Online health information behaviour and its association with statin adherence in patients with high cardiovascular risk: A prospective cohort study.

Objective: Statins are effective for preventing cardiovascular disease. However, many patients decide not to take statins because of negative influences, such as online misinformation. Online health information may affect decisions on medication adherence, but measuring it is challenging. This study aimed to examine the associations between online health information behaviour and statin adherence in patients with high cardiovascular risk. Methods: A prospective cohort study involving 233 patients with high cardiovascular risk was conducted at a primary care clinic in Malaysia. Participants used a digital information diary tool to record online health information they encountered for 2 months and completed a questionnaire about statin necessity, concerns and adherence at the end of the observation period. Data were analysed using structural equation modelling. Results: The results showed that 55.8% (130 of 233 patients) encountered online health information. Patients who actively sought online health information (91 of 233 patients) had higher concerns about statin use (ß = 0.323, p = 0.023). Participants with higher concern about statin use were also more likely to be non-adherent (ß = -0.337, p < 0.001). Patients who actively sought online health information were more likely to have lower statin adherence, mediated by higher concerns about statin use (indirect effect, ß = -0.109, p = 0.048). Conclusions: Our results suggest that patients with higher levels of concern about statins may be actively seeking online information about statins, and their concerns might influence how they search, what they find, and the potential to encounter misinformation. Our study highlights the importance of addressing patients' concerns about medications to improve adherence.

New Frontiers in Health Literacy: Using ChatGPT to Simplify Health Information for People in the Community.

BACKGROUND: Most health information does not meet the health literacy needs of our communities. Writing health information in plain language is time-consuming but the release of tools like ChatGPT may make it easier to produce reliable plain language health information. OBJECTIVE: To investigate the capacity for ChatGPT to produce plain language versions of health texts. DESIGN: Observational study of 26 health texts from reputable websites. METHODS: ChatGPT was prompted to 'rewrite the text for people with low literacy'. Researchers captured three revised versions of each original text. MAIN MEASURES: Objective health literacy assessment, including Simple Measure of Gobbledygook (SMOG), proportion of the text that contains complex language (%), number of instances of passive voice and subjective ratings of key messages retained (%). KEY RESULTS: On average, original texts were written at grade 12.8 (SD = 2.2) and revised to grade 11.0 (SD = 1.2), p < 0.001. Original texts were on average 22.8% complex (SD = 7.5%) compared to 14.4% (SD = 5.6%) in revised texts, p < 0.001. Original texts had on average 4.7 instances (SD = 3.2) of passive text compared to 1.7 (SD = 1.2) in revised texts, p < 0.001. On average 80% of key messages were retained (SD = 15.0). The more complex original texts showed more improvements than less complex original texts. For example, when original texts were ≥ grade 13, revised versions improved by an average 3.3 grades (SD = 2.2), p < 0.001. Simpler original texts (< grade 11) improved by an average 0.5 grades (SD = 1.4), p < 0.001. CONCLUSIONS: This study used multiple objective assessments of health literacy to demonstrate that ChatGPT can simplify health information while retaining most key messages. However, the revised texts typically did not meet health literacy targets for grade reading score, and improvements were marginal for texts that were already relatively simple.

Conflict of interest policies at Australian medical schools.

BACKGROUND: Pharmaceutical industry exposure is widespread during medical training and may affect education and clinical decision-making. Medical faculties' conflict of interest (COI) policies help to limit this exposure and protect students against commercial influence. AIMS: Our aim was to investigate the prevalence, content and strength of COI policies at Australian medical schools and changes since a previous assessment conducted in 2009. METHODS: We identified policies by searching medical school and host university websites in January 2021, and contacted deans to identify any missed policies. We applied a modified version of a scorecard developed in previous studies to examine the content of COI policies. All data were coded in duplicate. COI policies were rated on a scale from 0 (no policy) to 2 (strong policy) across 11 items per medical school. Oversight mechanisms and sanctions were also assessed, and current policies were compared with the 2009 study. RESULTS: Of 155 potentially relevant policies, 153 were university-wide and two were specific to medical schools. No policies covered sales representatives, on-site sponsored education or free samples. Oversight of consultancies had improved substantially, with 76% of schools requiring preapproval. Disclosure policies, while usually present, were weak, with no public disclosure required. CONCLUSION: We found little indication that Australian medical students are protected from commercial influence on medical education, and there has been limited COI policy development within the past decade. More attention is needed to ensure the independence of medical education in Australia.

