Citations

e17004 Background: Living clinical practice guidelines offer the opportunity to keep recommendations aligned with the latest evidence, but they also present a fundamental challenge: as evidence accumulates, panelists cannot sustainably query and synthesize a growing body of literature. In developing ASCO's living metastatic castration-resistant prostate cancer (mCRPC) guideline, we encountered this challenge firsthand. To address it, we developed and evaluated an interactive agentic framework that enables guideline panelists to query evidence in natural language and retrieve structured facts, tables, and figures—with the goal of extending this tool as a clinician-facing companion to enhance guideline accessibility. Methods: The mCRPC systematic review dataset supporting ASCO's clinical practice guideline was used, comprising 212 structured columns capturing trial characteristics, patient populations, and outcomes extracted from 188 references across 88 clinical trials. A large language model (LLM)-based agentic framework was developed employing a coding agent that translates natural language queries into executable scripts for structured evidence retrieval. Claude (Haiku 4.5, Sonnet 4.5, and Opus 4.5) served as the core LLM agents. We constructed a benchmark of 50 fact-based queries with gold-standard answers targeting trial identification, treatment modalities, and endpoint characteristics (e.g., "Which trials evaluated systemic monotherapy?"). Performance was evaluated using exact-match accuracy for final answers, and precision, recall, and F1 score for trial-level retrieval. Results: Across 50 queries, exact-match accuracy ranged from 82% to 86%. Haiku 4.5 and Sonnet 4.5 each achieved 82% (42/50), while Opus 4.5 achieved 86% (43/50). Precision ranged from 0.93–0.97, recall from 0.94–0.96, and F1 scores from 0.91–0.94. Conclusions: This agentic framework enables scalable, auditable, natural language interrogation of guideline evidence, addressing a critical bottleneck in living guideline maintenance. For panelists, it supports rapid, reproducible evidence surveillance during recommendation development. For clinicians, it offers a foundation for an interactive companion tool that transforms static guidelines into queryable resources—enabling personalized evidence retrieval at the point of care. LLM EM Precision Recall F1 claude-haiku-4-5 0.82(42/50) 0.95 0.94 0.91 claude-sonnet-4-5 0.82(42/50) 0.97 0.96 0.93 claude-opus-4-5 0.86(43/50) 0.93 0.96 0.94

e16229 Background: Liver cancer is a leading cause of cancer-related mortality worldwide, with rising incidence linked to metabolic disorders. Conditions such as obesity, diabetes mellitus, and metabolic syndrome contribute substantially to non-viral liver carcinogenesis. However, long-term temporal trends and mortality disparities among adults with metabolic disorders in the United States remain incompletely characterized. Methods: Mortality data for liver cancer among adults with metabolic disorders were obtained from the centers for disease control and prevention (CDC) mortality database. Adults aged 25 years and older were included and age-adjusted mortality rates (AAMRs) per 100,000 were calculated. Temporal trends were evaluated using joinpoint regression and average annual percent change (AAPC) with 95% confidence intervals. Results: From 1999–2023, AAMR for liver cancer in adults with metabolic disorders rose from 0.17 to 0.65 per 100,000 in the U.S., with an overall AAPC of 5.95% [95% CI 5.2–6.7]. Men had higher absolute mortality (0.25→0.93; AAPC 5.68%), while women showed a sharper recent rise (0.11→0.41; AAPC 7.08%), peaking at 11.86% annually after 2016. Hispanic/Latino (0.27→0.89; AAPC 6.73%) and Black populations (0.15→0.75; AAPC 6.15%) experienced steady increases, whereas White mortality accelerated post-2011 (AAPC 9.37%). Regionally, the West had highest rates, the South showed highest growth, and Northeast/Midwest had sharp recent increases. Rural mortality overtook urban by 2020 (0.50 vs. 0.48). Most deaths occurred in medical facilities (48.2%) or home (33.1%), with highest state rates in Hawaii (0.56), Texas (0.47), and Rhode Island (0.47). Conclusions: Mortality from liver cancer in adults with metabolic disorders rose sharply in the U.S. since 1999, with accelerating trends and marked disparities by sex, race/ethnicity, region, and urbanization. These findings highlight an urgent need for targeted metabolic risk reduction and equity-focused prevention strategies. Trends in Age-Adjusted Mortality Rates (AAMR) from 1999 to 2023 by demographic and geographic variables. Variable AAMR 1999 (per 100,000) AAMR 2023 (per 100,000) AAPC (95% CI) Overall 0.16 0.64 5.94* (5.22 – 6.67) Males 0.25 0.92 5.67* (4.82 – 6.5) Females 0.11 0.40 7.08* (5.79 – 8.38) White 0.16 0.63 6.12* (5.41 – 6.82) Black 0.15 0.75 6.15* (5.11 – 7.19) Hispanic or Latino 0.27 0.89 6.73* (5.98 – 7.48) Variable AAMR 1999 (per 100,000) AAMR 2020 (per 100,000) AAPC (95% CI) Urban 0.16 0.47 5.38* (4.43 – 6.33) Rural 0.19 0.50 5.73* (4.48 – 6.99) AAPC denotes the average annual percent change; * indicates p < 0.05.

