This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
More than 35% of American adults are obese. For African American and Hispanic adults, as well as individuals residing in poorer or more racially segregated urban neighborhoods, the likelihood of obesity is even higher. Information and communication technologies (ICTs) may substitute for or complement community-based resources for weight management. However, little is currently known about health-specific ICT use among urban-dwelling people with obesity.
We describe health-specific ICT use and its relationship to measured obesity among adults in high-poverty urban communities.
Using data collected between November 2012 and July 2013 from a population-based probability sample of urban-dwelling African American and Hispanic adults residing on the South Side of Chicago, we described patterns of ICT use in relation to measured obesity defined by a body mass index (BMI) of ≥30 kg/m2. Among those with BMI≥30 kg/m2, we also assessed the association between health-specific ICT use and diagnosed versus undiagnosed obesity as well as differences in health-specific ICT use by self-reported comorbidities, including diabetes and hypertension.
The survey response rate was 44.6% (267 completed surveys/598.4 eligible or likely eligible individuals); 53.2% were African American and 34.6% were Hispanic. More than 35% of the population reported an annual income of less than US $25,000. The population prevalence of measured obesity was 50.2%. People with measured obesity (BMI≥30 kg/m2) were more likely to report both general (81.5% vs 67.0%,
In conclusion, ICT-based health resources may be particularly useful for people in high-poverty urban communities with isolated measured obesity, a population that is at high risk for poor health outcomes.
Obesity and obesity-related chronic diseases are leading drivers of health care costs in the United States [
The 2012 Pew Internet Health Tracking Survey examined the relationship between types of ICT use, including seeking information online about conditions, medications, or the experiences of others, and self-reported chronic disease [
On Chicago’s South Side, 55% of the population (approximately 528,000) lives at or below 200% federal poverty level; 77% of residents are African American, 13% are Hispanic [
In this study, we first describe overall patterns of current ICT use by measured obesity status. Based on the Pew findings for other chronic conditions, we hypothesized that although overall ICT use would be lower among respondents with measured obesity, health-specific ICT use would be higher. Second, among people with measured obesity (BMI≥30 kg/m2), we described health-specific ICT use comparing those with and without a physician’s diagnosis of obesity and by the presence of self-reported comorbid conditions, specifically diabetes and hypertension. The Health Belief Model suggests that willingness to perform a health behavior, such as using health-specific ICT resources, depends on the perceived need for action [
This analysis is based on data from the South Side Health and Vitality Studies (SSHVS) [
Individuals eligible for this study included those 35 years of age or older, English or Spanish speaking, and residing within the target region.
Study participants were sampled from 2 distinct regions, a total of 7 census tracts, on the South Side of Chicago. The northwest region was almost entirely African American (98%), based on 2010 US Census data [
Eligible participants were recruited between November 2012 and July 2013 through mailed letters, telephone calls, and home visits. Once informed consent was obtained, participants took part in an in-person, interviewer-administered structured interview, lasting approximately 1 hour. Participants could choose to complete the interview in English or Spanish. The interview collected sociodemographic characteristics, details on ICT device ownership and use, and self-reported medical history. Physical measures were obtained, including height, weight, waist circumference, systolic and diastolic blood pressure, and finger-stick dried blood specimens.
General ICT use was defined as any use of cell phones for texting, emailing, going online, or downloading apps, or any use of the Internet (accessed via computer or cell phone). Health-specific ICT activities included the following activities: looking up health information online, using a health-related mobile app, Web-based purchasing of medications, Web-based communication with providers, participation in online health support groups, and Web-based management of health records and/or benefits. Questions included on the survey instrument relating to health-specific ICT use were primarily based on a prior national-level survey, the Health Information National Trends Survey, which was modified by the study team to increase cultural appropriateness and aid comprehension (
Instructions: I’m going to list some ways people use the Internet. Some people have done these things and some have not. Please use Card #X to tell me how often you did these things in the past 12 months. In the past 12 months, how often have you...
□ Every day □ At least once a week □ At least once a month □ <Once a month □ Never
1. Bought some kind of medicine online. This includes prescription medicines, over-the-counter medicines, or herbal supplements?