A comparison of machine learning methods to find clinical trials for inclusion in new systematic reviews from their PROSPERO registrations prior to searching and screening.

Searching for trials is a key task in systematic reviews and a focus of automation. Previous approaches required knowing examples of relevant trials in advance, and most methods are focused on published trial articles. To complement existing tools, we compared methods for finding relevant trial registrations given a International Prospective Register of Systematic Reviews (PROSPERO) entry and where no relevant trials have been screened for inclusion in advance. We compared SciBERT-based (extension of Bidirectional Encoder Representations from Transformers) PICO extraction, MetaMap, and term-based representations using an imperfect dataset mined from 3632 PROSPERO entries connected to a subset of 65,662 trial registrations and 65,834 trial articles known to be included in systematic reviews. Performance was measured by the median rank and recall by rank of trials that were eventually included in the published systematic reviews. When ranking trial registrations relative to PROSPERO entries, 296 trial registrations needed to be screened to identify half of the relevant trials, and the best performing approach used a basic term-based representation. When ranking trial articles relative to PROSPERO entries, 162 trial articles needed to be screened to identify half of the relevant trials, and the best-performing approach used a term-based representation. The results show that MetaMap and term-based representations outperformed approaches that included PICO extraction for this use case. The results suggest that when starting with a PROSPERO entry and where no trials have been screened for inclusion, automated methods can reduce workload, but additional processes are still needed to efficiently identify trial registrations or trial articles that meet the inclusion criteria of a systematic review.

How do patients with high cardiovascular risk evaluate online health information? A qualitative study.

BACKGROUND: People are exposed to variable health information from the Internet, potentially influencing their health decision-making and behaviour. It remains a challenge for people to discern between good- and poor-quality online health information (OHI). This study explored how patients evaluate and determine trust in statin-related OHI in patients with high cardiovascular risk. METHODS: This qualitative study used vignettes and think-aloud methods. We recruited patients from a primary care clinic who were at least 18 years old, had high cardiovascular risk and had previously sought OHI. Participants were given two statin-related vignettes: Vignette 1 (low-quality information) and Vignette 2 (high-quality information). Participants voiced their thoughts aloud when reading the vignettes and determined the trust level for each vignette using a 5-point Likert scale. This was followed by a semi-structured interview which was audio-recorded and transcribed verbatim. The transcripts were coded and analysed using thematic analysis. RESULTS: A total of 20 participants were recruited, with age ranging from 38-74 years. Among all the high cardiovascular-risk participants, eight had pre-existing cardiovascular diseases. For Vignette 1 (low-quality information), five participants trusted it while nine participants were unsure of their trust. 17 participants (85%) trusted Vignette 2 (high-quality information). Five themes emerged from the analysis of how patients evaluated OHI: (1) logical content, (2) neutral stance and tone of OHI content, (3) credibility of the information source, (4) consistent with prior knowledge and experience, and (5) corroboration with information from other sources. CONCLUSION: Patients with high cardiovascular risks focused on the content, source credibility and information consistency when evaluating and determining their trust in statin-related OHI. Doctors should adopt a more personalised approach when discussing statin-related online misinformation with patients by considering their prior knowledge, beliefs and experience of statin use.

What generative AI means for trust in health communications.

ABSTRACTLarge language models are fundamental technologies used in interfaces like ChatGPT and are poised to change the way people access and make sense of health information. The speed of uptake and investment suggests that these will be transformative technologies, but it is not yet clear what the implications might be for health communications. In this viewpoint, we draw on research about the adoption of new information technologies to focus on the ways that generative artificial intelligence (AI) tools like large language models might change how health information is produced, what health information people see, how marketing and misinformation might be mixed with evidence, and what people trust. We conclude that transparency and explainability in this space must be carefully considered to avoid unanticipated consequences.

