BACKGROUND:Metabolic dysfunction (MetD) and alcohol use disorder (AUD) frequently coexist as synergistic risk factors for steatotic liver disease. Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are established therapies for MetD, including type 2 diabetes mellitus (T2DM) and obesity. Recent studies suggested potential beneficial effects of GLP-1RA to decrease addictive behaviours in AUD. We evaluated the outcomes of GLP-1RA therapy compared with FDA-approved pharmacotherapies for AUD, including naltrexone, acamprosate, and disulfiram, in patients with dual risk factors of MetD and AUD. METHODS:We conducted a retrospective cohort study of patients at Stanford Health Care (2017-2025). Eligible patients had a concurrent diagnosis of alcohol-related complications meeting criteria for AUD and MetD, including obesity (BMI > 25) and/or a history of T2DM with HbA1c > 5.7. Those with advanced liver disease within 1 year of diagnosis were excluded. Exposure groups included ≥ 6 months of GLP-1RA therapy (semaglutide or tirzepatide) in comparison with FDA-approved pharmacotherapies for AUD. Propensity score matching was employed to reduce the effects of confounding factors. RESULTS:In total, 1946 patients were diagnosed concurrently with AUD and MetD. Of them, 274 patients were exposed to GLP-1RA, 1272 to naltrexone, 232 to acamprosate, and 168 to disulfiram. Patients were followed for an average of 1341 days. Patients exposed to GLP-1RA had higher BMI (35.5 vs. 30.1) and more T2DM (66% vs. 14%). GLP-1RA therapy was associated with lower 1-year AUD relapse [IRR 0.55, 95% CI 0.42-0.73; p < 0.01], greater BMI reduction (-1.3 vs. -0.3; p = 0.004), and HbA1c improvement (-1.0 vs. +0.1; p = 0.02). The incidence of decompensated cirrhosis trended lower but was not statistically significant [HR 0.52, p = 0.09]. Mortality was similar. CONCLUSIONS:GLP-1RAs are a promising option for patients with concurrent MetD and AUD, improving relapse rates and metabolic outcomes compared with currently FDA-approved pharmacotherapies for AUD. Trends toward better liver outcomes support further prospective evaluation.

Aims Use of sodium-glucose cotransporter 2 inhibitors (SGLT2i) in adults with Type 2 diabetes (AwT2D) has recently been associated with decreased stroke risk. In this study, we investigate the impact of SGLT2i on stroke. Materials and Methods We conducted a cohort study using an electronic health record (EHR) representing 66 million patients to examine whether AwT2D have a lower incidence of stroke and/or transient ischemic attack (TIA) if treated with SGLT2i. All analyses were completed utilizing high-dimensional propensity score matching to control for confounders. Results AwT2D treated with SGLT2i compared to other antidiabetic medications, assessed after 1 year, starting 2 weeks after the date of prescription of SGLT2i or other antidiabetic drug, had a significantly lower incidence of any stroke (Odds Ratio (OR): 0.84 with 95% Confidence Interval (CI) 0.79-0.91, p < 0.001), hemorrhagic stroke (OR: 0.72 with 95% CI 0.59-0.88, p < 0.001), ischemic stroke (OR: 0.86 with 95% CI 0.77-0.95, p = 0.005), and TIA (OR: 0.87 with 95% CI 0.80-0.95, p = 0.003). Conclusion Among AwT2D, those treated with SGLT2i had lower incidence of stroke and TIA compared with adults treated with other antidiabetic medications.