2. Taken part in an online support group for people with a health or medical issue?
3. Used e-mail or the Internet to talk with a doctor or a doctor’s office or hospital?
4. Looked for health or medical information online?
5. Looked at or managed your health records online
5a: If yes, do you ever use the Internet to find health information because you did not want to ask a doctor?
□ Yes □ No □ Don’t Know □ Refused
6. Looked at or managed your health benefits, like filing an insurance claim online
Body mass index was calculated using measured height and weight collected during in-home interviews [
Population-based probability sample enrollment flowchart. BMI: body mass index.
All analyses for this population-based probability sample were weighted to account for differential selection probabilities and differential nonresponse. The response rate and the cooperation rate were calculated using the American Association for Public Opinion Research (AAPOR) definitions for response rate (RR3) and cooperation rate (COOP3) [
The response rate was 44.6% (267 completed surveys/598.4 eligible or likely eligible individuals) and the cooperation rate was 61.5% (267 completed surveys/434 eligible individuals contacted). In total, 267 individuals participated in the biosocial study.
Associations between sociodemographic characteristics and measured obesity (BMI≥30 kg/m2) status in the population are summarized in
The prevalence of general ICT use in the population was high (75.0%) but was more common among people with obesity when compared with people without obesity (81.5% obese vs 67% nonobese,
Among people with measured obesity, a physician’s diagnosis of obesity was associated with educational attainment; people with obesity who had been diagnosed by a physician were more likely to have achieved a high school diploma or have passed a general educational development (GED) test (41.2% diagnosed vs 24.6% undiagnosed,
Sociodemographic characteristics of the population and by measured obesity status (results of weighted analysis).
Characteristic | Total population |
Nonobese |
Measured obese |
||
N=267c | n=115 |
n=140 |
|||
Age, years | |||||
35-40 | 11.8 (7.5-16.0) | 9.5 (3.9-15.1) | 13.9 (7.3-20.6) | .42 | |
41-50 | 36.1 (28.8-43.5) | 38.1 (26.8-49.3) | 35.7 (25.4-45.9) | ||
51-60 | 22.4 (17.3-27.6) | 20.8 (13.3-28.3) | 23.5 (16.1-30.8) | ||
61-70 | 13 (8.2-17.7) | 17.5 (9.4-25.6) | 9.1 (3.3-14.9) | ||
71+ | 16.7 (11.1-22.4) | 14.1 (6.4-21.8) | 17.8 (9.4-26.2) | ||
Gender | |||||
Male | 42.2 (34.9-49.4) | 49.5 (38.5-60.5) | 36.8 (26.6-47.0) | .01 | |
Female | 57.8 (50.6-65.1) | 50.5 (39.5-61.5) | 63.2 (53.0-73.4) | ||
Race/ethnicity | |||||
Black, non-Hispanic | 53.2 (48.9-57.5) | 51.4 (42.4-60.4) | 55.2 (46.4-63.9) | .30 | |
Hispanic | 34.6 (28.7-40.5) | 39.1 (29.0-49.3) | 29.4 (19.4-39.4) | ||
Other | 12.2 (7.8-16.6) | 9.5 (3.6-15.3) | 15.4 (8.4-22.4) | ||
Income, US $ | |||||
<$25K | 36.2 (29.5-43.0) | 40.7 (30.1-51.3) | 34.2 (24.8-43.5) | .59 | |
$25K-$49K | 28.4 (22.0-34.7) | 24.1 (15.4-32.8) | 33.9 (24.4-43.5) | ||
$50K-$99K | 16.7 (11.2-22.2) | 14.2 (6.7-21.8) | 17.1 (8.9-25.3) | ||
