National weighting of data from the Behavioral Risk Factor Surveillance System (BRFSS)

Can online self-reports assist in real-time identification of influenza vaccination uptake? A cross-sectional study of influenza vaccine-related tweets in the USA, 2013-2017

INTRODUCTION

The Centers for Disease Control and Prevention (CDC) spend significant time and resources to track influenza vaccination coverage each influenza season using national surveys. Emerging data from social media provide an alternative solution to surveillance at both national and local levels of influenza vaccination coverage in near real time.

OBJECTIVES

This study aimed to characterise and analyse the vaccinated population from temporal, demographical and geographical perspectives using automatic classification of vaccination-related Twitter data.

METHODS

In this cross-sectional study, we continuously collected tweets containing both influenza-related terms and vaccine-related terms covering four consecutive influenza seasons from 2013 to 2017. We created a machine learning classifier to identify relevant tweets, then evaluated the approach by comparing to data from the CDC's FluVaxView. We limited our analysis to tweets geolocated within the USA.

RESULTS

We assessed 1 124 839 tweets. We found strong correlations of 0.799 between monthly Twitter estimates and CDC, with correlations as high as 0.950 in individual influenza seasons. We also found that our approach obtained geographical correlations of 0.387 at the US state level and 0.467 at the regional level. Finally, we found a higher level of influenza vaccine tweets among female users than male users, also consistent with the results of CDC surveys on vaccine uptake.

CONCLUSION

Significant correlations between Twitter data and CDC data show the potential of using social media for vaccination surveillance. Temporal variability is captured better than geographical and demographical variability. We discuss potential paths forward for leveraging this approach.

Comparisons between 2015 US asthma prevalence and two measures of asthma burden by racial/ethnic group

Objective: Asthma affects approximately 20,383,000 Americans and costs the US $56 billion dollars annually. Asthma burden is not uniform across different racial/ethnic groups. Although annual asthma prevalences by racial/ethnic group are provided by the CDC, these numbers do not account for the influence of other sociodemographic and clinical factors. In this study, we evaluated several measures of asthma burden that controlled for sociodemographic and clinical factors and compared them to 2015 CDC asthma prevalences by racial/ethnic group. Methods: We obtained unadjusted asthma predictions directly from the 2015 Behavioral Risk Factor Surveillance System (BRFSS) survey. To estimate average adjusted predictions (AAP) and adjusted predictions at the means (APM) for asthma, we controlled for sociodemographic and clinical factors through logistic models and subsequently calculated marginal probabilities for these results. Results: We found substantial differences at the national and state level between unadjusted prevalences and AAPs and APMs for asthma by race/ethnicity group. In addition, states that had the highest and lowest unadjusted prevalences differed from states that had the highest and lowest AAPs and APMs. For unadjusted prevalences, AAPs, and APMs, we found that states that had the highest prevalences or probabilities of asthma for Whites tended to have the widest disparities between the estimates for Whites and those of other races. Conclusions: The study's results highlight the need to account for sociodemographic and clinical factors when addressing asthma burden by racial/ethnic group across the US.

Risk factors for cardiovascular disease among active adult US scuba divers

Cardiovascular factors among uninjured active adult recreational scuba divers in the USA are described. Scuba diving as an activity was included in 2011, 2013, and 2015 Behavioral Risk Factor Surveillance System data. One-third of active US scuba divers were aged ≥50 years and/or reported prior high cholesterol, around half were overweight, more than half reported having smoked cigarettes, and 32% reported hypertension or borderline hypertension. High cholesterol, hypertension, high body mass index, and smoking status should all be addressed during routine diving fitness physician assessments, to reduce the risk of mortality while diving.

An ontology-guided semantic data integration framework to support integrative data analysis of cancer survival

BACKGROUND

Cancer is the second leading cause of death in the United States, exceeded only by heart disease. Extant cancer survival analyses have primarily focused on individual-level factors due to limited data availability from a single data source. There is a need to integrate data from different sources to simultaneously study as much risk factors as possible. Thus, we proposed an ontology-based approach to integrate heterogeneous datasets addressing key data integration challenges.

