Background:

Project IMPACT, a CDC-funded 5-year modified stepped wedge quasi-experimental study, was launched to test the impact of integrating an electronic health record (EHR)-based and community health worker-led intervention to improve hypertension control among patients in small NYC practices with a large proportion of South Asian patients. Results of the EHR intervention at 12 months post-intervention implementation suggested a significant positive association to improve average BP and hypertension control. Literature on long-term outcomes of HIT interventions are sparse. This presentation responds to that gap by testing the effect of this intervention on BP outcomes 18-months after implementation.

Methods:

Aggregate practice-level EHR data were extracted every six months. The primary practice-level outcome was the change in proportion of hypertensive patients with controlled BP among all patients with hypertension. Patient-level EHR and claims data were extracted for patients with Healthfirst (HF) insurance and had at least one diagnosis of hypertension at a clinic visit in 2015. The main individual-level outcome for this study was average systolic BP and diastolic BP at last clinic visit. A secondary individual-level outcome was proportion of patients with controlled BP at last clinic visit, among all HF hypertensive patients.

Results:

At the practice level, there was a significant improvement in hypertension control through 18 months post-intervention implementation, adjusting for age and sex (aRR: 1.15, 95% CI: 1.07-1.25). Among HF patients, we observed a significant association on average SBP and DBP adjusting for time, age, and sex, with a decrease of 1.22 mmHg and 1.40 mmHg, respectively. The proportion of HF patients with well-controlled BP improved after adjusting for demographics (aOR: 1.25, 95% CI: 1.00-1.57).

Conclusion:

Early results suggest that practice-level outcome improvements are sustainable through 18-months post-intervention implementation. Future results to be presented include adherence of hypertension medication and appointments, and BP control through 24-months post-intervention implementation.

Background: The internet is an important source for health information, but online information quality varies substantially. Variations in ability to make sense of information may contribute to negative psychosocial outcomes and increase digital divide-related health disparities. Advances in health information technologies can increase access to high-quality health information and eliminate such disparities. The aim of this study was to understand how people trust and use the information they obtain from online sources by linking objective measures of information-seeking from an online artificial intelligence-powered system (Buoyhealth.com) with clinical/psychosocial outcomes.

Methods: 194 randomly-sampled users completed a Reasoned-Action-based survey immediately after their Buoy session and 2 weeks later. OLS regression models examined factors affecting confidence in recommendations and intentions to follow these.

Results: Buoy users were well-educated (86% college-educated), mostly white (77%) and female (83%). Most had insurance (87%), regular healthcare provider (78%), and reported good/very good/excellent health (82%). The most common symptoms reported were bloody vaginal discharge, pain (lower back/abdominal), cough, appetite loss. The most common recommendations were to seek primary care within 2 weeks (37%) or to self-treat (27%). Users found Buoy useful, easy-to-understand, and reported confidence in the recommendations, Respondents had strong intentions to follow Buoy’s recommendations, but were less inclined to discuss Buoy use with their doctors. Confidence (Β=0.46,p<.001), perceived-utility (Β=0.39,p<.001) and anxiety-reduction because of using Buoy (Β=0.39,p<.01) were associated with stronger intentions to follow recommendations. Those advised to wait/watch or self-treat had strongest intentions to comply, whereas those advised to seek care had weaker intentions (Β=-0.25,p<.001).

Conclusions: Results demonstrate the potential utility of an artificial-intelligence-powered online health information tool to improve psychological outcomes; less clear is whether this translates into health behaviors/outcomes. Analyses of the user base illustrate persistent second-level digital-divide effects.

Care coordination is a critical approach to improve the delivery of dental services and prevent disease. Particularly in the case of children, proper coordination across systems of care, health care providers, and caregivers is important to keep children healthy and pain free throughout their childhood. Utilization of Health Information Technology (HIT) is a key strategy to enable timely communication of care needs, yet clinical providers are at a disadvantage due to the lack of technology available to securely communicate between medical and dental settings.

Strengthening the connections between medical and dental settings is a necessary and important step to improving pediatric oral health. To overcome the communication gap between medical and dental clinicians, our team developed SmileConnect® Clinical. This dental registry aims to improve the timeliness of referrals to a dentist, and puts in place a way to monitor access to oral health services.

Classified as a public health tool, SmileConnect Clinical provides a complete oral health profile for a patient. SmileConnect Clinical can integrate with any patient care system that supports open standards for authentication (SAML 2. 0, OAUTH2), and any dental provider directory service that can support a SOAP or REST based protocol. Technology that creates connectivity between care settings is constantly evolving, and certain Electronic Health Record (EHR) vendors are breaking ground by deploying Fast Healthcare Interoperability Resources (FHIR) enabled EHR versions into production. SmileConnect Clinical’s architect is both scalable and extensible.