≥$100K | 6.7 (2.8-10.7) | 8.4 (1.3-15.6) | 5.8 (1.2-10.4) | ||
Don't know/refused | 12.0 (6.6-17.5) | 12.5 (3.7-21.4) | 9.0 (2.4-15.5) | ||
Education | |||||
Middle school/ |
30.4 (23.5-37.4) | 33.3 (22.6-44.0) | 27.2 (17.8-36.6) | .55 | |
High school graduate/GEDd | 34.8 (27.8-41.8) | 37 (26.7-47.4) | 36.2 (26.1-46.3) | ||
Associates/ |
34.7 (28.1-41.3) | 29.7 (19.9-39.5) | 36.6 (27.4-45.9) | ||
Employment status | |||||
Unemployed | 14.3 (9.5-19.2) | 15.5 (7.7-23.2) | 14.7 (8.0-21.4) | .85 | |
Employed | 45.5 (38.2-52.7) | 46.4 (35.5-57.3) | 42.5 (32.3-52.6) | ||
Retired | 18.7 (13.2-24.1) | 18.4 (10.6-26.2) | 18.6 (10.6-26.6) | ||
Unable to work | 10.0 (5.4-14.5) | 7.1 (0.4-13.8) | 12.5 (5.9-19.1) | ||
Other | 11.6 (6.5-16.6) | 12.6 (4.3-20.8) | 11.8 (4.9-18.7) | ||
Health insurance | |||||
Uninsured | 24.6 (18.1-31.0) | 33.8 (23.0-44.5) | 17.1 (9.7-24.5) | .01 | |
Medicaid only | 8.9 (5.0-12.8) | 4.5 (0.8-8.1) | 13.7 (6.9-20.6) | ||
Medicare only | 12.4 (8.0-16.8) | 15.4 (7.8-23.1) | 10.8 (5.7-16.0) | ||
Private/other | 38.0 (31.0-44.9) | 35.4 (25.0-45.7) | 40.5 (30.5-50.5) | ||
Multiple | 16.2 (10.9-21.4) | 11.0 (4.8-17.2) | 17.9 (9.8-25.9) | ||
Physician visit in past year (% yes) | 79.8 (73.6-86.0) | 78.6 (69.5-87.7) | 82.5 (73.6-91.4) | .55 | |
Source of regular care (% yes) | 91.4 (87.4-95.4) | 88.6 (82.2-94.9) | 96.7 (93.1-1.0) | .04 |
a BMI: body mass index.
b
c A total of 12 people were missing a BMI value. These individuals were excluded from the chi-square analysis.
d GED: general educational development.
Information and communication technology–based activities of the population and by measured obesity status (results of weighted analysis).
ICTa activities (% reported yes) | Total population |
Measured nonobese |
Measured obese |
||
N=267c | n=115 |
n=140 |
|||
General ICT use | 75.0 (68.5-81.3) | 67.0 (56.7-77.3) | 81.5 (73.1-90.0) | .04 | |
Any health-specific use | 51.7 (44.5-59.0) | 41.2 (30.6-51.9) | 61.1 (51.0-71.1) | .01 | |
Seek health info online | 47.2 (40.0-54.5) | 38.7 (28.1-49.3) | 54.4 (44.1-64.6) | .04 | |
Use Web-based resources to avoid asking doctor | 20.2 (14.8-25.7) | 16.0 (8.9-23.5) | 23.9 (15.7-32.1) | .17 | |
Web-based access of health benefit info | 12.2 (7.8-16.6) | 5.8 (0.9-10.7) | 18.6 (11.2-26.1) | .01 | |
Participate in online health support group | 9.3 (5.1-13.4) | 8.4 (2.8-14.0) | 10.5 (4.0-17.0) | .63 | |
Web-based access of health records | 8.8 (4.9-12.7) | 4.9 (−0.1 to 9.9) | 12.7 (6.5-18.9) | .08 | |
Web-based medication purchasing | 7.7 (4.3-11.1) | 5.1 (0.9-9.3) | 10.0 (4.5-15.5) | .17 | |
Web-based communication with providers | 9.7 (5.9-13.5) | 7.3 (2.5-12.1) | 12.4 (6.2-18.6) | .20 | |
Use health-related mobile app | 7.6 (3.8-11.4) | 5.0 (−0.03 to 10.0) | 9.0 (3.6-14.3) | .31 |
a ICT: information and communication technology.
b
c A total of 12 people were missing a body mass index value. These individuals were excluded from the chi-square analysis.
Sociodemographic characteristics of the measured obese population and by obesity diagnosis status (results of weighted analysis).