METHODS

Following best practices in ontology engineering, we created the Ontology for Cancer Research Variables (OCRV) adapting existing semantic resources such as the National Cancer Institute (NCI) Thesaurus. Using the global-as-view data integration approach, we created mapping axioms to link the data elements in different sources to OCRV. Implemented upon the Ontop platform, we built a data integration pipeline to query, extract, and transform data in relational databases using semantic queries into a pooled dataset according to the downstream multi-level Integrative Data Analysis (IDA) needs.

RESULTS

Based on our use cases in the cancer survival IDA, we created tailored ontological structures in OCRV to facilitate the data integration tasks. Specifically, we created a flexible framework addressing key integration challenges: (1) using a shared, controlled vocabulary to make data understandable to both human and computers, (2) explicitly modeling the semantic relationships makes it possible to compute and reason with the data, (3) linking patients to contextual and environmental factors through geographic variables, (4) being able to document the data manipulation and integration processes clearly in the ontologies.

CONCLUSIONS

Using an ontology-based data integration approach not only standardizes the definitions of data variables through a common, controlled vocabulary, but also makes the semantic relationships among variables from different sources explicit and clear to all users of the same datasets. Such an approach resolves the ambiguity in variable selection, extraction and integration processes and thus improve reproducibility of the IDA.

Surveillance for Certain Health Behaviors and Conditions Among States and Selected Local Areas - Behavioral Risk Factor Surveillance System, United States, 2015

PROBLEM

Chronic conditions and disorders (e.g., diabetes, cardiovascular diseases, arthritis, and depression) are leading causes of morbidity and mortality in the United States. Healthy behaviors (e.g., physical activity, avoiding cigarette use, and refraining from binge drinking) and preventive practices (e.g., visiting a doctor for a routine check-up, tracking blood pressure, and monitoring blood cholesterol) might help prevent or successfully manage these chronic conditions. Monitoring chronic diseases, health-risk behaviors, and access to and use of health care are fundamental to the development of effective public health programs and policies at the state and local levels.

REPORTING PERIOD

January-December 2015.

DESCRIPTION OF THE SYSTEM

The Behavioral Risk Factor Surveillance System (BRFSS) is an ongoing, state-based, random-digit-dialed landline- and cellular-telephone survey of noninstitutionalized adults aged ≥18 years residing in the United States. BRFSS collects data on health-risk behaviors, chronic diseases and conditions, access to and use of health care, and use of preventive health services related to the leading causes of death and disability. This report presents results for all 50 states, the District of Columbia, the Commonwealth of Puerto Rico (Puerto Rico), and Guam and for 130 metropolitan and micropolitan statistical areas (MMSAs) (N = 441,456 respondents) for 2015.