e16157 Background: Liver cancer mortality remains a major public health burden in the United States despite advances in prevention and treatment. Psychoactive substance use disorder (PSUD) is an increasingly prevalent comorbidity; however, national trends and disparities in liver cancer mortality among adults with PSUD remain poorly characterized. Methods: We conducted a retrospective analysis of liver cancer-related deaths among U.S. adults aged 25 years and older with PSUD from 1999 to 2023 using Centers of Disease Control (CDC) Wide Ranging Online Data for Epidemiological Research (WONDER) mortality database. Age-adjusted mortality rates (AAMRs) were calculated per 100,000 population and stratified by census region, state, urban–rural status, sex, and race/ethnicity. Temporal trends were assessed using Joinpoint regression to estimate average annual percent change (AAPC). Statistical significance was determined using p-value (<0.05). Results: Overall liver cancer related AAMR per 100,000 among adults with PSUD significantly increased from 0.15 in 1999 to 1.02 in 2023 (AAPC: 7.47; 95% CI: 6.33 to 8.61; p < 0.000001). The overall mortality burden was higher in males. However, females experienced larger rises in mortality (AAPC: 10.03; 95% CI: 8.85 to 11.24; p < 0.000001) as compared to males (AAPC: 7.30; 95% CI: 5.78 to 8.84; p < 0.000001). Marked racial disparities were observed, with non-Hispanic Black and non-Hispanic White individuals experiencing the highest mortality rates with persistent increases over time, while Hispanic/Latino populations exhibited modest and variable trends. A substantial geographic variability was observed, with the greatest increases in the Midwest and the South. State-level analysis revealed higher mortality rate in Oregon (1.27) followed by Washington (1.12) and Texas (1.06) respectively. Both rural and urban areas experienced a sharp rise in mortality with AAPC being slightly higher in rural areas (AAPC: 10.34) as compared to urban areas (AAPC: 9.67). Conclusions: Liver cancer mortality among U.S. adults with PSUD has risen markedly over the past two decades, with substantial demographic and geographic disparities. These findings highlight the need for integrated liver cancer screening and substance use disorder interventions targeting high-risk populations. Variables AAMR (95% CI) in 2023 AAPC (95% CI) Overall 1.02 (0.98 –1.05) 1999-2023: 7.47* (6.33 – 8.61) Female 0.42 (0.39 – 0.46) 1999-2023: 10.03* (8.85 – 11.24) Men 1.69 (1.62 –1.76) 1999-2023: 7.30* (5.78 – 8.84) White 1.05 (1.0 –1.09) 1999-2023: 9.52* (8.08 – 10.98) Black / African Americans 1.22 (1.09 – 1.35) 1999-2023: 6.59* (5.50 – 7.68) Hispanics / Latinos 0.75 (0.65 – 0.84) 1999-2023: 1.26 (-1.01 – 3.60) AAMR (95% CI) in 2020 AAPC (95% CI) Rural 1.51 (1.39 – 1.62) 1999-2020: 10.34* (5.05 – 15.90) Urban 0.92 (0.82 – 0.95) 1999-2020: 9.67* (7.32 – 12.06)

e16027 Background: Postoperative oral intake after esophagectomy has traditionally been delayed due to concerns about anastomotic leakage. Although Enhanced Recovery After Surgery (ERAS) pathways increasingly support early feeding, its safety remains debated. This study compared early versus delayed oral feeding following esophagectomy for esophageal cancer. Methods: PubMed, Embase, and Scopus were systematically searched through October 2025 for randomized controlled trials evaluating early oral feeding (postoperative days 0–3) versus delayed feeding (postoperative day ≥5) after esophagectomy. Primary outcomes were anastomotic leakage and pneumonia. Secondary outcomes included postoperative complications, gastrointestinal recovery, length of hospital stay, healthcare costs, and quality of life. Pooled