proportion of users with activity-related and sedentary-related interests on Facebook.Results:Higher proportion of the population with activity-related interests on Facebook was associated with a significant 12.0% (95% Confidence Interval (CI) 11.9 to 12.1) lower predicted prevalence of obese and/or overweight people across USA metros and 7.2% (95% CI: 6.8 to 7.7) across NYC neighborhoods. Conversely, greater proportion of the population with interest in television was associated with higher prevalence of obese and/or overweight people of 3.9% (95% CI: 3.7 to 4.0) (USA) and 27.5% (95% CI: 27.1 to 27.9, significant) (NYC). For activity-interests and national obesity outcomes, the average root mean square prediction error from 10-fold cross validation was comparable to the average root mean square error of a model developed using the entire data set.Conclusions:Activity-related interests across the USA and sedentary-related interests across NYC were significantly associated with obesity prevalence. Further research is needed to understand how the online social environment relates to health outcomes and how it can be used to identify or target interventions.

Original languageEnglish (US)
Article numbere61373
JournalPLoS One
Volume8
Issue number4
DOIs
StatePublished - Apr 24 2013

Fingerprint

Disease control
social environment
Social Environment
Television
Linear regression
Mean square error
obesity
Obesity
Health
monitoring
confidence interval
Confidence Intervals
social networks
Social Support
Population
Centers for Disease Control and Prevention
television
Centers for Disease Control and Prevention (U.S.)
cross-sectional studies
Linear Models

ASJC Scopus subject areas

  • Agricultural and Biological Sciences(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Medicine(all)

Cite this

Assessing the Online Social Environment for Surveillance of Obesity Prevalence. / Chunara, Rumi; Bouton, Lindsay; Ayers, John W.; Brownstein, John S.

In: PLoS One, Vol. 8, No. 4, e61373, 24.04.2013.

Research output: Contribution to journalArticle

Chunara, Rumi ; Bouton, Lindsay ; Ayers, John W. ; Brownstein, John S. / Assessing the Online Social Environment for Surveillance of Obesity Prevalence. In: PLoS One. 2013 ; Vol. 8, No. 4.
@article{f9857a34ec1e4314bce9b07a02b697fc,
title = "Assessing the Online Social Environment for Surveillance of Obesity Prevalence",
abstract = "Background:Understanding the social environmental around obesity has been limited by available data. One promising approach used to bridge similar gaps elsewhere is to use passively generated digital data.Purpose:This article explores the relationship between online social environment via web-based social networks and population obesity prevalence.Methods:We performed a cross-sectional study using linear regression and cross validation to measure the relationship and predictive performance of user interests on the online social network Facebook to obesity prevalence in metros across the United States of America (USA) and neighborhoods within New York City (NYC). The outcomes, proportion of obese and/or overweight population in USA metros and NYC neighborhoods, were obtained via the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance and NYC EpiQuery systems. Predictors were geographically specific proportion of users with activity-related and sedentary-related interests on Facebook.Results:Higher proportion of the population with activity-related interests on Facebook was associated with a significant 12.0{\%} (95{\%} Confidence Interval (CI) 11.9 to 12.1) lower predicted prevalence of obese and/or overweight people across USA metros and 7.2{\%} (95{\%} CI: 6.8 to 7.7) across NYC neighborhoods. Conversely, greater proportion of the population with interest in television was associated with higher prevalence of obese and/or overweight people of 3.9{\%} (95{\%} CI: 3.7 to 4.0) (USA) and 27.5{\%} (95{\%} CI: 27.1 to 27.9, significant) (NYC). For activity-interests and national obesity outcomes, the average root mean square prediction error from 10-fold cross validation was comparable to the average root mean square error of a model developed using the entire data set.Conclusions:Activity-related interests across the USA and sedentary-related interests across NYC were significantly associated with obesity prevalence. Further research is needed to understand how the online social environment relates to health outcomes and how it can be used to identify or target interventions.",
author = "Rumi Chunara and Lindsay Bouton and Ayers, {John W.} and Brownstein, {John S.}",
year = "2013",
month = "4",
day = "24",
doi = "10.1371/journal.pone.0061373",
language = "English (US)",
volume = "8",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "4",

}

TY - JOUR

T1 - Assessing the Online Social Environment for Surveillance of Obesity Prevalence

AU - Chunara, Rumi

AU - Bouton, Lindsay

AU - Ayers, John W.