RESULTS

The age-adjusted prevalence estimates of health-risk behaviors, self-reported chronic health conditions, access to and use of health care, and use of preventive health services varied substantially by state, territory, and MMSA in 2015. Results are summarized for selected BRFSS measures. Each set of proportions refers to the median (range) of age-adjusted prevalence estimates for health-risk behaviors, self-reported chronic diseases or conditions, or use of preventive health care services by geographic jurisdiction, as reported by survey respondents. Adults with good or better health: 84.6% (65.9%-88.8%) for states and territories and 85.2% (66.9%-91.3%) for MMSAs. Adults with ≥14 days of poor physical health in the past 30 days: 10.9% (8.2%-17.2%) for states and territories and 10.9% (6.6%-19.1%) for MMSAs. Adults with ≥14 days of poor mental health in the past 30 days: 11.3% (7.3%-15.8%) for states and territories and 11.4% (5.6%-20.5%) for MMSAs. Adults aged 18-64 years with health care coverage: 86.8% (72.0%-93.8%) for states and territories and 86.8% (63.2%-95.7%) for MMSAs. Adults who received a routine physical checkup during the preceding 12 months: 69.0% (58.1%-79.8%) for states and territories and 69.4% (57.1%-81.1%) for MMSAs. Adults who ever had their blood cholesterol checked: 79.1% (73.3%-86.7%) for states and territories and 79.5% (65.1%-87.3%) for MMSAs. Current cigarette smoking among adults: 17.7% (9.0%-27.2%) for states and territories and 17.3% (4.5%-29.5%) for MMSAs. Binge drinking among adults during the preceding 30 days: 17.2% (11.2%-26.0%) for states and territories and 17.4% (5.5%-24.5%) for MMSAs. Adults who reported no leisure-time physical activity during the preceding month: 25.5% (17.6%-47.1%) for states and territories and 24.5% (16.1%-47.3%) for MMSAs. Adults who reported consuming fruit less than once per day during the preceding month: 40.5% (33.3%-55.5%) for states and territories and 40.3% (30.1%-57.3%) for MMSAs. Adults who reported consuming vegetables less than once per day during the preceding month: 22.4% (16.6%-31.3%) for states and territories and 22.3% (13.6%-32.0%) for MMSAs. Adults who have obesity: 29.5% (19.9%-36.0%) for states and territories and 28.5% (17.8%-41.6%) for MMSAs. Adults aged ≥45 years with diagnosed diabetes: 15.9% (11.2%-26.8%) for states and territories and 15.7% (10.5%-27.6%) for MMSAs. Adults aged ≥18 years with a form of arthritis: 22.7% (17.2%-33.6%) for states and territories and 23.2% (12.3%-33.9%) for MMSAs. Adults having had a depressive disorder: 19.0% (9.6%-27.0%) for states and territories and 19.2% (9.9%-27.2%) for MMSAs. Adults with high blood pressure: 29.1% (24.2%-39.9%) for states and territories and 29.0% (19.7%-41.0%) for MMSAs. Adults with high blood cholesterol: 31.8% (27.1%-37.3%) for states and territories and 31.4% (23.2%-42.0%) for MMSAs. Adults aged ≥45 years who have had coronary heart disease: 10.3% (7.2%-16.8%) for states and territories and 10.1% (4.7%-17.8%) for MMSAs. Adults aged ≥45 years who have had a stroke: 4.9% (2.5%-7.5%) for states and territories and 4.7% (2.1%-8.4%) for MMSAs.

INTERPRETATION

The prevalence of health care access and use, health-risk behaviors, and chronic health conditions varied by state, territory, and MMSA. The data in this report underline the importance of continuing to monitor chronic diseases, health-risk behaviors, and access to and use of health care in order to assist in the planning and evaluation of public health programs and policies at the state, territory, and MMSA level.

PUBLIC HEALTH ACTION

State and local health departments and agencies and others interested in health and health care can continue to use BRFSS data to identify groups with or at high risk for chronic conditions, unhealthy behaviors, and limited health care access and use. BRFSS data also can be used to help design, implement, monitor, and evaluate health-related programs and policies.

The association between long work hours and leisure-time physical activity and obesity

Obesity affects approximately one-third of all U.S. adults, presenting a large economic and public health burden. Long work hours may be contributing to the rising obesity problem by reducing time for physical activity, particularly for individuals working in sedentary occupations. This study sought to investigate the association between long work hours, leisure-time physical activity (LTPA), and obesity across levels of occupational activity in order to identify potentially vulnerable groups. Cross sectional analysis was performed in 2017 using data from the 2015 Georgia Behavioral Risk Factor Surveillance System and prevalence ratios were estimated across work hour and occupational activity groups. Ability to meet guidelines for LTPA did not differ significantly across work hour categories overall. Those working in low activity occupations were more likely to meet aerobic guidelines for LTPA compared to those in intermediate and high activity occupations (χ2: 19.3; P-value: <0.01). Results of interaction assessment demonstrate that the effects of work hours on obesity risk and meeting aerobic guidelines are significantly different across OA categories, indicating OA to be an effect modifier of the relationship between long work hours and obesity (χ2: 13.33; P-value: <0.001; χ2: 4.42; P-value: <0.05). Employees in intermediate activity occupations working long hours were found to be at the greatest risk for obesity. Further research is required to better understand the mechanisms impacting the relationship between long work hours, domains of physical activity, and obesity risk as well as to identify effective intervention and prevention programs for employees in intermediate activity occupations.

