Search

BackgroundMultiple chronic conditions (MCCs) are common among older adults and expensive to manage. Two-thirds of Medicare beneficiaries have multiple conditions (eg, diabetes and osteoarthritis) and account for more than 90% of Medicare spending. Patients with MCCs also experience lower quality of life and worse medical and psychiatric outcomes than patients without MCCs. In primary care settings, where MCCs are generally treated, care often focuses on laboratory results and medication management, and not quality of life, due in part to time constraints. eHealth systems, which have been shown to improve multiple outcomes, may be able to fill the gap, supplementing primary care and improving these patients’ lives. ObjectiveThis study aims to assess the effects of ElderTree (ET), an eHealth intervention for older adults with MCCs, on quality of life and related measures. MethodsIn this unblinded study, 346 adults aged 65 years and older with at least 3 of 5 targeted high-risk chronic conditions (hypertension, hyperlipidemia, diabetes, osteoarthritis, and BMI ≥30 kg/m2) were recruited from primary care clinics and randomized in a ratio of 1:1 to one of 2 conditions: usual care (UC) plus laptop computer, internet service, and ET or a control consisting of UC plus laptop and internet but no ET. Patients with ET have access for 12 months and will be followed up for an additional 6 months, for a total of 18 months. The primary outcomes of this study are the differences between the 2 groups with regard to measures of quality of life, psychological well-being, and loneliness. The secondary outcomes are between-group differences in laboratory scores, falls, symptom distress, medication adherence, and crisis and long-term health care use. We will also examine the mediators and moderators of the effects of ET. At baseline and months 6, 12, and 18, patients complete written surveys comprising validated scales selected for good psychometric properties with similar populations; laboratory data are collected from eHealth records; health care use and chronic conditions are collected from health records and patient surveys; and ET use data are collected continuously in system logs. We will use general linear models and linear mixed models to evaluate primary and secondary outcomes over time, with treatment condition as a between-subjects factor. Separate analyses will be conducted for outcomes that are noncontinuous or not correlated with other outcomes. ResultsRecruitment was conducted from January 2018 to December 2019, and 346 participants were recruited. The intervention period will end in June 2021. ConclusionsWith self-management and motivational strategies, health tracking, educational tools, and peer community and support, ET may help improve outcomes for patients coping with ongoing, complex MCCs. In addition, it may relieve some stress on the primary care system, with potential cost implications. Trial RegistrationClinicalTrials.gov NCT03387735; https://www.clinicaltrials.gov/ct2/show/NCT03387735. International Registered Report Identifier (IRRID)DERR1-10.2196/25175

BackgroundDepressive symptoms are the most prevalent mental health concern among older adults (possibly heightened during the COVID-19 pandemic), which raises questions about how such symptoms can be lowered in this population. Existing research shows that offline social connectedness is a protective factor against depression in older adults; however, it is unknown whether web-based social connectedness can have similar effects. ObjectiveThis study investigates whether social connectedness on a support website protects older adults against depressive symptoms over the course of a year, above and beyond the protective effect of offline social connectedness. The secondary aim is to determine whether older adults with increased depressive symptoms are more likely to engage in social connectedness on this website. Thus, we examine depressive symptoms as both an outcome and predictor of web-based social connectedness to fully understand the chain of causality among these variables. Finally, we compare web-based social connectedness with offline social connectedness in their ability to lower depressive symptoms among older adults. MethodsA total of 197 adults aged 65 years or older were given access to a social support website, where they were able to communicate with each other via a discussion forum for a year. Participants’ social connectedness on the web-based platform, conceptualized as message production and consumption, was measured using behavioral log data as the number of messages participants wrote and read, respectively, during the first 6 months (t1) and the following 6 months (t2) of the study. Participants self-reported their offline social connectedness as the number of people in their support networks, and they reported their depressive symptoms using the Patient Health Questionnaire-8 both at baseline (t1) and at 12-month follow-up (t2). To ascertain the flow of causality between these variables, we employed a cross-lagged panel design, in which all variables were measured at t1 and t2. ResultsAfter controlling for the effect of offline support networks at t1, web-based message consumption at t1 decreased older adults’ depressive symptoms at t2 (β=−.11; P=.02), but web-based message production at t1 did not impact t2 depressive symptoms (β=.12; P=.34). Web-based message consumption had a larger effect (β=−.11; P=.02) than offline support networks (β=−.08; P=.03) in reducing older adults’ depressive symptoms over time. Higher baseline depressive symptoms did not predict increased web-based message consumption (β=.12; P=.36) or production (β=.02; P=.43) over time. ConclusionsThe more messages older adults read on the web-based forum for the first 6 months of the study, the less depressed they felt at the 1-year follow-up, above and beyond the availability of offline support networks at baseline. This pinpoints the substantial potential of web-based communication to combat depressive symptoms in this vulnerable population. International Registered Report Identifier (IRRID)RR2-10.1186/s13063-015-0713-2