K-PathVQA: Knowledge-Aware Multimodal Representation for Pathology Visual Question Answering.

Pathology imaging is routinely used to detect the underlying effects and causes of diseases or injuries. Pathology visual question answering (PathVQA) aims to enable computers to answer questions about clinical visual findings from pathology images. Prior work on PathVQA has focused on directly analyzing the image content using conventional pretrained encoders without utilizing relevant external information when the image content is inadequate. In this paper, we present a knowledge-driven PathVQA (K-PathVQA), which uses a medical knowledge graph (KG) from a complementary external structured knowledge base to infer answers for the PathVQA task. K-PathVQA improves the question representation with external medical knowledge and then aggregates vision, language, and knowledge embeddings to learn a joint knowledge-image-question representation. Our experiments using a publicly available PathVQA dataset showed that our K-PathVQA outperformed the best baseline method with an increase of 4.15% in accuracy for the overall task, an increase of 4.40% in open-ended question type and an absolute increase of 1.03% in closed-ended question types. Ablation testing shows the impact of each of the contributions. Generalizability of the method is demonstrated with a separate medical VQA dataset.

Measuring the burden of infodemics with a research toolkit for connecting information exposure, trust, and health behaviours.

BACKGROUND: During a public health emergency, accurate and useful information can be drowned out by questions, concerns, information voids, conflicting information, and misinformation. Very few studies connect information exposure and trust to health behaviours, which limits available evidence to inform when and where to act to mitigate the burden of infodemics, especially in low resource settings. This research describes the features of a toolkit that can support studies linking information exposure to health behaviours at the individual level. METHODS: To meet the needs of the research community, we determined the functional and non-functional requirements of a research toolkit that can be used in studies measuring topic-specific information exposure and health behaviours. Most data-driven infodemiology research is designed to characterise content rather than measure associations between information exposure and health behaviours. Studies also tend to be limited to specific social media platforms, are unable to capture the breadth of individual information exposure that occur online and offline, and cannot measure differences in trust by information source or content. Studies are also designed very differently, limiting synthesis of results. RESULTS: We demonstrate a way to address these requirements via a web-based study platform that includes an app that participants use to record topic-specific information exposure, a browser plugin for tracking access to relevant webpages, questionnaires that can be delivered at any time during a study, and app-based incentives for participation such as visual analytics to compare trust levels with other participants. Other features of the platform include the ability to tailor studies to local contexts, ease of use for participants, and frictionless sharing of de-identified data for aggregating individual participant data in international meta-analyses. CONCLUSIONS: Our proposed solution will be able to capture detailed data about information exposure and health behaviour data, standardise study design while simultaneously supporting localisation, and make it easy to synthesise individual participant data across studies. Future research will need to evaluate the toolkit in realistic scenarios to understand the usability of the toolkit for both participants and investigators.

Helping patient educators meet health literacy needs: End-user testing and iterative development of an innovative health literacy editing tool.

Objective: The Sydney Health Literacy Lab (SHeLL) Editor is an online text-editing tool that provides real-time assessment and feedback on written health information (assesses grade reading score, complex language, passive voice). This study aimed to explore how the design could be further enhanced to help health information providers interpret and act on automated feedback. Methods: The prototype was iteratively refined across four rounds of user-testing with health services staff (N = 20). Participants took part in online interviews and a brief follow-up survey using validated usability scales (System Usability Scale, Technology Acceptance Model). After each round, Yardley's (2021) optimisation criteria guided which changes would be implemented. Results: Participants rated the Editor as having adequate usability (M = 82.8 out of 100, SD = 13.5). Most modifications sought to reduce information overload (e.g. simplifying instructions for new users) or make feedback motivating and actionable (e.g. using frequent incremental feedback to highlight changes to the text altered assessment scores). Conclusion: terative user-testing was critical to balancing academic values and the practical needs of the Editor's target users. The final version emphasises actionable real-time feedback and not just assessment. Innovation: The Editor is a new tool that will help health information providers apply health literacy principles to written text.