Preliminary impact data of SmileConnect Clinical in Michigan shows that Medicaid children who receive a dental referral through SmileConnect Clinical (rather than receiving a paper or verbal referral) are accessing preventive dental services at a much more rapid rate than children not referred through SmileConnect Clinical.

Introduction. Examining geographic patterns and hotspots for pediatric injury hospital admissions can provide insight into potential injury prevention strategies. We identified pediatric injury factors associated with hotspots with high concentrations of injury admissions.

Methods. A retrospective analysis of a pediatric trauma registry of patients (n=3,714) admitted between 2010-2017 was conducted using geographic information systems (GIS) to map where injuries were occurring in relation to the pediatric trauma center and to identify clustering of injuries in hotspots. Factors independently associated with injury hotspots were identified using forward stepwise logistic regression modeling.

Results. Children injured in a geographic hotspot were more likely to be non-white (Adjusted Odds Ratio (AOR)=2.70; 95% Confidence Interval (CI): 2.30, 3.17), age 12 years or older (AOR= 1.58; 95% CI: 1.34, 1.87), and have government insurance (AOR=1.77; 95% CI: 1.50, 2.10). These children also had more severe injuries and were thus less likely to be discharged home (AOR= 0.63, 95% CI: 0.48, 0.84). The odds of being admitted for a gunshot wound (AOR=2.27, 95% CI: 1.42, 3.61) or a motorcycle injury (AOR=2.40; 95% CI: 1.47, 3.93) were more than twice as high among children injured in a hotspot compared to non-hotspot. Of the motorcycle injuries, 65% were attributed to motocross motorcycles, which do not require a driver’s license. Hotspot zip codes had overall lower socioeconomic status than the non-hotspots, with average median household income of $40,072 in hotspots compared to $50,191 in non-hotspots (p=0.04).

Conclusions. Pediatric gun violence injuries appear to be clustered in hotspots and areas of lower socioeconomic status, and thus, may be effective targets for gun safety and anti-violence campaigns. These areas also have a clustering of motorcycle injuries, which can be reduced in the future through safety programs focused on motocross since most of the motorcycle injuries occurred while participating in this sport.

Background/Purpose

The Popular Opinion Leader (POL) model developed by the Centers of Disease Control trains community leaders to increase health-related awareness in their social networks. An academic-community partnership in Chicago and Boston used the POL model to increase lupus knowledge and promote care-seeking behaviors among African Americans. Understanding that place influences health, it was important to assess sociodemographic characteristics of Chicago communities where the POLs disseminated lupus-related awareness information.

Methods

Ten POLs participated in a four-session training program. All POLs were African American females and the mean age was 59. Over 6 months, the POLs tracked their social network encounters by recording addresses where information was disseminated and number of contacts engaged. Data were entered into the Geographic Information System (GIS) to evaluate network reach. We utilized the Healthy Chicago Database to describe sociodemographic factors and health resources of neighborhoods identified by the POL’s social networks.

Results

The 10 POLs’ social networks were concentrated in four Chicago communities (Washington Heights, Morgan Park, Englewood and Roseland) in a 6-month time frame. In these Chicago South side neighborhoods health, educational, and financial resources are limited. All four communities were predominantly African American, range 54.6%-96.1%. Individuals with no health insurance ranged from 8.4%-17.2% in Morgan Park and Englewood, and from 10.5%-12.4% in Washington Heights and Roseland. Individual poverty was highest in Englewood (30.8%) and Roseland (27.2%) compared with 18% in the other two neighborhoods. The percentage of individuals not completing high school was highest in Englewood (19.7%) and Roseland (13.5%) compared with 10% in the other two areas.

Discussion/Conclusions

Engagement of community members through POL conversations tapped into neighborhoods that demonstrated significant individual poverty, lack of health insurance, and reduced high school education. The POLs targeted resource-limited areas as demonstrated by GIS for lupus awareness efforts. Next steps include incorporating Healthy Chicago Database knowledge to demonstrate how the POLs can sustain educational engagement considering local sociodemographic challenges.

Background: The prevalence of mental illness is increasing, and various treatment options exist. Individuals with mental health needs and illnesses report benefit from peer-to-peer interactions. Peer support is a unique categorization of support which involves a group recovery via shared experiences. With the advent of consumers undertaking the role of shared decision makers in their own health, there is a significant increase in usage of mobile and electronic technologies. Current literature is focused largely on recommendations, future projections, content analysis of user postings, and/or outcomes which validate the use of peer support. Some gaps in existing literature portray themselves in the lack of measures for meaningful data, and furthermore in the lack of existing methodologies to target, exclusively, the mental health needs of populations whilst utilizing peer-to-peer support.

Objectives: To develop a website/web-based portal for individuals with mental health needs, where peer support and peer learning are primary means of intervention.