Human groups are able to converge to more accurate beliefs through deliberation, even in the presence of polarization and partisan bias --- a phenomenon known as the ``wisdom of partisan crowds.'' Large Language Models (LLMs) are increasingly being used to simulate human collective behavior, yet few benchmarks exist for evaluating their dynamics against the behavior of human groups. In this paper, we examine the extent to which the wisdom of partisan crowds emerges in groups of LLM-based agents that are prompted to role-play as partisan personas (e.g., Democrat or Republican). We find that they not only display human-like partisan biases, but also converge to more accurate beliefs through deliberation, as humans do. We then identify several factors that interfere with convergence, including the use of chain-of-thought prompting and lack of details in personas. Conversely, fine-tuning on human data appears to enhance convergence. These findings show the potential and limitations of LLM-based agents as a model of human collective intelligence.

Accurately simulating human opinion dynamics is crucial for understanding a variety of societal phenomena, including polarization and the spread of misinformation. However, the agent-based models (ABMs) commonly used for such simulations often over-simplify human behavior. We propose a new approach to simulating opinion dynamics based on populations of Large Language Models (LLMs). Our findings reveal a strong inherent bias in LLM agents towards producing accurate information, leading simulated agents to consensus in line with scientific reality. This bias limits their utility for understanding resistance to consensus views on issues like climate change. After inducing confirmation bias through prompt engineering, however, we observed opinion fragmentation in line with existing agent-based modeling and opinion dynamics research. These insights highlight the promise and limitations of LLM agents in this domain and suggest a path forward: refining LLMs with real-world discourse to better simulate the evolution of human beliefs.

BackgroundeHealth interventions appear and change so quickly that they challenge the way we conduct research. By the time a randomized trial of a new intervention is published, technological improvements and clinical discoveries may make the intervention dated and unappealing. This and the spate of health-related apps and websites may lead consumers, patients, and caregivers to use interventions that lack evidence of efficacy. ObjectiveThis paper aims to offer strategies for increasing the speed and usefulness of eHealth research. MethodsThe paper describes two types of strategies based on the authors’ own research and the research literature: those that improve the efficiency of eHealth research, and those that improve its quality. ResultsEfficiency strategies include: (1) think small: conduct small studies that can target discrete but significant questions and thereby speed knowledge acquisition; (2) use efficient designs: use such methods as fractional-factorial and quasi-experimental designs and surrogate endpoints, and experimentally modify and evaluate interventions and delivery systems already in use; (3) study universals: focus on timeless behavioral, psychological, and cognitive principles and systems; (4) anticipate the next big thing: listen to voices outside normal practice and connect different perspectives for new insights; (5) improve information delivery systems: researchers should apply their communications expertise to enhance inter-researcher communication, which could synergistically accelerate progress and capitalize upon the availability of “big data”; and (6) develop models, including mediators and moderators: valid models are remarkably generative, and tests of moderation and mediation should elucidate boundary conditions of effects and treatment mechanisms. Quality strategies include: (1) continuous quality improvement: researchers need to borrow engineering practices such as the continuous enhancement of interventions to incorporate clinical and technological progress; (2) help consumers identify quality: consumers, clinicians, and others all need to easily identify quality, suggesting the need to efficiently and publicly index intervention quality; (3) reduce the costs of care: concern with health care costs can drive intervention adoption and use and lead to novel intervention effects (eg, reduced falls in the elderly); and (4) deeply understand users: a rigorous evaluation of the consumer’s needs is a key starting point for intervention development. ConclusionsThe challenges of distinguishing and distributing scientifically validated interventions are formidable. The strategies described are meant to spur discussion and further thinking, which are important, given the potential of eHealth interventions to help patients and families.

To examine the effects of numerical evidence and message framing in communicating vaccine efficacy information about infectious diseases, an online experiment presented to U.S. adults different versions of a vaccination promotional message that vary by numerical vaccine efficacy evidence: (low efficacy rate: 60% vs. high efficacy rate: 95%), outcome framing (preventing disease-related infection vs. preventing disease-related severe illness), and gain vs. loss framing, using a factorial between-subjects design. While there was no significant interaction between numerical vaccine efficacy evidence and message framing, findings showed that a higher vaccine efficacy rate increased positive beliefs about vaccination and outcome framing emphasizing infection prevention increased message processing fluency. Given that infectious diseases pose higher risks for severe illness among older adults, follow-up analyses by age showed that only younger adults were sensitive to message framing where outcome framing emphasizing infection prevention increased processing fluency. [ABSTRACT FROM AUTHOR]

