Correction: Sex-specific outcomes in myocardial infarction: a dual-cohort analysis using clinical and real-world data

Background: Heart failure with mildly reduced ejection fraction (HFmrEF), defined by a left ventricular ejection fraction (LVEF) of 41–49%, is a growing clinical entity. However, limited real-world data exist regarding sex-specific outcomes in this subgroup. This study aimed to evaluate differences in cardiovascular outcomes between male and female patients with HFmrEF. Methods: We conducted a retrospective cohort study using de-identified electronic health records from the TriNetX US Collaborative Network, comprising data from 68 healthcare organizations. Adults aged 18–90 years with a diagnosis of heart failure (ICD-10-CM: I50) and a documented left ventricular ejection fraction (LVEF) between 41–49% were included. Patients with prior heart transplant, mechanical circulatory support, artificial heart, end-stage renal disease, or neoplasms were excluded. Female (n=15,169) and male (n=26,658) cohorts were identified and matched 1:1 using propensity scores, yielding 12,783 patients per group. Matching was based on demographics, comorbidities, and cardiovascular medication use. The index event was defined as the first recorded HF diagnosis with LVEF 41–49%, and outcomes were assessed starting one day after the index date until final follow-up. Primary outcomes included all-cause mortality, cardiac arrest, acute myocardial infarction (MI), cerebral infarction, and acute kidney injury (AKI), analyzed using Kaplan-Meier survival analysis and risk-based comparisons. Natriuretic peptide B (NPB) levels were also compared between groups. Results: After matching, females had a higher hazard for overall mortality (HR 1.051, 95% CI: 0.994–1.111, p =0.048) with lower survival probability (44.2% vs. 22.3%). Females also had a higher risk and hazard of acute MI (HR 1.092, 95% CI: 1.022–1.165, p =0.008). No significant differences were observed for cardiac arrest (HR 0.917, p =0.604), cerebral infarction (HR 1.020, p =0.665), or AKI (HR 1.020, p =0.476). NPB levels were slightly higher in females (mean: 1533.5 vs. 1355.6 pg/mL), but the difference was not statistically significant ( p =0.192). Conclusion: In this large, real-world analysis, female patients with HFmrEF had a significantly higher hazard of mortality and acute MI compared to males, despite matching for baseline characteristics. These findings highlight the need for sex-specific risk stratification and tailored management strategies in HFmrEF.

BackgroundRandomized controlled trials (RCTs) have stringent inclusion criteria and may not fully represent patients seen in everyday clinical practice. Real-world data (RWD) can provide supportive evidence for the effectiveness of medical interventions in more heterogeneous populations than RCTs. Sunitinib is a widely used first-line treatment for patients with metastatic renal cell carcinoma (mRCC).ObjectiveThis is the first comprehensive meta-analysis to evaluate the efficacy of sunitinib using the novel approach of combining RCTs and RWD.MethodsRCTs and RWD studies published between 2000 and 2017 were identified from PubMed, Ovid, MEDLINE, and EMBASE. Eligible studies contained a cohort of ≥ 50 adult patients with mRCC receiving first-line sunitinib treatment. The meta-analysis combined RWD and RCT treatment groups, adjusting for data type (RCT or RWD). Recorded outcomes were median progression-free survival (mPFS), median overall survival (mOS), and objective response rate (ORR). Publication bias was assessed via review of funnel plots for each outcome measure. A random effects model to account for study heterogeneity was applied to each endpoint. Sensitivity analyses evaluated the robustness of the overall estimates.ResultsOf the 3611 studies identified through medical database searches, 22 (15 RWD studies, 7 RCTs) met eligibility criteria and were analyzed. mPFS (18 studies), mOS (19 studies), and ORR (15 studies) were reported for aggregate measures based on 4815, 5321, and 4183 patients, respectively. Reported mPFS (RWD, 7.5–11.0 months; RCTs, 5.6–15.1 months) and ORR data (RWD, 14.0–34.6%; RCTs, 18.8–46.9%) were consistent with the overall confidence estimates (95% confidence interval [CI]) of 9.3 (8.6–10.2) months and 27.9% (24.2–32.0), respectively. Reported mOS showed greater variation in RWD (6.8–33.2 months) compared with RCTs (21.8–31.5 months), with an overall confidence estimate (95% CI) of 23.0 (19.2–27.6) months. Inspection of funnel plots and sensitivity analyses indicated that there was no publication bias for any efficacy endpoint. Sensitivity analyses showed no evidence of lack of robustness for mPFS, mOS, or ORR. Interpretation of these results is limited by differences in trial design, cohort characteristics, and missing data.ConclusionsThis novel, comprehensive meta-analysis validates sunitinib as an effective first-line treatment for patients with mRCC in both RCTs and everyday clinical practice. The methodology provides a framework for future analyses combining data from RCTs and RWD.Electronic supplementary materialThe online version of this article (10.1007/s11523-019-00653-5) contains supplementary material, which is available to authorized users.

