Introduction: Measuring consumption of fruit and vegetables (FV) is both challenging and critical to the evaluation of community-based interventions, such as FRESH Austin. This study presents the development and validation of the FRESH Austin FFQ, compared to repeated 24 hour Dietary Recalls (24hDR).

Methods: FV consumption and demographic information of fifty-six residents in low-income, ethnically diverse communities were collected using the FV FFQ and three 24hDRs over thirty days. Summary statistics, correlations, Mean Absolute Errors (MAE) and Bland-Altman plots were used to characterize associations and agreement among food categories.

Results: Study participants were diverse (22% Hispanic, 5.5% Black, 54% White, 19% Other), and 24% reported less than $25K annual income. The FFQ and 24hDR produced similar estimates of average total FV servings per day (6.68 and 6.40 servings per day, respectively.) Spearman or Pearson correlations produced measures from = 0.01 for “Potatoes” and r = 0.59 for “Other Vegetables”, indicating that the FRESH Austin FFQ provides moderate correlation for most, but not all, FV surveyed. MAE values were small for all FV, suggesting the variance of the error estimates were also small. Bland-Altman plots indicate acceptable levels of agreement between the two methods.

Conclusion: The development and validation of the FRESH Austin FV FFQ compared to repeated 24hDRs may provide important insights for evaluating and improving community-based efforts to increase FV consumption in diverse populations.

Background. Timely and representative data is essential to reduce health disparities, yet data are outdated and minorities are underrepresented in research and surveys. Objective. This study was conducted to determine if a nonprobability geographically targeted social media ad (SMAs) method, could provide better representation of certain groups than address-based sampling (ABS), for a rapid approach to health data collection. Methods. Focus groups (n=21) were conducted to develop the SMAs. SMAs and traditional ABS were used to recruit participants from 1.9 million residents in Santa Clara County, California, to download the HealthKey mobile app and complete two mobile surveys over a few weeks. Demographics were compared to the census-based American Community Survey and costs for recruitment were calculated. Results. Of the 496 mobile app downloads, 363 participants were SMA recruits compared to 133 for ABS. In the combined sample, the survey participants were representative of the population by race and ethnicity. However, ages 50+ and men were underrepresented in the SMA sample compared to ABS and the Census data. The SMA method was three-fold less expensive compared to ABS costs. Conclusions. Given persistent health disparities and a need for cost effective measures to collect real-time data, SMA recruitment with smartphone data collection may be a path to improving representation in research and could be an important methodology to shrink health disparities, particularly for efforts that rely on real time data. Further investigation is warranted as many disciplines rely on survey data for health care resource allocation and policy decisions.

Background: Since disabled individuals experience health disparities, national health surveys, such as the Behavioral Risk Factor Surveillance System (BRFSS), are attempting to capture disability status. However, measuring disability is complex due to the range of biological manifestations.

Objective: This study aimed to compare agreement between two BRFSS measures of disability and a previously used proxy measure of disability, Medicare enrollment before the age of 65.

Methods: This secondary data analysis used 2016 BRFSS data to compare the two-item and six-item disability measures assessed across 19 states in 2016. Further, these measures were compared to the Medicare-proxy for disability collected in three states that year. Agreement between disability measures was assessed using percent agreement and kappa coefficients. Statistical analysis were performed in SAS 9.4.

Results: Among the 2016 BRFSS respondents with information for both disability measures (n=221,520), most were white (76.2%), and aged less than 65 years (63.4%). Almost one-third reported a disability with the six-item measure (30.5%), while fewer were classified as disabled with the two-items (28.9%). Agreement was moderate between the six- and two-item measures overall (81.5% agreement and k=0.559); however the mobility question agreed more with the two-item measure than the other five questions. Additionally, agreement was lower between the Medicare proxy and both other measures (78.7% agreement, k=0.352 and 78.7% agreement, k=0.344 for the six-item and two-item measures, respectively).

Conclusions: The six-item disability measure identified more individuals with disabilities than the two-item measure. Medicare coverage before the age of 65 may be inadequate proxy measure for disability.

Background

Tuberculosis (TB) remains a public health concern in the Philippines, with 26,000 recorded deaths in 2016. The Philippines monitors its TB situation through annual incidence rates, measured via new TB cases per 100,000 population. However, using incidence rates tends to understate the magnitude of the TB problem and obscures performance of TB control efforts. Moreover, using TB incidence in strategy development may overlook considerations on managing cases who remain infectious.

Objectives

The study aims to describe the limitations of incidence-based TB burden monitoring using a TB transitions model.

Methods

This study employs a TB transitions model that describes the distribution of TB cases to show the percentage of those who would be cured, dead, and those that would remain infective. The model uses TB prevalence and incidence data, and figures on cured TB cases in health facilities, TB mortality, and treatment-seeking behavior.

Results

The model estimates the drug-sensitive TB burden to be around 900,000 individuals, with almost 400,000 being prevalent TB cases uncured from previous years. This estimated figure is significantly higher than the WHO reported 2018 TB incidence of 591,000. This nuance is critical since medium-term targets under the Strategic Plan for TB Elimination is based on incidence rates.

Conclusion

This study recommends the use of a transitions model to estimate the actual TB burden using prevalence data and other parameters. A more accurate estimate of TB burden will help policy makers in planning for TB interventions and would result in greater coverage and improved outcomes for TB.