Background: The observational literature on comparative effectiveness is expanding rapidly but remains difficult to synthesize. Discordant findings often stem from structural differences in cohort definitions, inclusion criteria, and follow up windows, leaving stakeholders without a cohesive evidence base. Furthermore, studies typically focus on a narrow subset of outcomes, neglecting the broader needs of diverse healthcare stakeholders 1,2,3,4. Methods We developed a high throughput evidence generation workflow using linked EHR and administrative claims data. The cornerstone is a prespecified measurement architecture applied uniformly across clinical scenarios: six post index windows (acute to two year follow.up); 28 Elixhauser comorbidities; 14 healthcare resource utilization (HCRU) categories; 29 laboratory measures with 52 binary thresholds; and 42 adverse event categories. We generated unadjusted treatment comparisons across ~1,038 outcomes per scenario, including effect-measure modification (EMM) assessments across 130 baseline features. Results Across 40 clinical domains, the workflow produced approximately 32,982,552 outcome evaluations. An evaluation included a treatment comparison outcome population effect estimate with uncertainty bounds and supporting diagnostics. Approximately 5,000 narrative summaries underwent structured clinical and statistical quality control before dissemination. Conclusions Standardized, high throughput workflows can shift evidence generation away from fragmented studies toward comprehensive evidence packages. This shared evidence base supports precision medicine by making treatment effect heterogeneity visible across clinically meaningful subpopulations, reducing the need for redundant, stakeholder-specific studies. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study was funded by Atropos Health. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: IRB Stanford University School of Medicine waived ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors.

BACKGROUND:Intestinal stricture affects over half of patients with Crohn's disease (CD) and has significant associated morbidity. Statins possess both anti-inflammatory and anti-fibrotic properties and may improve CD outcomes, although current data are limited. This study assessed whether statin use is associated with a reduced risk of new stricture development in CD in a diverse US population, and evaluated the role of IBD therapy in any association. METHODS:We conducted a retrospective cohort study comparing patients with CD with and without statin exposure using the EVERSANA US electronic health records database. Findings were independently validated in a second database (Merative MarketScan). Patient demographics, co-morbidities, laboratory measurements, and CD medications were assessed. The primary outcome was development of new stricture as defined by a composite endpoint of an encounter for stricture diagnosis or occurrence of a stricture-related procedure. Propensity score (PS)-matched Cox proportional hazards models were used to estimate associations. RESULTS:The EVERSANA cohort comprised 1210 statin recipients and 25 000 non-statin users. Over a mean follow-up of 3.8 years, PS-matched statin users had a 28% reduction in the risk of new-onset stricture (hazard ratio [HR] 0.72, 95% confidence interval [CI] 0.53-0.99, P = .043). The Merative cohort contained 9577 statin users and 56 918 non-statin users. Over a mean follow-up of 3.6 years, PS-matched statin use had a 29% risk reduction in new-onset stricture (HR 0.71, 95% CI 0.66-0.77, P < .001). CONCLUSIONS:Statin use is independently associated with reduced progression to stricture formation in two PS-matched large and diverse cohorts of patients with CD.

Background Paroxysmal Nocturnal Haemoglobinuria (PNH) is a rare (3.81 per 100K), treatable clonal hematopoietic stem cell (HSC) disorder characterized by intravascular hemolysis, thrombosis, and smooth muscle dystonias, with bone marrow failure occurring in some cases. Patients undergo lengthy diagnostic journeys frequently exceeding a year. Often diagnosis is made following a high morbidity/mortality event, such as a stroke. Earlier identification and treatment may improve disease burden. Worker, et al. 2024 demonstrated feasibility with a sample of 131 patients with PNH and 74 features. We sought to replicate and extend the finding for use in a relevant clinical workflow by reducing time from symptomology to diagnosis. We evaluate whether we identify patients likely to benefit from workup for PNH because they are likely to receive an initial PNH diagnosis 3 to 12 months from initial workup. Methods The study used the Atropos Health Aracadia data asset, which is derived from population health management software used across the United States (67MM total and 1,208 PNH patients). We used a set of 416K patients with a Hb (index date) between 2016 and 2023 and no prior PNH diagnosis. Cases were defined by the presence of ICD-10 D59.5 from 3 to 12 months following their index date (n=306 of 1,049 total PNH patients). Controls were randomly sampled from the total population of patients with Hb. For clinical features, the look back period was up to 24 months prior to the index date. We used a 70/30 train/test split. Several machine learning techniques were attempted, and the results of an XGBoost model are reported here. Models were built to maximize the areas under the precision recall curve (AUPRC), clinical resonance, and ease of deployment. In addition, two physicians manually reviewed 50 individuals selected for maximum prediction discord (e.g., high predicted likelihood of PNH but no PNH diagnosis found). Results We identified 306 PNH patients (0.07% of sample) with an initial diagnosis of PNH 3 to 12 months after a Hb. AUPRC and AUROC were 0.09 and 0.77, respectively for a full model with 375 features and 0.08 and 0.77 for a model with 13 features. Important features included: history of anemia (aplastic and broadly defined), overall disease burden, age, haptoglobin, and having a diagnosis for myelodisplastic syndrome (MDS). Clinical review agreed that 100% of “unlikely” PNH patients did not warrant further workup. Among the “false positives” (>34% probability of PNH, but no PNH diagnosis found in 3 to 12 months), physicians indicated 75% to 90% (two physicians) of these patients warranted further workup. The predictive model is available at https://github.com/atroposhealth/pnh-undiagnosed. Conclusion The performance of this machine learning algorithm is sufficient to deploy to suggest potential workup for PNH in order to reduce diagnostic delays for patients with PNH. Setting the threshold for further workup should be tested within a local environment due to variation in lab testing (i.e., Hb) patterns.