Although low socio-economic status (SES) persons with diabetes show low levels of physical activity (PA), there is limited knowledge regarding which media is effective in improving their PA. This study aimed to determine the appropriate media for providing PA-related information to persons with diabetes in low SES compared with those in high SES. The data of 770 persons with diabetes in low or high SES were extracted from Multimedia Audience Research Systems 2013, a nationwide cross-sectional study. The relationships among media use for health information (i.e. magazine, television and Internet use), PA, and high and low SES were examined using regression analysis. Additional analysis was conducted to examine whether the above relationships differ by age, which influences the use of media. The relationships of media use for health information with PA levels varied by SES; while television use was associated with increased PA levels, Internet use was associated with decreased PA levels in older, low SES persons with diabetes. The findings indicated that television can be a useful resource to provide PA-related information to low SES persons with diabetes and suggest the significance of choosing appropriate media to provide PA-related information for them.

The intersection of consumer culture and civic life has long been a topic of academic discubion. This study revisits the relationship between consumption and civic engagement and investigates the moderators of this relationship. Specifically, we focus on news consumption and opinion leadership as intervening factors that condition the way consumption and civic life are interconnected. Our data reveal that both socially conscious consumption and status-oriented consumption are positively related with civic participation. The positive relationships become stronger when news consumption increases or when one's opinion leadership is strong. Implications for research on consumer culture and civic engagement are discubed.

The 2020 election and its aftermath present an opportunity to understand how audiences' changing news consumption patterns within an expanded, digitized, and polarized media environment shape electoral judgments. This paper introduces measures that capture individuals' partisan slant, partisan extremity, and overall diversity of news media use to understand how people interact with the contemporary news ecology. The measures combine self-reported media use with the partisan slant of news outlets along the ideological spectrum. Using these measures, we analyze a two-wave panel survey conducted before and after the 2020 US election to investigate how slant, diversity, and extremity are related to post-election beliefs, including public confidence in the election and the acceptance of fraud claims. Our findings show that Republicans have more insulated news use behaviors in terms of slant and diversity. The analysis also reveals that the slant of people's news use is associated with post-election fraud beliefs, with right-wing news consumers more likely to believe such claims. However, a diverse news consumption style can moderate misinformation beliefs. Panel analysis points to the role of extreme-right news use in decreasing confidence in the legitimacy of the election. [ABSTRACT FROM AUTHOR]

This study explores the framing strategies and language features of U.S. news coverage surrounding sexual violence and gender issues across the ideological media spectrum at two pivotal phases of the #MeToo movement: (1) during its initial rise as a hashtag-driven social movement in 2017 and (2) during the Kavanaugh nomination and confirmation in 2018. Using structural topic modeling, community detection, and feature extraction, we reveal a heightened employment of political framing during the Kavanaugh accusations. Topical prevalence and language use in news treatment also showed clear partisan differences, consistent with theories of moral foundation and issue ownership. Implications for research on news coverage of gender rights and framing of social movements are discussed. [ABSTRACT FROM AUTHOR]

Extending literature on political polarization and political expression, we study patterns of polarized expression by vocal partisans from opposing camps on social media. Specifically, we argue that polarized partisan expression can be characterized by three divergences: 1) different thematic emphases on the same issue; 2) response to different real-world events on the same issue; and 3) a temporal disconnect at the aggregate level. Highlighting how online expression by different partisan groups is animated by discrete concerns and events and exhibits different temporality, the three divergences in polarized partisan expression not only reflect and explain existing polarization concepts but also speak to the epistemological chasm between partisan groups. Our empirical analysis is based on Twitter discussion about the issue of immigration in the U.S. and applies topic modeling and time series analysis. Results demonstrate that liberal and conservative tweets exhibit different thematic emphases, are often spurred by different event features, and remain largely temporally independent, though both Trump's tweets and emotionally evocative events can draw simultaneous reaction from both sides. These findings suggest that opposing partisan groups not only hold different views on the same issue, but also weave different events and facts about the issue into partisan expression in response to different exogenous factors. In short, they "talk past each other." These polarized partisan expression patterns indicate a splintered public sphere, a concerning quality for deliberative democracy. [ABSTRACT FROM AUTHOR]

