Riaz, Irbaz

Abstract The adoption of electronic health records (EHRs) has transformed health care, improving efficiency and chart accessibility. However, the widespread reliance on unstructured data entry and the lack of standardized documentation frameworks have resulted in significant data fragmentation across health care systems. The prevalence of unstructured data in EHRs limits their potential for clinical decision support, trial matching, real‐world evidence (RWE) generation, and quality measurement. Data fragmentation in health care triggers a cascade of challenges that ultimately compromise patient care. Clinicians face an excessive documentation burden and struggle to locate critical information buried in unstructured notes. Researchers encounter difficulties in extracting reliable clinical data. EHR vendors grapple with standardizing unstructured information for interoperability, and payers are unable to process unstructured clinical data efficiently to support value‐based care models. These challenges are particularly acute in oncology, where complex clinical elements like cancer staging, disease status, and treatment changes require precise, structured documentation. Emerging artificial intelligence (AI) technologies, such as large language models (LLMs) and ambient listening, offer a path to automate structured data generation while reducing the workload on providers. Here, the authors propose LLM‐based workflows that balance automation with clinician verification, streamlining data entry without compromising accuracy. Realizing these benefits requires coordinated efforts among clinicians, researchers, EHR vendors, payers, and policymakers to align regulatory frameworks with AI‐driven innovations. This article outlines a strategy to enhance structured data capture within EHRs, ultimately improving patient care, research, and health care efficiency.

Objective To determine the harms of ezetimibe in people who need lipid-lowering treatment. Design Systematic review and meta-analysis. Data sources Randomised controlled trials and cohort studies. Eligibility criteria for selecting studies Studies comparing ezetimibe with placebo, standard care, or other lipid-lowering agents in people who need lipid-lowering treatment with a follow-up duration of at least six months (or 24 weeks). The relative effects for potential harms of ezetimibe were pooled by use of random effect pairwise meta-analyses for randomised controlled trials and the evidence from observational studies was narratively summarised. The certainty of evidence was assessed using the Grading of Recommendation Assessment, Development, and Evaluation. Results 48 randomised controlled trials with 28 444 participants (median follow-up 34 weeks, range 24-312 weeks) and four observational studies with 1667 participants (median follow-up 282 weeks, range 72-400 weeks) were included. The meta-analyses of randomised trials showed moderate to high certainty that ezetimibe was not associated with cancer (relative risk 1.01; 95% confidence interval 0.92 to 1.11), fractures (0.90; 0.74 to 1.10), discontinuation due to any adverse event (0.87; 0.74 to 1.03), gastrointestinal adverse events leading to discontinuation (1.34; 0.58 to 3.08), myalgia or muscular pain leading to discontinuation (0.82; 0.51 to 1.33), neurocognitive events (1.48; 0.58 to 3.81), or new-onset diabetes (0.88; 0.61 to 1.28). The narrative analysis of observational studies provided consistent findings. No credible subgroup effects were identified for the harm outcomes, including shorter versus longer follow-up duration of trials. Conclusions Ezetimibe results in little to no difference in adverse events or other undesirable effects compared with placebo, usual care or other lipid-lowering agents. Review registration PROSPERO CRD42020187437.