News media coverage of COVID-19 public health and policy information

The study investigates how traditional print and online newspapers in Canada, the United Kingdom, and the United States covered COVID-19 public health and policy information from March to mid-August 2020. It addresses the research question: what were the amount, scientific quality, sensationalism, and topical emphases of COVID-19 reporting across newspapers spanning the political spectrum, and how did this relate to public health contexts? The context is a rapidly evolving pandemic characterized by scientific uncertainty, policy coordination challenges, and widespread misinformation/disinformation. News media serve as a key conduit among health professionals, policymakers, and the public, influencing risk perceptions, accountability, and policy debates. The purpose is to systematically quantify scientific quality (alignment with current evidence and uncertainty) and sensationalism (discursive strategies amplifying extraordinariness) in pandemic reporting, providing insights to inform future media practices and public health communication strategies.

Prior research on pandemic media coverage (H1N1 2009, SARS 2003, Ebola 2014) has identified problems such as inadequate scientific quality, over- or understatement of risks, insufficient communication of protective measures, false balance between expert consensus and uninformed opinion, and rapid decline in coverage despite ongoing risks. Media dramatization and politicization can shape public perceptions and policy responses, affecting trust in science and coordination. National newspapers can set agendas across media ecosystems. These literatures motivate assessing both scientific quality and sensationalism using validated tools and situating coverage within varying governance contexts of Canada, the U.K., and the U.S.

Design and scope: Comparative content analysis of print and online newspaper articles across 12 national outlets spanning the political spectrum in Canada, the U.K., and the U.S. Timeframe: 1 March–15 August 2020, starting two weeks before WHO’s pandemic declaration and covering five subsequent months.

Public health context: Compiled and visualized existing datasets on COVID-19 cases, deaths, and testing (e.g., Our World in Data, national dashboards) and cataloged key public health declarations, policies, and guidance at international, national, and subnational levels to contextualize media trends.

Search strategy and inclusion criteria: Retrieved records from Factiva using terms: coronavirus, COVID-19, epidemic, outbreak, pandemic, SARS-CoV-2. Eligible articles were original news reporting or news analysis in English that (1) had a direct focus on public health implications of COVID-19 or attempts to control its spread, and (2) focused on the newspaper’s country of publication; excluded were opinion pieces, editorials, letters, interviews, microblogs, snippets, roundups, obituaries, ads, corrections.

Sampling: Randomly sampled one day per week for six consecutive 4-week periods (Monday–Saturday only), yielding 24 sampled days. For each sampled day and outlet, retrieved all COVID-19-related records, screened for eligibility, and randomly selected up to five eligible articles per outlet per day for coding; if fewer than five were available, all eligible articles were included. Across sampled days and outlets, the outlets published an estimated 18,430 COVID-19-related articles, with 23.4% (≈4321) eligible; 1331 articles were coded.

Coding tool and measures: Adapted the Hoffman and Justicz (2016) tool to COVID-19 context (SI Coding Tool). Scientific quality was assessed with six 1–5 Likert items: applicability, opinion vs facts, validity, precision, context, and a global assessment (higher scores = higher quality). Sensationalism was assessed with six 1–5 Likert items: exposing, speculating, generalizing, warning, extolling, and a global assessment (higher scores = higher sensationalism). Article metadata recorded included coder ID, title, date, outlet (and whether syndicated), societal sectors (up to two), and public health measures mentioned.

Coder training and reliability: Three coders underwent six weeks of training and calibration, including iterative coding of example articles, group discussions, development of COVID-19-specific examples and decision rules. Interrater reliability was monitored; discrepancies of 3–4 points triggered reconciliation discussions. Reliability was quantified using weighted Cohen’s kappa (quadratic weights) suitable for ordinal, high-inference coding. Interrater reliability was moderate to substantial for global assessments and similarly high for most specific items, with lower reliability where score distributions were highly skewed.

Statistical analysis: Used Kruskal–Wallis one-way ANOVA for group comparisons with post hoc multiple comparisons (kruskalmc, pgirmess in R). Also compared original vs syndicated articles. Temporal trends and topic associations were examined by categorizing societal sectors and public health measures coded per article.