BackgroundOtologic adverse events (AEs) have been occasionally reported as sequalae of COVID-19 vaccination, although their incidence in comparison with that of preexisting vaccines with high uptake remains unclear. This study compared the rates of new-onset otologic AEs among matched adults receiving mRNA COVID-19 vaccination versus influenza vaccination.MethodsThis retrospective cohort study used electronic health records (EHR) data from Stanford Health Care to identify adults aged 50–89 years with no history of otologic disorders prior to first Pfizer/Moderna COVID-19 vaccine (December 2020–January 2022) or any pre-pandemic influenza vaccine (January 2016–December 2019). Patients were categorized by vaccination into FluVax or COVIDVax cohorts. A 90-day history pre-vaccination (baseline period) and ≥6 months follow-up post-vaccination were required. Event rates of new-onset hearing loss (HL), sudden HL, tinnitus, vertigo/dizziness, aural fullness, and otalgia in the 6 months post-vaccination were compared between groups after high-dimensional propensity score (hdPS) matching. A sensitivity analysis was conducted among patients with no COVID-19 infection at any time. Odds ratios (ORs) were calculated using logistic regression for the hdPS matched cohorts.ResultsAfter hdPS matching, 20,325 patients were included into the FluVax and COVIDVax cohorts, respectively (mean age: 65.5 and 65.2 years; 53.1 and 53.8% females). The rates of otologic AEs in the 6 months post-vaccination were similarly low for the FluVax and COVIDVax cohorts: 1.16% vs. 1.16% for any HL, 0.01% vs. 0.02% for sudden HL, 0.41% vs. 0.47% for tinnitus, 1.96% vs. 1.59% for vertigo, 0.27% vs. 0.25% for otalgia, and 0.09% vs. 0.2% for aural fullness. COVIDVax patients had lower odds of vertigo [OR 95% CI: 0.81 (0.70, 0.94)] and higher odds of aural fullness [2.16 (1.25–3.72)] than the FluVax patients (both p < 0.05). The results of the sensitivity analysis limited to patients with no COVID-19 infection at any time (N = 17,530 each cohort) were consistent with the primary results, but aural fullness was the only AE with statistically higher risk in the COVIDVax vs. FluVax cohort [OR (95% CI): 1.90 (1.09–3.31); p = 0.021].ConclusionNew-onset otologic AEs were rare among a large cohort of hdPS-matched patients who received mRNA COVID-19 or pre-pandemic flu vaccination at a single institution. Although aural fullness was statistically more common in the COVIDVax vs. FluVax cohort, regardless of COVID-19 infection status, it remained extremely rare (<0.22%) in any cohort. These results indicate a similar otologic safety profile of the two vaccines, although future research is recommended in larger EHR databases to corroborate the findings.