Utility and usability evaluation of an information diary tool to measure health information access and exposure among patients with high cardiovascular risk.

Background: Online health misinformation about statins potentially affects health decision-making on statin use and adherence. We developed an information diary platform (IDP) to measure topic-specific health information exposure where participants record what information they encounter. We evaluated the utility and usability of the smartphone diary from the participants' perspective. Methods: We used a mixed-method design to evaluate how participants used the smartphone diary tool and their perspectives on usability. Participants were high cardiovascular-risk patients recruited from a primary care clinic and used the tool for a week. We measured usability with the System Usability Scale (SUS) questionnaire and interviewed participants to explore utility and usability issues. Results: The information diary was available in three languages and tested with 24 participants. The mean SUS score was 69.8 ± 12.9. Five themes related to utility were: IDP functions as a health information diary; supporting discussion of health information with doctors; wanting a feedback function about credible information; increasing awareness of the need to appraise information; and wanting to compare levels of trust with other participants or experts. Four themes related to usability were: ease of learning and use; confusion about selecting the category of information source; capturing offline information by uploading photos; and recording their level of trust. Conclusion: We found that the smartphone diary can be used as a research instrument to record relevant examples of information exposure. It potentially modifies how people seek and appraise topic-specific health information.

Measuring the Burden of Infodemics: Summary of the Methods and Results of the Fifth WHO Infodemic Management Conference.

Background: An infodemic is excess information, including false or misleading information, that spreads in digital and physical environments during a public health emergency. The COVID-19 pandemic has been accompanied by an unprecedented global infodemic that has led to confusion about the benefits of medical and public health interventions, with substantial impact on risk-taking and health-seeking behaviors, eroding trust in health authorities and compromising the effectiveness of public health responses and policies. Standardized measures are needed to quantify the harmful impacts of the infodemic in a systematic and methodologically robust manner, as well as harmonizing highly divergent approaches currently explored for this purpose. This can serve as a foundation for a systematic, evidence-based approach to monitoring, identifying, and mitigating future infodemic harms in emergency preparedness and prevention. Objective: In this paper, we summarize the Fifth World Health Organization (WHO) Infodemic Management Conference structure, proceedings, outcomes, and proposed actions seeking to identify the interdisciplinary approaches and frameworks needed to enable the measurement of the burden of infodemics. Methods: An iterative human-centered design (HCD) approach and concept mapping were used to facilitate focused discussions and allow for the generation of actionable outcomes and recommendations. The discussions included 86 participants representing diverse scientific disciplines and health authorities from 28 countries across all WHO regions, along with observers from civil society and global public health-implementing partners. A thematic map capturing the concepts matching the key contributing factors to the public health burden of infodemics was used throughout the conference to frame and contextualize discussions. Five key areas for immediate action were identified. Results: The 5 key areas for the development of metrics to assess the burden of infodemics and associated interventions included (1) developing standardized definitions and ensuring the adoption thereof; (2) improving the map of concepts influencing the burden of infodemics; (3) conducting a review of evidence, tools, and data sources; (4) setting up a technical working group; and (5) addressing immediate priorities for postpandemic recovery and resilience building. The summary report consolidated group input toward a common vocabulary with standardized terms, concepts, study designs, measures, and tools to estimate the burden of infodemics and the effectiveness of infodemic management interventions. Conclusions: Standardizing measurement is the basis for documenting the burden of infodemics on health systems and population health during emergencies. Investment is needed into the development of practical, affordable, evidence-based, and systematic methods that are legally and ethically balanced for monitoring infodemics; generating diagnostics, infodemic insights, and recommendations; and developing interventions, action-oriented guidance, policies, support options, mechanisms, and tools for infodemic managers and emergency program managers.