Epidemiology of areca (betel) nut use in the mariana islands: Findings from the University of Guam/University of Hawai`i cancer center partnership program

BACKGROUND

Areca (betel) nut is considered a Group 1 human carcinogen shown to be associated with other chronic diseases in addition to cancer. This paper describes the areca (betel) nut chewing trend in Guam, and health behaviors of chewers in Guam and Saipan.

METHODS

The areca (betel) nut module in the Guam Behavioral Risk Factor Surveillance Survey was used to calculate the 5-year (2011-2015) chewing trend. To assess the association between areca (betel) nut chewing and health risks in the Mariana Islands, a cross-section of 300 chewers, ≥18years old, were recruited from households in Guam and Saipan. Self-reported socio-demographics, oral health behaviors, chronic disease status, diet, and physical activity were collected. Anthropometry was measured. Only areca (betel) nut-specific and demographic information were collected from youth chewers in the household.

RESULTS

The 5-year areca (betel) nut chewing prevalence in Guam was 11% and increased among Non-Chamorros, primarily other Micronesians, from 2011 (7%) to 2015 (13%). In the household survey, most adult chewers (46%) preferred areca nut with betel leaf, slaked lime, and tobacco. Most youth chewers (48%) preferred areca nut only. Common adult chronic conditions included diabetes (14%), hypertension (26%), and obesity (58%).

CONCLUSION

The 5-year areca (betel) nut chewing prevalence in Guam is comparable to the world estimate (10-20%), though rising among Non-Chamorros. Adult and youth chewers may be at an increased risk for oral cancer. Adult chewers have an increased risk of other chronic health conditions. Cancer prevention and intervention strategies should incorporate all aspects of health.

Comparing cancer screening estimates: Behavioral Risk Factor Surveillance System and National Health Interview Survey

Cancer screening prevalence from the Behavioral Risk Factor Surveillance System (BRFSS), designed to provide state-level estimates, and the National Health Interview Survey (NHIS), designed to provide national estimates, are used to measure progress in cancer control. A detailed description of the extent to which recent cancer screening estimates vary by key demographic characteristics has not been previously described. We examined national prevalence estimates for recommended breast, cervical, and colorectal cancer screening using data from the 2012 and 2014 BRFSS and the 2010 and 2013 NHIS. Treating the NHIS estimates as the reference, direct differences (DD) were calculated by subtracting NHIS estimates from BRFSS estimates. Relative differences were computed by dividing the DD by the NHIS estimates. Two-sample t-tests (2-tails), were performed to test for statistically significant differences. BRFSS screening estimates were higher than those from NHIS for breast (78.4% versus 72.5%; DD=5.9%, p<0.0001); colorectal (65.5% versus 57.6%; DD=7.9%, p<0.0001); and cervical (83.4% versus 81.8%; DD=1.6%, p<0.0001) cancers. DDs were generally higher in racial/ethnic minorities than whites, in the least educated than most educated persons, and in uninsured than insured persons. For example, the colorectal cancer screening DD for whites was 7.3% compared to ≥8.9% for blacks and Hispanics. Despite higher prevalence estimates in BRFSS compared to NHIS, each survey has a unique and important role in providing information to track cancer screening utilization among various populations. Awareness of these differences and their potential causes is important when comparing the surveys and determining the best application for each data source.

Population characteristics associated with pharmacy-based influenza vaccination in United States survey data

OBJECTIVES

To examine the population characteristics associated with the health behavior of receiving an influenza vaccine from a pharmacy-based setting.