AU - Brownstein, John S.

PY - 2013/4/24

Y1 - 2013/4/24

N2 - Background:Understanding the social environmental around obesity has been limited by available data. One promising approach used to bridge similar gaps elsewhere is to use passively generated digital data.Purpose:This article explores the relationship between online social environment via web-based social networks and population obesity prevalence.Methods:We performed a cross-sectional study using linear regression and cross validation to measure the relationship and predictive performance of user interests on the online social network Facebook to obesity prevalence in metros across the United States of America (USA) and neighborhoods within New York City (NYC). The outcomes, proportion of obese and/or overweight population in USA metros and NYC neighborhoods, were obtained via the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance and NYC EpiQuery systems. Predictors were geographically specific proportion of users with activity-related and sedentary-related interests on Facebook.Results:Higher proportion of the population with activity-related interests on Facebook was associated with a significant 12.0% (95% Confidence Interval (CI) 11.9 to 12.1) lower predicted prevalence of obese and/or overweight people across USA metros and 7.2% (95% CI: 6.8 to 7.7) across NYC neighborhoods. Conversely, greater proportion of the population with interest in television was associated with higher prevalence of obese and/or overweight people of 3.9% (95% CI: 3.7 to 4.0) (USA) and 27.5% (95% CI: 27.1 to 27.9, significant) (NYC). For activity-interests and national obesity outcomes, the average root mean square prediction error from 10-fold cross validation was comparable to the average root mean square error of a model developed using the entire data set.Conclusions:Activity-related interests across the USA and sedentary-related interests across NYC were significantly associated with obesity prevalence. Further research is needed to understand how the online social environment relates to health outcomes and how it can be used to identify or target interventions.

AB - Background:Understanding the social environmental around obesity has been limited by available data. One promising approach used to bridge similar gaps elsewhere is to use passively generated digital data.Purpose:This article explores the relationship between online social environment via web-based social networks and population obesity prevalence.Methods:We performed a cross-sectional study using linear regression and cross validation to measure the relationship and predictive performance of user interests on the online social network Facebook to obesity prevalence in metros across the United States of America (USA) and neighborhoods within New York City (NYC). The outcomes, proportion of obese and/or overweight population in USA metros and NYC neighborhoods, were obtained via the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance and NYC EpiQuery systems. Predictors were geographically specific proportion of users with activity-related and sedentary-related interests on Facebook.Results:Higher proportion of the population with activity-related interests on Facebook was associated with a significant 12.0% (95% Confidence Interval (CI) 11.9 to 12.1) lower predicted prevalence of obese and/or overweight people across USA metros and 7.2% (95% CI: 6.8 to 7.7) across NYC neighborhoods. Conversely, greater proportion of the population with interest in television was associated with higher prevalence of obese and/or overweight people of 3.9% (95% CI: 3.7 to 4.0) (USA) and 27.5% (95% CI: 27.1 to 27.9, significant) (NYC). For activity-interests and national obesity outcomes, the average root mean square prediction error from 10-fold cross validation was comparable to the average root mean square error of a model developed using the entire data set.Conclusions:Activity-related interests across the USA and sedentary-related interests across NYC were significantly associated with obesity prevalence. Further research is needed to understand how the online social environment relates to health outcomes and how it can be used to identify or target interventions.

UR - http://www.scopus.com/inward/record.url?scp=84876540246&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84876540246&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0061373

DO - 10.1371/journal.pone.0061373

M3 - Article

VL - 8

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 4

M1 - e61373

ER -