Methods: We identified focus areas for the website/web-based portal that include resources and peer-to-peer support channels. The web portal is to be developed with HTML and JavaScript alongside other scripting languages. A key area of focus is on the peer-to-peer support intervention, which would take place in the form of channels or threads where individuals can anonymously connect and share experiences. Peer learning is another aspect present in this peer-to-peer support intervention as there is an absence of an implied authority or caregiver. Furthermore, the second area of focus is on resources, which include various scoring methodologies alongside crisis information.

Discussion: Outcome measures include self-reported depression screening scores, distress screening scores, self-reported progress, and levels of user engagement.

Conclusions: Technology can be especially beneficial for individuals with mental health needs and based on an evaluation of the effects of peer-to-peer support, peer-to-peer support is an important intervention which needs to be explored.

Understanding community health needs and preparing for evolving public health threats are complex challenges for any health organization. While the scope of these challenges varies across organizations and geographies; similar tool-sets and capabilities are needed. As both the number and intensity of emergency events increase, organizations are increasingly looking at updating and enhancing their existing plans and capabilities. This is necessary to not only safeguard public health during disasters, but to ensure resources are being allocated efficiently under blue skies and that both regulatory and grant requirements are being met. To address these challenges, organizations need to integrate massive amounts of disparate, and often siloed datasets into one comprehensive view that is accessible by a variety of stakeholders with different skill sets, responsibilities and access levels. Location is often the only common feature in these data-sets and Geographic Information System (GIS) technology provides a framework to bring this data together and transform it into actionable information for analysis, visualization and reporting purposes. This session will help attendees understand how to leverage modern GIS tools to aid in all aspects of public health preparedness planning and response while simultaneously enhancing routine operations with the same set of tools. The speaker will provide insight into available tools to rapidly assess community health needs, improve resiliency, enable a mobile workforce and create a common operating picture that provides real-time insight into rapidly unfolding events. Background, context and real-world case-study examples will be reinforced with demonstrations that address the six key preparedness domains identified by the CDC. The speaker will explain how to collect, access, map and analyze different types of data in real-time through easily configured applications. Additional topics covered will include how to easily share the resulting data, maps and applications via the web and mobile devices to promote cross agency collaboration and citizen engagement.

Background

Carissa carandas L., an herb commonly found in Thailand, has been claimed to have anti-diabetic property potentially from alpha-glucosidase inhibition. While a clinical experiment in diabetic patients is not readily possible, in addition to exploring whether individuals who are in proximity to Carissa carandas L. plantation are less likely to become diabetic, this study shares an experience in collecting spatial data by village health volunteers (VHVs).

Methods

Demographics and spatial data of diabetic patients in Samut Songkhram province were retrieved and analyzed. Spatial data of growers and number of Carissa carandas planted by each grower were efficiently surveyed by VHVs who were familiar with the community. A plantation area was defined as having at least one Carissa carandas L. aged at least one year. Spatial correlation analysis was performed.

Results

Of 172,648 population (66,902 households) in Samut Songkhram province, 10,706 diabetic patients had spatial data retrieved whereas 597 out of 944 Carissa carandas growers identified from the survey had spatial data available for the analysis. Spatial autocorrelation analysis revealed that the Carissa carandas plantation locations were not significantly different than random (Moran’s Index -0.053; p=0.7745) whereas the diabetic patients residences were significantly different than random (Moran’s Index 0.3659; p<0.001).

Conclusion

Village health volunteers could assist in collecting spatial data, which is useful for a preliminary assessment of a claimed diabetic prevention property of Carissa carandas L.

Background: Despite the potential value of data visualization for improving understanding of data in a manner that results in better decision-making in health and healthcare, there is little understanding of its applications in public health practice, particularly for public health program planning and resource allocation.

Objectives: This study examined how data are understood differently when displayed in a table format versus visualized format by public health practitioners. This study also examined their confidence in understanding, and perceived ease of use with table and visualization presentations.

Methods: An exploratory mixed-methods design, using online surveys was used. The study developed an online dashboard including a set of a tables and an interactive visualization for use in an online survey. The online survey included scenarios and related questions asking participants to explore the dashboard by comparing their agency with others within and across states.

Results: The survey included 22 public health practitioners. Content analysis was used in coding qualitative data from open-ended questions and descriptive analyses with short answer questions. Analyses revealed comparisons that participants made through looking at data in a table and a visualization. Public health practitioners made similar types of comparisons with both the table and visualization data format, such as comparing to the mean, median, min, or max; comparing by population size categories; or between subcategories. Data visualization has apparent benefits in advancing the understanding of data among public health practitioners. Data visualization makes it seemingly easy for participants to find information from data, reduce errors in assessing information, as well as process data and help generate more meaningful knowledge.

Discussion/Conclusions: Findings suggest that data visualizations could be useful in advancing understanding of data for local public health leaders, and in filling a gap in understanding how to improve their use of data and data visualization for decision-making.