risk ratios (RRs), mean differences (MDs), or standardized mean differences (SMDs) with 95% confidence intervals were calculated using Review Manager 5.4. Results: Thirteen trials involving 1,468 patients were included. Early oral feeding was not associated with increased anastomotic leakage (RR: 0.89, 95% CI: 0.55–1.45; I² = 0%) and showed consistent safety across surgical approaches and anastomotic sites. The incidence of pneumonia was significantly reduced (RR: 0.66, 95% CI: 0.47–0.92; p = 0.01). Early feeding accelerated gastrointestinal recovery, including earlier first bowel movement (MD: −0.50 days, p < 0.00001) and passage of flatus (MD: −0.61 days, p = 0.0006). Hospital stay was shortened (MD: −1.89 days, p = 0.002), with lower medical costs (SMD: −1.19, p = 0.009). Quality-of-life measures improved, including global health status, dysphagia, pain, and swallowing function. Conclusions: Early oral feeding after esophagectomy appears safe and beneficial, reducing pulmonary complications and enhancing postoperative recovery. Further high-quality multicenter trials are needed to confirm these findings.

e16024 Background: Patients with advanced esophageal squamous cell carcinoma (ESCC) continue to have poor outcomes with chemotherapy alone. The addition of PD-L1 inhibitors to first-line chemotherapy has shown encouraging results in individual trials, but the overall magnitude and consistency of benefit across key clinical endpoints have not been fully clarified. To better understand the effects of PD-L1 inhibitors used in combination with chemotherapy, we performed a systematic review and meta-analysis to compare the efficacy and safety of PD-L1-based combination therapies to chemotherapy alone in patients with advanced or metastatic ESCC. Methods: A systematic search of PubMed, Embase, and Scopus identified randomized controlled trials comparing PD-L1 inhibitor plus chemotherapy versus chemotherapy alone in patients with advanced or metastatic ESCC. Three eligible studies met inclusion criteria. We calculated pooled hazard ratios (HRs) for overall survival (OS) and progression-free survival (PFS); and pooled risk ratios (RRs) for objective response rate (ORR) and Grade 3-5 treatment-related adverse events (TRAEs) using random effects models. We quantified the heterogeneity among studies using the I² test. Results: Across three randomized trials, the addition of a PD-L1 inhibitor to chemotherapy resulted in a statistically significant and highly consistent survival advantage. OS was longer with combination therapy (HR 0.67, 95% CI 0.60–0.76; p < 0.00001), representing a 33% reduction in the risk of death, with no heterogeneity across studies (I² = 0%). Similarly, PFS was prolonged with combination therapy (HR 0.61, 95% CI 0.55–0.68; p < 0.00001; I² = 0%), corresponding to a 39% reduction in the risk of disease progression or death. Tumor responses occurred at higher frequencies in patients who received PD-L1 inhibitor-based therapy than in patients who received chemotherapy alone (RR = 1.49, 95% CI = 1.35-1.63; p < .00001; I² = 0%). The increased efficacy observed with combination therapy was not accompanied by an increase in severe toxicities. The rates of Grade 3-5 TRAEs were similar between the two groups (RR = 1.06, 95% CI = 0.99-1.13; p = 0.09; I² = 0%). Conclusions: Based on evidence from three randomized trials, the addition of PD-L1 inhibitors to standard chemotherapy improves both survival and response rates in patients with advanced ESCC without an increase in severe treatment related adverse events. Therefore, PD-L1-based combination therapies represent a promising new therapeutic option for patients with advanced ESCC.