Characteristic | Total measured obesea |
Undiagnosed obese |
Diagnosed obese |
|||
n=140 |
n=43 |
n=97 |
||||
Age, years | ||||||
35-40 | 13.9 (7.3-20.6) | 12.3 (1.4-23.3) | 14.6 (6.3-23.0) | .78 | ||
41-50 | 35.7 (25.4-45.9) | 32.2 (14.2-50.3) | 37.2 (24.7-49.7) | |||
51-60 | 23.5 (16.1-30.8) | 26.9 (12.4-41.4) | 22.0 (13.5-30.5) | |||
61-70 | 9.1 (3.3-14.9) | 5.6 (−0.9 to 12.2) | 10.5 (2.8-18.3) | |||
71+ | 17.8 (9.4-26.2) | 22.9 (5.1-40.7) | 15.6 (6.6-24.7) | |||
Gender | ||||||
Male | 36.8 (26.6-47.0) | 45.5 (26.9-64.1) | 33.0 (20.7-45.4) | .15 | ||
Female | 63.2 (53.0-73.4) | 54.5 (35.9-73.1) | 67.0 (54.6-79.3) | |||
Race/ethnicity | ||||||
Black, non-Hispanic | 55.2 (46.4-63.9) | 46.3 (30.3-62.3) | 59.0 (49.0-69.0) | .32 | ||
Hispanic | 29.4 (19.4-39.4) | 40.5 (22.6-58.4) | 24.6 (12.7-36.5) | |||
Other | 15.4 (8.4-22.4) | 13.1 (1.4-24.9) | 16.4 (7.9-25.0) | |||
Income, US $ | ||||||
<$25K | 34.2 (24.8-43.5) | 32.6 (16.7-48.5) | 34.9 (23.3-46.5) | .41 | ||
$25K-$49K | 33.9 (24.4-43.5) | 36.4 (18.0-54.7) | 32.9 (21.7-44.0) | |||
$50K-$99K | 17.1 (8.9-25.3) | 10.3 (−1.1 to 21.7) | 20.1 (9.5-30.6) | |||
≥$100K | 5.8 (1.2-10.4) | 4.3 (−1.8 to 10.3) | 6.5 (0.4-12.5) | |||
Don't know/refused | 9.0 (2.4-15.5) | 16.5 (0.5-32.4) | 5.7 (−0.4 to 11.9) | |||
Education | ||||||
Middle school/ |
27.2 (17.8-36.6) | 49.5 (30.8-68.1) | 17.5 (8.0-27.1) | .01 | ||
High school graduate/GEDd | 36.2 (26.1-46.3) | 24.6 (9.0-40.1) | 41.2 (28.7-53.7) | |||
Associates/ |
36.6 (27.4-45.9) | 26.0 (11.1-40.8) | 41.2 (29.6-52.8) | |||
Employment status | ||||||
Unemployed | 14.7 (8.0-21.4) | 20.3 (5.3-35.3) | 12.3 (5.3-19.2) | .22 | ||
Employed | 42.5 (32.3-52.6) | 37.5 (20.0-55.1) | 44.6 (32.2-56.9) | |||
Retired | 18.6 (10.6-26.6) | 8.9 (0.5-17.4) | 22.8 (12.1-33.5) | |||
Unable to work | 12.5 (5.9-19.1) | 20.2 (3.2-37.2) | 9.1 (3.8-14.5) | |||
Other | 11.8 (4.9-18.7) | 13.0 (0.7-25.3) | 11.2 (2.8-19.7) | |||
Health insurance | ||||||
Uninsured | 17.1 (9.7-24.5) | 14.0 (0.8-27.3) | 18.4 (9.3-27.5) | .56 | ||
Medicaid only | 13.7 (6.9-20.6) | 21.5 (5.9-37.1) | 10.4 (3.6-17.2) | |||
Medicare only | 10.8 (5.7-16.0) | 9.9 (1.4-18.4) | 11.2 (4.8-17.7) | |||
Private/other | 40.5 (30.5-50.5) | 33.5 (16.9-50.0) | 43.5 (31.6-55.4) | |||
Multiple | 17.9 (9.8-25.9) | 21.1 (3.6-38.6) | 16.5 (7.7-25.2) | |||
Physician visit in past year (% yes) | 82.5 (73.6-91.4) | 79.3 (63.7-95.0) | 83.9 (72.9-94.8) | .63 | ||
Source of regular care (% yes) | 96.7 (93.1-1.0) | 89.1 (77.3-100) | 100 (100-100) | .01 |
a A total of 12 people were missing a body mass index value. These individuals were excluded from the chi-square analysis.
b BMI: body mass index.
c
d GED: general educational development.