DESIGN

Secondary analysis of data from states that participated in an optional influenza module in the 2014 Behavioral Risk Factor Surveillance System, a state-based observational survey of U.S. adults.

SETTING AND PARTICIPANTS

Analytic sample of 28,954 respondents from 8 states and Puerto Rico who reported receiving an influenza vaccination in the past year.

MAIN OUTCOME MEASURES

The main outcome was a self-reported categoric variable indicating the setting of the most recent seasonal influenza vaccination: doctor's office, pharmacy-based store, or other setting.

RESULTS

Multinomial logistic regression results showed that environmental, predisposing, enabling, and need factors in the Andersen model were salient features associated with odds of using pharmacy-based influenza vaccination settings instead of a doctor's office. Residents of states that allowed pharmacists as immunizers before 1999 reported greater use of pharmacy-based store settings (odds ratio [OR] 1.31). Compared with young adults, individuals 65 years of age and older were more likely to choose a pharmacy-based store than a doctor's office (OR 1.41) and less likely to use other community settings (OR 0.45). Compared with non-Hispanic whites, black respondents were less likely to use pharmacy-based store vaccination (OR 0.51), and multiracial and Hispanic respondents were more likely to use other settings (ORs 1.47 and 1.60, respectively). Enabling and need factors were also associated with setting.

CONCLUSION

Based on this dataset of selected states from 2014, almost one-fourth of U.S. adults who reported receiving an annual influenza vaccination did so from a pharmacy-based store; 35% reported using other community-based settings that may enlist pharmacists as immunizers. There were striking disparities in use of nontraditional vaccination settings by age and race or ethnicity. Pharmacists and pharmacies should address missed opportunities for vaccination by targeting outreach efforts based on environmental and predisposing characteristics.

Mobile Phone Surveys for Collecting Population-Level Estimates in Low- and Middle-Income Countries: A Literature Review

Background

National and subnational level surveys are important for monitoring disease burden, prioritizing resource allocation, and evaluating public health policies. As mobile phone access and ownership become more common globally, mobile phone surveys (MPSs) offer an opportunity to supplement traditional public health household surveys.

Objective

The objective of this study was to systematically review the current landscape of MPSs to collect population-level estimates in low- and middle-income countries (LMICs).

Methods

Primary and gray literature from 7 online databases were systematically searched for studies that deployed MPSs to collect population-level estimates. Titles and abstracts were screened on primary inclusion and exclusion criteria by two research assistants. Articles that met primary screening requirements were read in full and screened for secondary eligibility criteria. Articles included in review were grouped into the following three categories by their survey modality: (1) interactive voice response (IVR), (2) short message service (SMS), and (3) human operator or computer-assisted telephone interviews (CATI). Data were abstracted by two research assistants. The conduct and reporting of the review conformed to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.

Results

A total of 6625 articles were identified through the literature review. Overall, 11 articles were identified that contained 19 MPS (CATI, IVR, or SMS) surveys to collect population-level estimates across a range of topics. MPSs were used in Latin America (n=8), the Middle East (n=1), South Asia (n=2), and sub-Saharan Africa (n=8). Nine articles presented results for 10 CATI surveys (10/19, 53%). Two articles discussed the findings of 6 IVR surveys (6/19, 32%). Three SMS surveys were identified from 2 articles (3/19, 16%). Approximately 63% (12/19) of MPS were delivered to mobile phone numbers collected from previously administered household surveys. The majority of MPS (11/19, 58%) were panel surveys where a cohort of participants, who often were provided a mobile phone upon a face-to-face enrollment, were surveyed multiple times.

Conclusions

Very few reports of population-level MPS were identified. Of the MPS that were identified, the majority of surveys were conducted using CATI. Due to the limited number of identified IVR and SMS surveys, the relative advantages and disadvantages among the three survey modalities cannot be adequately assessed. The majority of MPS were sent to mobile phone numbers that were collected from a previously administered household survey. There is limited evidence on whether a random digit dialing (RDD) approach or a simple random sample of mobile network provided list of numbers can produce a population representative survey.