e16023 Background: Stroke is an underrecognized cause of death among patients with gastric cancer. While overall cancer mortality has improved, less is known about long-term trends and disparities in stroke-related deaths. We examined national patterns in stroke mortality among gastric cancer patients from 1999 to 2023. Methods: U.S. national mortality data were used to calculate age-adjusted mortality rates (AAMR) per 100,000. Joinpoint regression identified temporal changes and estimated annual percent change (APC) and average annual percent change (AAPC). Analyses were stratified by age, sex, race and ethnicity, census region, and urbanization. Results: Between 1999 and 2023, an overall decline in stroke-related mortality was noted with AAMR falling from 167.29 to 111.35 per 100,000 people (AAPC −1.70%, 95% CI −1.9 to −1.5). The decline was greatest between 2002 and 2009 (−5.28% per year) after which it began to plateau until it rose significantly between 2018 and 2021 (APC 6.56%) followed by a decline after 2021 (−3.63%). Older adults had the highest level of stroke related mortality, and therefore the greatest reduction over time (from 756.36 to 479.63) with a total reduction of −36.6% yet they saw a significant increase in mortality between 2018 and 2021. Middle aged patients had little overall improvement and demonstrated the steepest recent increase (APC 8.46%). Younger patients concluded the study period at rates similar to those seen in 1999. Men demonstrated a greater incidence of stroke related mortality compared to women (118.67 vs 104.15 in 2023) with similar trends over time. Racial disparities were present with Black patients showing the highest stroke related mortality in all years (from 242.47 to 156.26) and the largest recent increase (APC 7.37%). Regionally, the South and rural areas carried the highest mortality burden; however, all geographic areas demonstrated an identical increase between 2018 and 2021. Conclusions: Although stroke-related mortality among gastric cancer patients has declined over two decades, recent nationwide increases highlight ongoing vulnerability and persistent disparities. Targeted integration of cardiovascular risk assessment and stroke prevention into gastric cancer care may help sustain progress and reduce inequities. Group (Category A / Category B) AAMR 1999 (A / B) AAMR 2023 (A / B) Recent Surge Trend 2018–2021 (APC) Overall 167.3 111.4 +6.6%* Sex (Male / Female) 181.5 / 156.2 118.7 / 104.2 +6.8%* / +6.2%* Race (NH White / NH Black) 160.4 / 242.5 108.2 / 158.4 +6.3%* / +7.4%* Race (Hispanic / Asian) 140.3 / 153.6 97.7 / 80.0 +6.9%* / +5.2%* Age (Young / Middle) 6.7 / 47.7 6.7 / 42.3 +7.8%* / +8.5%* Age (Old 65+) 756.4 479.6 +6.2%* Region (Northeast / Midwest) 141.5 / 172.0 82.7 / 115.3 +4.4%* / +6.7%* Region (South / West) 177.5 / 171.4 125.7 / 107.2 +6.9%* / +6.2%* An asterisk (*) indicates that the APC is statistically significant (p < 0.05).

e13121 Background: Triple-negative breast cancer (TNBC) lacks reliable prognostic biomarkers beyond standard clinical staging, limiting individualized risk stratification and treatment planning. The neutrophil-to-lymphocyte ratio (NLR), a readily available marker of systemic inflammation, has been proposed as a prognostic indicator in TNBC; however, existing evidence remains inconsistent. We performed an updated systematic review and meta-analysis to evaluate the prognostic value of pretreatment NLR in the contemporary treatment era. Methods: PubMed, Embase, Cochrane CENTRAL, and ClinicalTrials.gov were systematically searched from database inception through January 1, 2026, for randomized and observational studies reporting associations between pretreatment NLR and survival outcomes in TNBC. Two reviewers independently screened studies and extracted data with discrepancies resolved by consensus. Outcomes included overall survival (OS), disease-free survival (DFS), and progression-free survival (PFS). Hazard ratios (HRs) were pooled using random-effects models. Risk of bias was assessed using ROBINS-I tool. Prespecified subgroup and sensitivity analyses were performed based on NLR cut-off values, follow-up duration, geographic region, and risk of bias. Results: Twenty-three retrospective studies comprising 4,731 patients with TNBC were included. On multivariate analysis, elevated pretreatment NLR was independently associated with poorer OS (HR 2.21, 95% CI: 1.56–3.13) and DFS (HR 1.97, 95% CI: 1.65–2.37. Similar associations were observed on univariate analysis (OS: HR 1.70, 95% CI: 1.07–2.72; DFS: HR 1.84, 95% CI: 1.14–2.98). No statistically significant or consistent association was observed between NLR and PFS. Associations between high NLR and adverse OS and DFS remained robust across prespecified subgroup and sensitivity analyses, representing the largest contemporary synthesis of multivariable-adjusted outcomes in TNBC. Key pooled outcome associations are summarized in Table 1. Conclusions: Elevated pretreatment NLR is independently associated with worse OS and DFS in patients with TNBC, supporting its potential role as a low-cost tool for clinical risk stratification and treatment planning. Given its availability from routine blood tests, NLR may assist in identifying high-risk patients who could benefit from intensified surveillance or therapeutic strategies. Prospective validation in the modern chemo-immunotherapy era is warranted. Summary of pooled associations between pretreatment NLR and survival outcomes in TNBC. Outcome Studies (n) Interpretation Overall survival 14 Worse survival Disease-free survival 15 Worse survival Progression-free survival 3 Inconclusive