The landscape for evidence generation in hematologic malignancies is rapidly evolving. While randomized controlled trials (RCTs) remain the gold standard in support of drug efficacy, approval and use, the supplemental use of real-world data (RWD), generated as part of routine healthcare delivery, and real-world evidence (RWE), the insights derived from RWD, in this setting has become increasingly common. There is a wide variety of sources of RWD, each with its own strengths and weaknesses that need to be considered when determining its appropriate use in RWE generation. RWD and RWE have historically been utilized in the post-approval setting to assess real-world application, efficacy, and safety of approved therapies. However, due to increasing awareness of the advantages of additional sources of information, RWE sourced from clinical data are being increasingly used to provide context for regulatory decision-making across several diseases including hematologic malignancies. Today, many commercial vendors offer fully aggregated, de-identified and standardized real-world clinical data. To maximize the potential of RWD and RWE, important considerations are needed to ensure patient privacy and to reduce the potential for biases and residual confounding. Continued collaboration among researchers, regulators and industry partners are needed to optimize evidence generation to ensure that new therapies reach patients as quickly and safely as possible.

Non-vitamin K oral anticoagulants (NOACs) are increasingly used as alternatives to conventional therapies and have considerable accumulated real-world clinical data in patients with non-valvular atrial fibrillation (NVAF) or venous thromboembolism (VTE). However, it is not easy to make a complete changeover to NOACs in real-world clinical practice because NOACs still have challenges in specific patient populations (eg, Asian patients, NVAF patients presenting with acute coronary syndrome [ACS], dialysis patients with NVAF, patients with cancer-associated VTE, etc.). Clinical data on the optimal dose of NOACs in Asian patients with NVAF are not sufficient. The intensity of NOAC and antiplatelet treatment and the duration of antiplatelet treatment should be adjusted according to the bleeding and thrombotic risk profiles of the individual NVAF patient presenting with ACS. Increased bleeding risk and unclear efficacy of NOACs in dialysis patients with NVAF should be considered when making decisions on whether to give NOACs for these patients. If dialysis patients with NVAF require anticoagulant for stroke prevention, then apixaban could be considered while awaiting more clinical efficacy and safety data. Additional studies are needed to determine the utility of continuing treatment with reduced-dose NOACs for long-term therapy after VTE. We have enough experiences in using NOACs in cancer patients showing the benefit of antithrombotic treatment counterbalanced the bleeding risk; however, some challenges of cancer-associated VTE management exist due to differences in cancer types or chemotherapy regimens and comorbidities. Different dosing regimens among NOACs may impact on medication adherence; thus, individual patient preference should be considered in choosing a particular NOAC. A significant proportion of patients remain on warfarin because of the high price of NOACs and variability in reimbursement coverage. To compensate clinical-evidence and achieve optimal use of NOACs, we should pay attention to the outcomes of ongoing studies and evaluate more real-world data.

Background: The Platelet Inhibition and Patient Outcomes (PLATO) trial showed that ticagrelor improved clinical outcomes in patients (pts) with acute coronary syndrome compared with clopidogrel. There were few real world data on the efficacy and safety of ticagrelor in East Asians in whom bleeding is a major concern with potent antithrombotic drugs. Methods: We conducted a nationwide population-based cohort study by using 2012 to 2014 Taiwan National Health Insurance Research Database. The 18-month clinical outcomes in pts with acute myocardial infarction (MI) treated with ticagrelor vs. clopidogrel were compared. Overall, we identified 27,339 acute MI pts (mean age 62.2 years, 71% male) who survived 30 days after the acute MI attack and took dual antiplatelet agents. Cohorts of ticagrelor and clopidogrel were 1:8 matched based on propensity score. The primary efficacy endpoints were all-cause death, recurrent MI or ischemic stroke. The safety endpoints consisted of major gastrointestinal (GI) bleeding and intracerebral hemorrhage (ICH). Results: In unmatched cohorts, ticagrelor-treated pts (n = 2,844) were younger with fewer comorbidities than clopidogrel-treated pts (n = 24,495). The 18-month primary efficacy endpoint rate was lower in ticagrelor group (10.1 vs. 18.3%, p < 0.001). In ticagrelor group, the risks of ICH (0.3%) and ICH + GI bleeding (2.9%) were similar to the data reported in the PLATO trial (ICH 0.3% and non-CABG related TIMI major bleeding 2.8%). After propensity matching, we obtained 2,389 ticagrelor users and 19,112 clopidogrel users in the cohort. After adjusting for clinical characteristics, the 18-month primary efficacy endpoint rate was 21% lower in ticagrelor than in clopidogrel group (10.5 vs. 16.0%, adjusted HR, 0.79; 95% CI, 0.68-0.93). The composite safety endpoint rate was not different between ticagrelor and clopidogrel group (3.2 vs. 4.1%, adjusted HR, 1.01; 95% CI, 0.74-1.37). Subgroup analyses show consistent benefits of ticagrelor in old pts (≥ 75 years) and in pts with diabetes, heart failure or previous MI. Conclusions: In the real world acute MI Taiwanese pts, ticagrelor offered better cardiovascular protection than clopidogrel without increasing the major bleeding risk.