Background Risk models to predict perioperative mortality rates (POMR) are critical to surgical quality improvement yet are not widely adapted for use in humanitarian and low-resource settings (LRS). We developed a POMR and corresponding nomogram and calculator for use in humanitarian surgical care. Methods Electronic health record data from a high-income academic medical center from 2015 to 2019 were retrospectively extracted, selecting variables and operations specific to LRS. This development dataset was used to create a logistic regression POMR model, which was then prospectively validated using data from 2022 to 2023 from the same institution. Results EHR from a total of 49,277 patients were used. The model fitted eight variables feasibly obtainable in LRS: age > 65 years (OR = 4.05 and 95% CI: 3.71-4.43), male sex (OR = 1.32 and 95% CI: 1.25-1.40), GCS < 13 (OR = 5.20 and 95% CI: 4.73-5.73), glucose > 200 mg/dL (OR = 2.19 and 95% CI: 2.01-2.38), Hgb <= 11 g/dL (OR = 2.65 and 95% CI: 2.43-2.89), HR > 120 bpm (OR = 2.49 and 95% CI: 2.35-2.64), T > 38 degrees Celsius (OR = 1.32 and 95% CI: 1.19-1.45), and SBP > 180 mmHg (OR = 1.18 and 95% CI: 1.02-1.37). The model demonstrated a high area under the curve (0.847, 0.867, and 0.925), sensitivity (0.739, 0.886, and 0.844), specificity (0.807, 0.780, and 0.864), and negative predictive value (0.750, 0.997, and 0.999) on training, holdout, and prospective validation sets. Conclusion We validated a POMR model for use in LRS using eight variables that are readily available in the target environment. This model's predictors and accompanying clinical tools of an Excel calculator and nomogram make it simultaneously comprehensive and accessible in LRS.

Surveillance colonoscopy reduces colorectal cancer (CRC) risk in inflammatory bowel disease (IBD), but anecdotal evidence suggests a link between surveillance colonoscopy and symptoms flare, which may affect patient compliance with CRC surveillance. We investigated the effects of surveillance colonoscopy in quiescent IBD using a national database to conduct a retrospective, propensity score-matched (PSM) cohort study of 1,717 patients with IBD and 1,717 patients without IBD who underwent colonoscopy. Strict inclusion criteria were used to select patients with quiescent IBD undergoing surveillance colonoscopy. We demonstrated that patients with quiescent IBD prior to colonoscopy received significantly more post-colonoscopy steroid prescriptions compared to matched controls (OR 1.46 [1.20, 1.77], p<0.001). Steroid prescriptions were more likely in patients with IBD at longer follow-up intervals, suggesting a delayed mechanism of onset for post-colonoscopy inflammation. No significant differences between groups were observed for the other outcomes of post-procedure emergency room and hospital visits or enteric infections. Our results suggest a potential risk of symptom exacerbation, evidenced by increased steroid prescriptions, following surveillance colonoscopy. Given that nearly 25% of patients with IBD do not receive surveillance colonoscopy at the recommended intervals, active post-procedure symptoms may be a contributing factor that warrants further investigation. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study was funded by The Leona M. and Harry B. Helmsley Charitable Trust (Grant 2007-04026), Paul G. Allen Frontiers Group, and Stanford Plant Based Diet Initiative. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Institutional Review Board of Stanford University gave ethical approval for this work. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The data that support the findings of this study are available from EVERSANA but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. The authors will make a good faith effort to provide access to the data upon reasonable request.