• Volume and eligibility: Across 24 sampled days, 12 outlets published 18,430 COVID-19-related articles; an estimated 23.4% (≈4321) met eligibility criteria. A total of 1331 eligible articles were coded.
• Scientific quality overall: Moderate overall scientific quality across outlets. Populist-right outlets had the lowest global scientific quality within each country: Toronto Sun mean 2.58 (n=106), Daily Mail 2.67 (n=115), New York Post 2.28 (n=118); Kruskal–Wallis p≤0.001; within-country pairwise p≤0.05 (except Daily Mail vs Times of London and Telegraph). Lower quality was especially evident in validity, precision, and context.
• Sensationalism overall: Sensationalism was low across outlets. In Canada and the U.S., outlets on the left and center had somewhat higher sensationalism than populist-right outlets. Canada: Toronto Sun had the lowest global sensationalism (mean 1.77, n=106; Kruskal–Wallis p≤0.001; pairwise p≤0.05 vs Globe and Mail and National Post). U.S.: Wall Street Journal 2.03 (n=118) and New York Post 2.13 (n=118) were lower than New York Times 2.40 (n=120) and Washington Post 2.38 (n=119); Kruskal–Wallis p≤0.001; pairwise p≤0.05. U.K.: no significant variation across outlets (p=0.283).
• Exposing and warning: In the U.S., left-leaning outlets (New York Times, Washington Post) had more exposing and warning coverage (e.g., highlighting policy failures, misinformation, and disease risks) than right-leaning outlets.
• Original vs syndicated: Original reporting had higher scientific quality than syndicated articles (mean global 2.93, n=1278 vs 2.71, n=54; p=0.020). Syndicated articles had lower sensationalism (1.82, n=54) than original reporting (2.14, n=1278; p≤0.001). The Toronto Sun had the highest proportion of syndicated content among coded items (34% of its 106 articles), followed by Toronto Star (6%) and National Post (11%).
• Topic associations: Articles on healthcare and institutions showed higher scientific quality (mean global 3.23) and health-related research even higher (3.72) than other topics (Kruskal–Wallis p≤0.001; most pairwise p≤0.05). Sensationalism was highest for politics (2.53) and foreign affairs (2.49) (Kruskal–Wallis p<0.001). U.S. outlets devoted the greatest share to politics/elections (63.8% of coverage across U.S. outlets).
• Temporal patterns: Coverage volume rose sharply in March and declined through May–June, not rebounding despite U.S. case surges in June–July. Scientific quality and sensationalism were generally stable over time, with an instance of lower quality on 3 July 2020 linked to limited healthcare-sector coverage.
• Policy measures covered: Consistent coverage of social distancing, testing/tracing, PPE/disinfection; increasing coverage of mask guidance and reopening policies over time.

The analysis shows that during early COVID-19 response, newspaper coverage across Canada, the U.K., and the U.S. exhibited moderate scientific quality and low sensationalism, with significant variation by political orientation and topic. Populist-right outlets consistently provided lower scientific quality—particularly in validity, precision, and contextualization—while also exhibiting very low sensationalism. In the U.S., higher sensationalism elements such as exposing and warning appeared more in left-leaning outlets amid pronounced failures of national leadership and widespread disinformation, coinciding with greater politicization and a high proportion of political coverage.

These patterns have implications for public understanding and accountability: low sensationalism is generally valued, but when combined with low scientific quality and few warnings or exposés, coverage may under-communicate risks and policy failures, potentially exacerbating adverse public health outcomes. Topic focus matters: healthcare and research coverage tended to be of higher scientific quality, whereas politics-related coverage featured higher sensationalism. Reliance on syndicated content—particularly notable in some Canadian outlets—was associated with lower scientific quality but also lower sensationalism, reflecting structural pressures in the news industry.

Findings underscore the interplay among evolving scientific evidence, politicization, and media practices. Given the agenda-setting role of national newspapers, understanding these dynamics can inform strategies by public health institutions and policymakers to craft clearer guidance and facilitate coordinated responses.

This study systematically quantified the amount, scientific quality, sensationalism, and topical emphases of early COVID-19 newspaper coverage across politically diverse outlets in Canada, the U.K., and the U.S. Coverage was generally of moderate scientific quality and low sensationalism, with populist-right outlets exhibiting the lowest scientific quality and, in some cases, very low sensationalism. U.S. left-leaning outlets featured more exposing and warning content against a backdrop of high disease burden and politicization. Healthcare and research topics were associated with higher scientific quality; politics and foreign affairs with higher sensationalism. Original reporting outperformed syndicated content on scientific quality.

These results highlight a tension: efforts to avoid sensationalism can, when paired with low scientific rigor, diminish the salience of genuine public health risks and policy shortcomings. Insights from this analysis can guide future pandemic reporting and public health communication strategies to better align with evolving science, mitigate politicization, and effectively alert the public to risks and evidence-based measures.