An evaluation of existing text de-identification tools for use with patient progress notes from Australian general practice.

INTRODUCTION: Digitized patient progress notes from general practice represent a significant resource for clinical and public health research but cannot feasibly and ethically be used for these purposes without automated de-identification. Internationally, several open-source natural language processing tools have been developed, however, given wide variations in clinical documentation practices, these cannot be utilized without appropriate review. We evaluated the performance of four de-identification tools and assessed their suitability for customization to Australian general practice progress notes. METHODS: Four tools were selected: three rule-based (HMS Scrubber, MIT De-id, Philter) and one machine learning (MIST). 300 patient progress notes from three general practice clinics were manually annotated with personally identifying information. We conducted a pairwise comparison between the manual annotations and patient identifiers automatically detected by each tool, measuring recall (sensitivity), precision (positive predictive value), f1-score (harmonic mean of precision and recall), and f2-score (weighs recall 2x higher than precision). Error analysis was also conducted to better understand each tool's structure and performance. RESULTS: Manual annotation detected 701 identifiers in seven categories. The rule-based tools detected identifiers in six categories and MIST in three. Philter achieved the highest aggregate recall (67%) and the highest recall for NAME (87%). HMS Scrubber achieved the highest recall for DATE (94%) and all tools performed poorly on LOCATION. MIST achieved the highest precision for NAME and DATE while also achieving similar recall to the rule-based tools for DATE and highest recall for LOCATION. Philter had the lowest aggregate precision (37%), however preliminary adjustments of its rules and dictionaries showed a substantial reduction in false positives. CONCLUSION: Existing off-the-shelf solutions for automated de-identification of clinical text are not immediately suitable for our context without modification. Philter is the most promising candidate due to its high recall and flexibility however will require extensive revising of its pattern matching rules and dictionaries.

Experiences and influences of online health information-seeking about statin use in patients with high cardiovascular risk: a qualitative study.

OBJECTIVES: Online health information (OHI) has been shown to influence patients' health decisions and behaviours. OHI about statins has created confusion among healthcare professionals and the public. This study explored the views and experiences of patients with high cardiovascular risk on OHI-seeking about statins and how OHI influenced their decision. DESIGN: This was a qualitative study using semi-structured in-depth interviews. An interpretive description approach with thematic analysis was used for data analysis. SETTING: An urban primary care clinic in Kuala Lumpur, Malaysia. PARTICIPANTS: Patients aged 18 years and above who had high cardiovascular risk and sought OHI on statins were recruited. RESULTS: A total of 20 participants were interviewed. The age of the participants ranged from 38 to 74 years. Twelve (60%) participants took statins for primary cardiovascular disease prevention. The duration of statin use ranged from 2 weeks to 30 years. Six themes emerged from the data analysis: (i) seeking OHI throughout the disease trajectory, (ii) active and passive approaches to seeking OHI, (iii) types of OHI, (iv) views about statin-related OHI, (v) influence of OHI on patients' health decisions, and (vi) patient-doctor communication about OHI. CONCLUSION: This study highlights the changing information needs throughout patient journeys, suggesting the opportunity to provide needs-oriented OHI to patients. Unintentional passive exposure to OHI appears to have an influence on patients' adherence to statins. The quality of patient-doctor communication in relation to OHI-seeking behaviour remains a critical factor in patient decision-making.

Multiple Automated Health Literacy Assessments of Written Health Information: Development of the SHeLL (Sydney Health Literacy Lab) Health Literacy Editor v1.