8021 Background: Lung cancer is the leading cause of cancer-related mortality worldwide, primarily due to delayed diagnosis and diagnostic variability in histopathological assessment. Advances in deep learning applied to medical imaging offer the potential to enhance early detection accuracy, reduce interobserver variability, and improve diagnostic efficiency in lung cancer. Methods: In this retrospective study, we evaluated DAVINCI using a dataset of 8,962 patients with confirmed lung malignancies. The model utilizes a specialized architecture combining convolutional neural networks (CNNs) and residual blocks to optimize feature extraction from high-resolution pathological slides. Performance was measured via detection accuracy, recall, specificity, and Area Under the Curve (AUC). Results: A total of 8,962 patients were included in the analysis. The DAVINCI model correctly identified lung cancer in 93% of cases, corresponding to 8,334 of 8,962 patients, with a 95% confidence interval (CI) of 86%–100% (7,707–8,962 patients). Model specificity was also 93% (95% CI, 86%–100%), indicating a high true-negative rate. Sensitivity (recall) was 87%, with correct detection in 7,797 patients and a 95% CI of 80%–94% (7,170–8,425 patients). The model demonstrated strong discriminatory performance, achieving an area under the curve (AUC) of 0.91 (95% CI, 0.84–0.98). In comparison, active pathologists achieved a mean diagnostic accuracy of 83%, corresponding to 7,438 of 8,962 cases, with a 95% CI of 76%–90% (6,811–8,066 cases). In terms of efficiency, DAVINCI processed whole-slide images in 16–19 seconds, substantially faster than conventional manual review, which typically requires several minutes per slide. Conclusions: DAVINCI demonstrated superior accuracy, robust discriminatory performance, and markedly faster processing compared with human review, supporting its potential role in early lung cancer detection at scale.

7563 Background: Subcutaneous (SC) drug delivery options have become more widely available in multiple myeloma (MM). SC delivery of isatuximab (Isa), an anti-CD38 monoclonal antibody, has been investigated across clinical trials via a novel hands-free on-body injector (OBI). Isa SC OBI features a small, retractable needle, establishing the first device-enabled delivery of an oncologic treatment. The Phase 3 IRAKLIA, Phase 2 IZALCO, and Phase 1b (TCD15484) trials consistently demonstrated similar efficacy and safety of Isa SC OBI vs Isa IV in relapsed/refractory MM (RRMM) patients (pts). Here, we present a cross-trial analysis exploring safety and reliability of Isa SC OBI in RRMM pts. Methods: Pts with RRMM in IRAKLIA (N=263) and the Phase 1b expansion cohort (N=22) received 1400 mg Isa SC OBI plus pomalidomide and dexamethasone (d). In IZALCO, RRMM pts (N=64) received 1400 mg Isa SC OBI + carfilzomib (K)-d. Pts were monitored for 4 hrs (IRAKLIA) or 1-2 hrs (IZALCO) following first administration and for 1 hr for the following cycle (C) 1 administrations to assess local tolerability. Pre-infusion medications included montelukast (C1), steroids, acetaminophen and H1 antihistamines. Pts without systemic infusion-related reactions (IRRs) after 4 consecutive administrations of Isa SC OBI had subsequent pre-medication reassessed at investigator’s discretion. Safety and device performance explored in this cross-trial analysis of company sponsored studies were assessed in the Isa SC OBI cohorts by the duration of injection, IRRs, local injection-site reactions (ISRs), and injection/device success, which were defined similarly across all trials. Results: The median duration of Isa SC OBI injection was 13 mins in IRAKLIA, 12 mins in IZALCO, and 10 mins in the Phase 1b trial (6426 total injections, N=349 pts). IRRs across trials occurred in 6/6426 (0.09%) injections and 4/349 (1.15%) pts. IRRs were mostly grade 1 (G; 3/6426, 0.05%) with 1 G3 (0.02%); none leading to discontinuation. In IRAKLIA, 80/263 (30.4%) pts did not require premedication (931/5145 injections, 18.1%); no IRRs occurred in this group. Across trials, ISRs occurred in 21/349 (6.02%) pts resulting in 35 (0.54%) ISR events. Most ISRs were G1 (33, 0.51%), the remaining were G2 (2, 0.03%) and generally occurred on the day of injection. 99.9% (6421/6426) of injections in 98.6% (344/349) pts were successful (completed without interruption) across all trials. Conclusions: These findings support overall low, self-limiting rates of IRRs and ISRs across trials, with >6000 (99.9%) Isa SC OBI injections successfully delivered. Isa SC OBI shows reproducible safety and reliability in IRAKLIA, IZALCO and Phase 1b trials, supporting it as a novel administration option for MM pts and its potential for future exploration in at-home administration. Clinical trial information: NCT04045795 , NCT05405166 , and NCT05704049 .