Among people with measured obesity, a self-reported diagnosis of hypertension (46.8%) or diabetes (16.8%) was prevalent. Only 36.8% of people with measured obesity had no diagnosis of diabetes or hypertension; 20.5% of the population had all 3 conditions. Isolated measured obesity was associated with higher rates of health-specific ICT use than measured obesity plus comorbid diabetes and/or hypertension diagnosis (77.1% obesity only vs 47.4% obesity and hypertension or diabetes vs 60.7% obesity, hypertension, and diabetes,
Comparing information and communication technology activities by presence of comorbid conditions (results of weighted analysis).
ICTa activities (% yes) | Measured obesity onlyc |
Measured obesity and hypertension or diabetes |
Measured obesity, hypertension, and diabetes |
||
n=44 |
n=68 |
n=28 |
|||
General ICT use | 93.7 (87.2-100) | 75.0 (60.3-89.6) | 73.4 (53.1-93.7) | .05 | |
Any health-specific use | 77.1 (61.4-92.7) | 47.4 (33.1-61.8) | 60.7 (39.3-82.0) | .04 | |
Seek health info online | 67.5 (49.8-85.2) | 45.9 (31.7-60.0) | 48.4 (26.6-70.3) | .15 | |
Use Web-based resources to |
30.8 (15.1-46.5) | 21.9 (10.5-33.4) | 15.6 (−1.2 to 32.4) | .41 | |
Web-based access of health |
27.6 (12.4-42.8) | 18.0 (7.1-28.9) | 4.0 (−1.8 to 9.8) | .04 | |
Participate in online health |
15.3 (1.7-28.9) | 10.5 (1.1-19.8) | 2.0 (−2.0 to 6.0) | .22 | |
Web-based access of health |
16.4 (4.2-28.6) | 9.1 (2.0-16.1) | 13.7 (−1.3 to 28.7) | .57 | |
Web-based medication |
10.8 (0.78-20.7) | 8.1 (1.4-14.7) | 12.8 (−2.0 to 27.5) | .81 | |
Web-based communication |
13.1 (1.8-24.5) | 10.2 (3.0-17.4) | 15.7 (−1.0 to 32.4) | .80 | |
Use health-related mobile |
9.7 (0.8-18.6) | 11.7 (2.2-21.3) | 2.0 (−2.0 to 6.0) | .29 |
a ICT: information and communication technology.
b
c A total of 12 people were missing a body mass index value. These individuals were excluded from the chi-square analysis.
This study describes ICT use in high-poverty African American and Hispanic communities on Chicago’s South Side with a disproportionate burden of obesity and obesity-related diseases and examines the association between obesity and ICT use. To our knowledge, this is the only study to ascertain measured BMI, self-reported obesity diagnoses, and ICT use from the same sample. This design, albeit limited by a relatively small sample size, enabled us to generate three new findings. First, we found that ICT use patterns differed by measured obesity status; people with obesity had statistically significant higher rates of both general ICT and health-specific ICT use compared with people without obesity. Second, among people with measured obesity, a physician’s diagnosis of obesity was not associated with higher rates of health-specific ICT use or use of Web-based health-related information sources, but it was associated with a higher rate of using Web-based resources to avoid asking questions of a doctor. Finally, an unexpected association between comorbidity burden and health-specific ICT use was found. The highest rates of ICT use were among people with measured obesity only, as compared with those with measured obesity who reported one or two common comorbidities.
In contrast to the Pew study findings that showed lower ICT use among people with other common chronic conditions [
Unlike other chronic conditions (eg, hypertension and diabetes), obesity is an outward-facing condition. The social stigma of obesity may lead individuals to ICT-based resources rather than medical care for their health needs [
Regardless of the drivers behind increased health-specific ICT use, our study suggests that obesity may be a useful target for health-specific ICT-based interventions. The high prevalence of obesity among residents on Chicago’s South Side and in other high-poverty, minority communities, along with the high rate of health-specific use, indicates an already online population with high health needs and risks. ICT-based resources could potentially not only aid in the self-care and management of obesity but also serve as an entry point to provide information and support for routine preventive care and other important health topics in this population.