Beyond a consciousness-raising hashtag, #MeToo has become a transnational movement, crossing the borders of many societies. However, outsized attention has been paid to the manifestations of #MeToo in the US and on Twitter when the reach of this movement was not restricted to a single country, language, or platform. Drawing from the concept of hybridization, we seek to understand how the uses of #MeToo are contextualized across cultures, languages, and social media platforms. By establishing a macroscopic computational approach, we examine the global diffusion of #MeToo as a hybrid communicative process across different language groups (English, Spanish, and Korean) and social media platforms (Twitter, Facebook, and Instagram). Through time-series analysis and comparative descriptions of language groups and platforms, we demonstrate how discourse flows, language characteristics, and actors differ across cultural and platform contexts and how public discourse of #MeToo was reappropriated and re-signified in different parts of the world to localize connective action. [ABSTRACT FROM AUTHOR]

Medication for opioid use disorder (MOUD) improves treatment retention and reduces illicit opioid use. A-CHESS is an evidence-based smartphone intervention shown to improve addiction-related behaviors. The authors tested the efficacy of MOUD alone versus MOUD plus A-CHESS to determine whether the combination further improved outcomes. In an unblinded parallel-group randomized controlled trial, 414 participants recruited from outpatient programs were assigned in a 1:1 ratio to receive either MOUD alone or MOUD+A-CHESS for 16 months and were followed for an additional 8 months. All participants were on methadone, buprenorphine, or injectable naltrexone. The primary outcome was abstinence from illicit opioid use; secondary outcomes were treatment retention, health services use, other substance use, and quality of life; moderators were MOUD type, gender, withdrawal symptom severity, pain severity, and loneliness. Data sources were surveys comprising multiple validated scales, as well as urine screens, every 4 months. There was no difference in abstinence between participants in the MOUD+A-CHESS and MOUD-alone arms across time (odds ratio=1.10, 95% CI=0.90–1.33). However, abstinence was moderated by withdrawal symptom severity (odds ratio=0.95, 95% CI=0.91–1.00) and MOUD type (odds ratio=0.57, 95% CI=0.34–0.97). Among participants without withdrawal symptoms, abstinence rates were higher over time for those in the MOUD+A-CHESS arm than for those in the MOUD-alone arm (odds ratio=1.30, 95% CI=1.01–1.67). Among participants taking methadone, those in the MOUD+A-CHESS arm were more likely to be abstinent over time (b=0.28, SE=0.09) than those in the MOUD-alone arm (b=0.06, SE=0.08), although the two groups did not differ significantly from each other (∆b=0.22, SE=0.11). MOUD+A-CHESS was also associated with greater meeting attendance (odds ratio=1.25, 95% CI=1.05–1.49) and decreased emergency department and urgent care use (odds ratio=0.88, 95% CI=0.78–0.99). Overall, MOUD+A-CHESS did not improve abstinence relative to MOUD alone. However, MOUD+A-CHESS may provide benefits for subsets of patients and may impact treatment utilization. [ABSTRACT FROM AUTHOR]

Political polarization surrounding the COVID-19 health crisis has been on the rise since the beginning of the pandemic. We combine prior research on motivated reasoning, selective exposure, and news framing to understand the association between partisan media use and social distancing behavior related to COVID-19. To do so, we collected media content data and national survey data during the onset of the pandemic. We employed structural topic modeling (STM), dependency parsing, word co-occurrence, and manual coding to examine the media coverage. Next, we analyzed survey data collected with a Qualtrics panel from a sample of U.S. residents for factors explaining social distancing behaviors. Results reveal coverage from the right leaning outlets downplayed the virus and highlighted the consequences of lockdowns on the economy. Our survey findings show that even after accounting for a range of demographic, political orientation, and COVID-19 awareness variables, conservative media use was linked, although modestly, with a lower likelihood of social distancing behavior. Our findings echo past research on media framing of pandemics and their association with public attitudes and behavior. [ABSTRACT FROM AUTHOR]

Understanding the implications of aggressive political debate style amid corrosive modes of campaign politics requires fine-grained analyses of political performance, attending to multiple communication modalities. Politicians’ facial expressions, emotional tone, and speech content can all independently convey aggression and dominance, and often work in combination for purposes of emphasis. Yet micro-coding individual visual, tonal, and verbal features across more than a handful of debate segments becomes extremely labor intensive, hampering research, especially historical, longitudinal, and cross-cultural work. To address this limitation, we develop a novel multimodal classifier using an Interaction Canonical Correlation Network (ICCN) that incorporates video and audio features with speech coding of candidate debate performance, trained on a 20% sample of 10- second segments from each of the first televised U.S presidential debates between 1980 and 2020. In the analysis, we demonstrate this classifier can accurately detect aggression by political candidates in U.S. debates. We sharpen its performance by distinguishing between debate eras characterized by lower and higher levels of aggression and validate the approach by comparing the performance of unimodal with multimodal classification. This classifier opens new avenues for computational social science research, including explaining candidate behavior within debates at a larger scale and across different eras. [ABSTRACT FROM AUTHOR]