7501 Background: AL amyloidosis is a plasma cell dyscrasia (PCD) characterized by κ or λ Ig light chain amyloid fibrils that deposit in vital organs. Anselamimab is an investigational mAb designed to complement anti-PCD therapy through accelerated removal of amyloid fibrils. Cardiac Amyloid Reaching for Extended Survival (CARES) studies (NCT04512235 and NCT04504825) investigated the efficacy and safety of anselamimab compared with placebo, in addition to anti-PCD therapy. Methods: Alongside anti-PCD therapy with CyBorD (daratumumab was permitted) newly diagnosed patients with European modification of Mayo stage IIIa or IIIb AL amyloidosis were randomized 2:1 to receive anselamimab (1000 mg/m 2 ) or placebo (normal saline) IV every 7 days for 4 weeks followed by every 14 days for the remainder of the blinded study (18 months after the last participant enrolled). The primary endpoint was a hierarchical combination of time to all-cause mortality (ACM) and frequency of adjudicated cardiovascular hospitalizations (CVH) using the Finkelstein-Schoenfeld test and win-ratio in the combined study population. Subgroup analysis by involved κ or λ free light chain (iFLC) was prespecified. Results: 271 patients were randomized to receive anselamimab and 135 placebo. Baseline characteristics were similar between treatment arms. The median duration of treatment was 21.4 months and 21.1 in the anselamimab and placebo arms, respectively. Daratumumab was used in 79.3% and 83.7% of the anselamimab and placebo arms, respectively. Results of the primary efficacy analyses are shown in the table. Anselamimab-treated patients with κ iFLC had a 62% reduction in ACM (HR = 0.38; 95% CI, 0.17, 0.86; nominal P = 0.012) and a 71% reduction in risk for CVH (estimated CVH rate ratio = 0.29; 95% CI, 0.10, 0.87; nominal P = 0.028). Patients with λ iFLC had a 10% reduction in risk for CVH; none in ACM. Safety data were comparable between both treatment arms. In anselamimab-treated patients, safety findings were similar between κ subgroup and full population. Conclusions: Anselamimab, a κ light chain–directed anti-fibril monoclonal antibody used alongside standard of care chemotherapy, offers a first-in-class, new therapeutic option for patients with newly diagnosed κ AL amyloidosis. Clinical trial information: NCT04512235 and NCT04504825 . Summary of primary efficacy endpoints. Endpoint Overall κ Isotype λ Isotype Anselamimab (N=271) Placebo (N=135) P -value Anselamimab (N=48) Placebo (N=24) P -value Anselamimab (N=219) Placebo (N=109) P -value ACM and CVH win ratio (95% CI) 1.1 (0.8, 1.5) 0.332 2.06 (0.98, 4.31) 0.100 0.9 (0.6, 1.3) 0.848 ACM n (%) 90 (33.2) 52 (38.5) 15 (31.3) 14 (58.3) 73 (33.3) 37 (33.9) HR (95% CI) 0.8 (0.6, 1.1) 0.290 0.38 (0.2, 0.9) 0.012 1.0 (0.7, 1.5) 0.647 CVH Frequency/year (95% CI) 0.6 (0.4, 0.8) 0.9 (0.6, 1.4) 0.145 0.41 (0.2, 0.8) 1.4 (0.6, 3.4) 0.028 0.7 (0.4, 1.0) 0.8 (0.4, 1.3) 0.664 Incidence risk ratio (95% CI) 0.7 (0.4, 1.2) 0.29 (0.1, 0.9) 0.9 (0.5, 1.7)