Producing health information that people can easily understand is challenging and time-consuming. Existing guidance is often subjective and lacks specificity. With advances in software that reads and analyzes text, there is an opportunity to develop tools that provide objective, specific, and automated guidance on the complexity of health information. This paper outlines the development of the SHeLL (Sydney Health Literacy Lab) Health Literacy Editor, an automated tool to facilitate the implementation of health literacy guidelines for the production of easy-to-read written health information. Target users were any person or organization that develops consumer-facing education materials, with or without prior experience with health literacy concepts. Anticipated users included health professionals, staff, and government and nongovernment agencies. To develop this tool, existing health literacy and relevant writing guidelines were collated. Items amenable to programmable automated assessment were incorporated into the Editor. A set of natural language processing methods were also adapted for use in the SHeLL Editor, though the approach was primarily procedural (rule-based). As a result of this process, the Editor comprises 6 assessments: readability (school grade reading score calculated using the Simple Measure of Gobbledygook (SMOG)), complex language (percentage of the text that contains public health thesaurus entries, words that are uncommon in English, or acronyms), passive voice, text structure (eg, use of long paragraphs), lexical density and diversity, and person-centered language. These are presented as global scores, with additional, more specific feedback flagged in the text itself. Feedback is provided in real-time so that users can iteratively revise and improve the text. The design also includes a "text preparation" mode, which allows users to quickly make adjustments to ensure accurate calculation of readability. A hierarchy of assessments also helps users prioritize the most important feedback. Lastly, the Editor has a function that exports the analysis and revised text. The SHeLL Health Literacy Editor is a new tool that can help improve the quality and safety of written health information. It provides objective, immediate feedback on a range of factors, complementing readability with other less widely used but important objective assessments such as complex and person-centered language. It can be used as a scalable intervention to support the uptake of health literacy guidelines by health services and providers of health information. This early prototype can be further refined by expanding the thesaurus and leveraging new machine learning methods for assessing the complexity of the written text. User-testing with health professionals is needed before evaluating the Editor's ability to improve the health literacy of written health information and evaluating its implementation into existing Australian health services.

Evolving Face Mask Guidance During a Pandemic and Potential Harm to Public Perception: Infodemiology Study of Sentiment and Emotion on Twitter.

BACKGROUND: Throughout the COVID-19 pandemic, US Centers for Disease Control and Prevention policies on face mask use fluctuated. Understanding how public health communications evolve around key policy decisions may inform future decisions on preventative measures by aiding the design of communication strategies (eg, wording, timing, and channel) that ensure rapid dissemination and maximize both widespread adoption and sustained adherence. OBJECTIVE: We aimed to assess how sentiment on masks evolved surrounding 2 changes to mask guidelines: (1) the recommendation for mask use on April 3, 2020, and (2) the relaxation of mask use on May 13, 2021. METHODS: We applied an interrupted time series method to US Twitter data surrounding each guideline change. Outcomes were changes in the (1) proportion of positive, negative, and neutral tweets and (2) number of words within a tweet tagged with a given emotion (eg, trust). Results were compared to COVID-19 Twitter data without mask keywords for the same period. RESULTS: There were fewer neutral mask-related tweets in 2020 (ß=-3.94 percentage points, 95% CI -4.68 to -3.21; P<.001) and 2021 (ß=-8.74, 95% CI -9.31 to -8.17; P<.001). Following the April 3 recommendation (ß=.51, 95% CI .43-.59; P<.001) and May 13 relaxation (ß=3.43, 95% CI 1.61-5.26; P<.001), the percent of negative mask-related tweets increased. The quantity of trust-related terms decreased following the policy change on April 3 (ß=-.004, 95% CI -.004 to -.003; P<.001) and May 13 (ß=-.001, 95% CI -.002 to 0; P=.008). CONCLUSIONS: The US Twitter population responded negatively and with less trust following guideline shifts related to masking, regardless of whether the guidelines recommended or relaxed mask usage. Federal agencies should ensure that changes in public health recommendations are communicated concisely and rapidly.

Natural language processing for clinical notes in dentistry: A systematic review.

