The impact of COVID-19-related changes in media consumption on public knowledge: results of a cross-sectional survey of Pennsylvania adults

COVID-19 knowledge correlates with trusted news source. To increase public knowledge of COVID-19 in order to maximize information dissemination and compliance with COVID-19-related public health recommendations, those who provide health information should consider use of the public’s most trusted sources of information, as well as monitoring and correcting misinformation presented by other sources. Independent content review for accuracy in media may be warranted in public health emergencies to improve knowledge.

Those who selected government health websites as their most trusted source were more likely to answer COVID-19 questions correctly than those who selected other internet news sources or television news (OR 1.21, p < .05; 1.08, p > .05; and 0.87, p < .05, respectively). Those who used Facebook as an additional source of news in any way were less likely to answer COVID-19 questions correctly than those who did not (OR 0.93, p < .05).

A cross-sectional online survey was sent to a convenience sample from a list of adults on a central Pennsylvania health system’s marketing database 25–31 March 2020. Respondents were grouped by their trusted news sources and comparison of respondent COVID-19 knowledge was made between these groups for 5948 respondents.

During the COVID-19 pandemic, Americans have increasingly relied on internet versus television news. The extent to which this change in health news consumption practice impacts health knowledge is not known. This study investigates the relationship between most trusted information source and COVID-19 knowledge.

Introduction The COVID-19 pandemic has become a global tragedy. Worldwide, 124.3 million cases have been reported and over 2,736,668 have lost their lives as of March 24, 20211. As nations continue to address this pandemic, effective communication between healthcare experts and national populations is paramount to public health2–7. Past pandemics have shown that ineffective messaging may result in public misconceptions, poor compliance with precautionary measures, overuse of health services, and facilitate inadequate health and public policy decisions8–11. For health messaging to be effective, it must be received, understood and believed by the public. Therefore, understanding public information sources is an important component of health messaging campaigns, and understanding their relationship to knowledge informs how to approach addressing public misconceptions. In 2003, 80% of adult internet users searched for at least one of 16 major health topics online12. In 2009, 61% of Americans searched online for health information13. In 2017, 50% of Americans got their news primarily from television, 43% primarily from the internet and 66% got at least some of their news from social media14. Information sources for COVID-19 appear to be quite different. The single most trusted news sources about COVID-19 for central Pennsylvania adults are government websites (42.8%), television (27.2%) and health system communications (9.3%)15. This change in information sources may play a significant role in how the public has understood and acted (or not acted) upon public health messaging, and may partially explain the public misconception and poor compliance already reported during the COVID-19 pandemic15–18. The extent to which preferred information sources relate to knowledge and behavior during the COVID-19 pandemic is a small, but growing, area of research. A political survey found that Americans appear to have maintained similar news habits19. One large study of social media users evaluated information sources and knowledge related to COVID-19 and found that traditional media was the most frequently utilized source of information (91.2%), and that the most trusted source of information was government websites (45.2%)20. That study also found that the use of social media information may not be associated with significantly different awareness about an emerging health crisis like COVID-19, but the authors note that because their sample was a nonprobability convenience sample of Facebooki (and affiliated platforms used), their results may not be generalizable20. To overcome potential biases inherent to political and nonprobability social medial platform surveys, this COVID-19 study explores relationships between information sources and knowledge in an apolitical survey of a patient population. Detailed understanding of public information consumption and knowledge will inform community outreach and health messaging strategies necessary to improve compliance with public health recommendations, and give insights into the generalizability of current research in the field.

Methods The development, design and implementation of the survey used are described in detail elsewhere15. Briefly, a cross-sectional online survey was sent to a convenience sample from a list of adults on a central Pennsylvania health system’s marketing database 25–31 March 2020. From a database of 121,573 unique, valid email addresses, 5948 respondents completed the survey (73.7% of all those who opened the survey, 50.9% of those who opened the survey link, 11.1% of those who opened the email and 4.8% of the entire email address list recipients). This study focuses on associations between knowledge, information sources and demographic characteristics. The Pennsylvania State University College of Medicine Institutional Review Board approved the study protocol. This study adhered to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE). Knowledge questions were scored as incorrect or correct, and each had an associated 5 point confidence score (1 = Extremely confident, 5 = Not at all confident [just guessing]). Simple imputation was used for skipped or missed confidence levels for questions answered. Binary knowledge answers were analyzed via a generalized linear mixed-effects model (GLMM) with a logistic link function and a random effect for the participant, modeling the probability of a correct response. The logarithm of the estimated odds ratios from the GLMM were transformed to probability estimates, with 95% confidence intervals (CIs). To report adjusted p values, the false-discovery rate was applied. SAS Version 9.4 was used.

Discussion Participants who used government health websites were more likely to correctly answer COVID-19 questions than other groups. This is similar to Ali et al., who reported that use of government websites resulted in significantly better knowledge for 3 of their 7 COVID-19 questions (with no difference in the other questions)20. Those who relied on television news as a primary source, or who used Facebook as an additional news source generally, were less likely to answer COVID-19 questions correctly than other groups. This is concerning, as half of Americans report using television as a source of news14,15 and 66% of Americans use social media such as Facebook as a news source14. This finding differs from that reported by Ali et al.20, which may be a reflection of their sampling method (derived from Facebook advertising). Further evaluation of knowledge across multiple social media sites, and the impact of Facebook’s efforts at “fact-checking” in part due to misinformation reported during COVID-19, is indicated. Lower knowledge in these groups is consistent with a 2018 study that found an inverse relationship between health literacy and using television or social media as a source of information21. Whether these relationships are due to misinformation, a lack of good information or re-enforcement of false beliefs is unknown but, given these associations, it would behoove health communicators to monitor these venues to correct misinformation when it appears. Our study has several strengths. It is one of the largest pandemic information studies reported to date. The survey was rapidly disseminated amidst the early weeks of the pandemic, offering insight into the public use of information in the midst of a national emergency. Further, we elicited both the single most trusted source as well as secondary sources, which may offer a better indication of primary information source influence over behavior. Our study is limited by a lack of racial diversity or urban representation. Also, during the survey period, the central Pennsylvania area had not directly experienced significant impacts of COVID-19, thus responses from populations with greater burden may be different. These limitations may decrease the generalizability of these findings. Further, selection bias may contribute to the relatively high knowledge scores, as those who chose to respond to the survey may be more attentive to COVID-19-related issues. The survey was cross-sectional and results may not be generalizable over time.