Surveillance colonoscopy is routinely performed in patients with inflammatory bowel disease (IBD) to decrease the risk of colorectal cancer (CRC). There is limited data regarding exacerbation of gastrointestinal symptoms as an adverse effect of colonoscopy, which may impact patient compliance with CRC surveillance. Using a large national database, we conducted a propensity score-matched, retrospective cohort study of 1,604 patients with IBD and 25,000 patients without IBD who underwent surveillance or screening colonoscopy from 2010 to 2023. Outcomes of composite emergency room (ER) visits and hospitalizations of any cause, steroid prescriptions, and gastroenterology (GI) office visits were assessed over a follow-up interval of days 7 to 30 after colonoscopy. Strict inclusion criteria were used to select for patients with inactive IBD disease undergoing surveillance exams by requiring at least 8 years of IBD disease history, no calprotectin level >120 in the 90-day period preceding colonoscopy, and use of billing codes associated with surveillance exams. The inclusion criteria were chosen to avoid inclusion of patients who received interventions for pre-existing endoscopic inflammation discovered at time of colonoscopy, rather than for symptoms developed after colonoscopy. Our inclusion criteria were validated using the Stanford Research Repository (STARR) and had 94% accuracy for patients in endoscopic remission. In propensity score-matched analysis, patients with IBD received a significantly higher rate of steroid prescriptions following colonoscopy compared to the control group (9.4% vs 6.4%, OR 1.51 [1.16-1.96], p=0.002). There was no significant difference in composite ER visits and hospitalizations or GI office visits between the two groups. Patients with IBD experienced the greatest increased risk of steroid prescription during the latter half of the 30-day follow-up interval compared to the 1-week (OR 1.37 [1.01-1.85], p=0.045) or 2-week follow-up interval (OR 1.24 [0.93-1.66], p=0.143). Patients with IBD were carefully selected with strict inclusion criteria to have quiescent disease. Patients with inactive IBD were prescribed steroids at a 51% higher rate compared to other patients following colonoscopy, though rates of ER visits and hospitalizations were similar to controls. These results suggest that patients with IBD may be at higher risk of symptom flares following colonoscopy, though symptoms do not appear to be severe enough to warrant ER or hospital presentation. Given that nearly 25% of patients with IBD do not receive surveillance colonoscopy at the recommended intervals, active symptoms after colonoscopy may be a contributing factor that warrants further investigation.

Objective:The practice of evidence-based medicine can be challenging when relevant data are lacking or difficult to contextualize for a specific patient. Large language models (LLMs) could potentially address both challenges by summarizing published literature or generating new studies using real-world data. Materials and Methods:We submitted 50 clinical questions to five LLM-based systems: OpenEvidence, which uses an LLM for retrieval-augmented generation (RAG); ChatRWD, which uses an LLM as an interface to a data extraction and analysis pipeline; and three general-purpose LLMs (ChatGPT-4, Claude 3 Opus, Gemini 1.5 Pro). Nine independent physicians evaluated the answers for relevance, quality of supporting evidence, and actionability (i.e., sufficient to justify or change clinical practice). Results:General-purpose LLMs rarely produced relevant, evidence-based answers (2-10% of questions). In contrast, RAG-based and agentic LLM systems, respectively, produced relevant, evidence-based answers for 24% (OpenEvidence) to 58% (ChatRWD) of questions. OpenEvidence produced actionable results for 48% of questions with existing evidence, compared to 37% for ChatRWD and <5% for the general-purpose LLMs. ChatRWD provided actionable results for 52% of questions that lacked existing literature compared to <10% for other LLMs. Discussion:Special-purpose LLM systems greatly outperformed general-purpose LLMs in producing answers to clinical questions. Retrieval-augmented generation-based LLM (OpenEvidence) performed well when existing data were available, while only the agentic ChatRWD was able to provide actionable answers when preexisting studies were lacking. Conclusion:Synergistic systems combining RAG-based evidence summarization and agentic generation of novel evidence could improve the availability of pertinent evidence for patient care.

The growing use of large language models (LLMs) for biomedical question answering raises concerns about the accuracy and evidentiary support of their responses. To address this, we present Answered with Evidence, a framework for evaluating whether LLM-generated answers are grounded in scientific literature. We analyzed thousands of physician-submitted questions using a comparative pipeline that included: (1) Alexandria, fka the Atropos Evidence Library, a retrieval-augmented generation (RAG) system based on novel observational studies, and (2) two PubMed-based retrieval-augmented systems (System and Perplexity). We found that PubMed-based systems provided evidence-supported answers for approximately 44