Among studies of health-related ICT use, this survey was unique in its combination of ICT use measures with anthropometric measures and assessment of self-reported chronic diseases [
Obesity status may also be more subject to individual perceptions than other chronic conditions. Past research has demonstrated that people have difficulty in assessing ideal weight [
Comorbid diagnoses of diabetes and/or hypertension were not found to be associated with a higher likelihood of health-specific ICT use among individuals with measured obesity. This finding contrasts with findings from a 2007 phone survey of US adults that demonstrated a positive correlation between the number of chronic conditions (did not include obesity) and engagement in selected health-specific ICT activities [
On the basis of the study findings, extant literature, and clinical experience, we propose a preliminary conceptual framework for the relationship between obesity and use of health-specific ICT (
The study has several limitations. First, the conservative AAPOR response rate calculation (44.6%) is lower than desired, but it is also consistent with or higher than that reported for other similar surveys in this and other urban populations [
In this high-poverty urban population, the majority of people with measured obesity reported use of technology for health-specific reasons. This high-risk, already online population presents an opportunity for ICT-based health resources to impact health, especially in communities where the burden of obesity is high. However, understanding current use patterns and potential opportunities for health-specific ICT-based resources is only a first step. The critical next step is evaluating the ability of these technology-based resources to meaningfully impact health care and health outcomes in this high-need, high-risk population.
Proposed conceptual framework for the relationship between obesity and health-specific information and communication technology (ICT) use derived from literature, study results, and clinical experience. Adapted from Andersen’s Behavioral Model of Health Services Utilization, the proposed model incorporates obesity as a specific use case. The dashed lines highlight two incompletely understood domains: (1) the relationship between obesity and health-specific ICT use and (2) the potential dual role of health-specific ICT as both an access point to and a replacement for traditional health resources. BMI: body mass index.
Frequency of health-specific information and communication technology activities.
American Association for Public Opinion Research
body mass index
information and communication technology
Institutional Review Board
South Side Health and Vitality Studies
general educational development
The authors would like to acknowledge the entire South Side Health and Vitality Studies (SSHVS) research team for their contribution to this work. We would also like to acknowledge the University of Chicago Survey Lab for its role in field operations of the SSHVS population health study, including conducting the survey interviews and biological measure collection.
The efforts of AG on this manuscript were supported by the Department of Veterans Affairs and by the Kaiser Permanente Northern California Division of Research. Support for the SSHVS population health studies was provided by the University of Chicago Medicine Urban Health Initiative, individual philanthropy to the Lindau Laboratory at the University of Chicago, and the Chicago Core on Biomarkers in Population-Based Aging Research supported by the National Institutes of Health (NIH)/National Institute on Aging grant P30 AG012857 at NORC and the University of Chicago. The efforts of STL, JM, and VE on this manuscript were also supported in part by the NIH/National Institute on Aging grant 1R01 AG 047869-01 at the University of Chicago.
The contents of this presentation are solely the responsibility of the authors and do not necessarily represent the official views of the US Department of Veterans Affairs, or the US Department of Health and Human Services or any of its agencies.
AG contributed to study conception, design, data analysis and interpretation, and manuscript construction. JM contributed to the study design and execution, data analysis and interpretation, and manuscript content. LBG and VE contributed to the data analysis and interpretation, and manuscript content. RM contributed to study design, data interpretation, and manuscript content. RH and MBW contributed to design of the population health study data collection instrument and protocol, community engagement activities in the preparation and dissemination phases of the study, interpretation of the analytic findings, and review and editing of this manuscript. STL led the research team that administered MAPSCorps and the South Side Health and Vitality Studies population health study data collection. STL also provided material support, contributed to design and execution of the study, and oversaw all aspects of data collection, analysis, and interpretation. She has contributed original writing to and edited drafts of the manuscript for important intellectual content. She has provided material support to the execution of this study.
RM has grant support from NIH, National Heart, Lung, and Blood Institute (NHLBI) grant K23 (109083) and NHLBI R01 (122457), but this funding was not used to conduct the study. Her time on this project was part of her leadership role with the Robert Wood Johnson Clinical Scholars program. STL is founder and co-owner of a social impact company NowPow, LLC, developed as the sustainable business model expected by a Centers for Medicare & Medicaid Services (CMS) Health Care Innovation Award (1C1CMS330997-03-00, 2012-15). The CMS award did not directly support the research described in this manuscript. The other authors report no conflicts of interest.