OBJECTIVE: To identify and synthesise research on applications of natural language processing (NLP) for information extraction and retrieval from clinical notes in dentistry. MATERIALS AND METHODS: A predefined search strategy was applied in EMBASE, CINAHL and Medline. Studies eligible for inclusion were those that that described, evaluated, or applied NLP to clinical notes containing either human or simulated patient information. Quality of the study design and reporting was independently assessed based on a set of questions derived from relevant tools including CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS). A narrative synthesis was conducted to present the results. RESULTS: Of the 17 included studies, 10 developed and evaluated NLP methods and 7 described applications of NLP-based information retrieval methods in dental records. Studies were published between 2015 and 2021, most were missing key details needed for reproducibility, and there was no consistency in design or reporting. The 10 studies developing or evaluating NLP methods used document classification or entity extraction, and 4 compared NLP methods to non-NLP methods. The quality of reporting on NLP studies in dentistry has modestly improved over time. CONCLUSIONS: Study design heterogeneity and incomplete reporting of studies currently limits our ability to synthesise NLP applications in dental records. Standardisation of reporting and improved connections between NLP methods and applied NLP in dentistry may improve how we can make use of clinical notes from dentistry in population health or decision support systems. PROTOCOL REGISTRATION: PROSPERO CRD42021227823.

Characteristics of clinical trials associated with early results reporting at ClinicalTrials.gov.

OBJECTIVE: We aimed to investigate the trial characteristics associated with earlier results reporting on ClinicalTrials.gov. STUDY DESIGN AND SETTING: We sampled interventional trials registered with ClinicalTrials.gov and examined the time from trial completion to results reporting on ClinicalTrials.gov as the event of interest. A Cox proportional hazards model was used to examine associations between the time to results reporting on ClinicalTrials.gov with funding type, intervention type, number of enrolled participants, trial phase, trial allocation status, and the year of trial completion. The model accounts for multiple risk factors simultaneously. RESULTS: Among 102,404 completed trials, the median follow-up for the result reporting event was 18.5 months (IQR 12.7-33.6), during which time 25% (26,608 of 102,404) had results available on ClinicalTrials.gov. Compared to industry funded trials (18.1 months), non-industry trials (median 18.8 months) had results reported slower (HR 0.35, 95% CI 0.34-0.36); compared to drug trials (18.4 months) non-drug trials (19.0 months) were reported slower (HR 0.61, 95% CI 0.59-0.64); compared to trials with more than 50 participants (18.0 months), smaller trials (19.3 months) were reported slower (HR 0.97, 95% CI 0.94-0.99). CONCLUSION: Non-industry, non-drug, and earlier phase trials reported results on ClinicalTrials.gov more slowly if at all. Much of the efforts aimed at improving trial reporting through structured reporting on ClinicalTrials.gov have been focused on industry funded drug trials, but these results suggest that incentives and tools targeting non-industry and non-drug trials are also needed.

Association between online health information-seeking and medication adherence: A systematic review and meta-analysis.

Background: The evidence of the impact of online health information-seeking (OHIS) on health outcomes has been conflicting. OHIS is increasingly recognised as a factor influencing health behaviour but the impact of OHIS on medication adherence remains unclear. Objectives: We conducted a systematic review and meta-analysis to examine the associations between OHIS and medication adherence. Methods: We searched Medline, Embase, Web of Science, Scopus, CINAHL and Psychology and Behavioural Science Collection for studies published up to December 2020. The inclusion criteria were studies that reported the associations of OHIS and medication adherence, quantitative design, reported primary data only, related to any health condition where medications are used and conducted on patients either in clinical or community settings. A meta-analysis was used to examine the association between OHIS and medication adherence. Results: A total of 17 studies involving 24,890 patients were included in this review. The study designs and results were mixed. In the meta-analysis, there was no significant association (n = 7, OR 1.356, 95% CI 0.793-2.322, p = 0.265), or correlation (n = 4, r = -0.085, 95% CI -0.572-0.446, p = 0.768) between OHIS and medication adherence. In the sub-group analysis of people living with HIV/AIDS, OHIS was associated with better medication adherence (OR 1.612, 95% CI 1.266-2.054, p < 0.001). Conclusions: The current evidence of an association between OHIS and medication adherence is inconclusive. This review highlights methodological issues on how to measure OHIS objectively and calls for in-depth exploration of how OHIS affects health decisions and behaviour.