Mortality rate differences observed between men and women presenting with acute myocardial infarction (AMI) have been the focus of clinical trials and registry analyses over the past two decades. Despite advances in cardiovascular therapies over this time, studies have shown that women presenting with AMI have increased early mortality compared to their male counterparts. The cause of mortality discrepancy after AMI in women has been the subject of much debate and has become one of many areas of interest regarding sex-specific outcomes in cardiovascular disease (CVD). Research suggests potential baseline risk profile differences and treatment biases among women compared to men as a reason for the AMI mortality gap. While the exploration of gender differences in coronary artery disease (CAD) has increased, the low proportion of female patients included in CVD trials as well as the lack of published studies results sorted by sex, leaves this an area in need of further study. The current issue of EuroIntervention includes two papers addressing this topic. Both studies assessed different European populations for gender-related differences in mortality after STsegment elevation myocardial infarction (STEMI) and come to the conclusions that female gender is an independent predictor of inhospital mortality following AMI. Benamer et al analysed 16,760 patients (3,664 women) treated by PCI for STEMI from 2003 to 2007 in the CARDIO-ARHIF registry and determined that women tended to be older, had more diabetes, were more likely to present in cardiogenic shock and had a lower success rate of PCI when compared to men. After multivariate logistic regression adjusted for these risks, the authors concluded female gender was associated with higher in-hospital mortality (OR 1.38 [1.16-1.63], p=0.0002) but only after age of 75. Sadowski and colleagues analysed 26,035 consecutive patients (8,989 women) with STEMI enrolled in the Polish Registry of Acute Coronary Syndromes (PL-ACS). Like Benamer’s study, Sadowski et al determined these women were older with higher incidences of comorbidities such as hypertension, diabetes, obesity (BMI>30kg/m) and cardiogenic shock at presentation compared to males, and found a higher in-hospital mortality (1.13 [1.01-1.26], p<0.03), but not 12-month mortality (1.02 [0.96-1.09], p=NS) after multivariate adjustment. Given the data presented from these two studies, three important issues need further discussion. First, CVD is the most common cause of death among women, yet women remain poorly represented in randomised clinical trials investigating cardiovascular disease and treatment. As a society we have enjoyed a steady reduction in age-adjusted mortality for CVD, but only recently has the rate of mortality decline for women approached that of men. Reasons for this delay may include less frequent use of revascularisation procedures or evidence-based therapies in female patients with CVD. Others postulate an overall lack of awareness of the prevalence of CVD in women, on the part of patients, policy makers and healthcare professionals. A 2005 online survey of primary care physicians, gynaecologists and cardiologists found that fewer than 20% of physicians were aware that more women die annually of CAD than men – a finding that highlights the need for awareness campaigns among healthcare providers. There is a paucity of well-designed clinical trials assessing outcomes in women with cardiovascular disease and as a result, women have been under-represented in CVD clinical trials. In early studies investigating therapies employed in the treatment of AMI, fewer than 20% of subjects were women. Reasons for this

Female gender is reported as an independent risk factor for a poor outcome after coronary artery bypass grafting. We analysed the influence of gender on surgical outcome in patients with single-vessel disease undergoing minimally invasive direct coronary artery bypass (MIDCAB). From January 1998 to December 2016, a total of 607 consecutive patients with single-vessel disease (31.9% women) underwent MIDCAB at our institution. Major adverse cardiac and cerebrovascular events (MACCE) were recorded during a median follow-up period of 8.0 years. Survival time was estimated for all patients and after applying propensity score matching (138 women vs 138 men). Multivariable Cox regression analysis identified risk factors predicting a long-term mortality rate. In women, a longer surgical time (125 vs 113 min; P < 0.001) and a higher transfusion rate were recorded (13.0% vs 5.1%; P = 0.001) with similar rates of in-hospital deaths (1.0% vs 0.5%; P = 0.60) and MACCE (1.5% vs 0.7%; P = 0.39). Survival and MACCE-free survival during the follow-up period did not differ significantly between genders (P = 1.0, P = 0.36). Survival and MACCE-free survival rates after 5 years were 94% and 90% in women compared to 91% and 86% in men. Propensity score matching demonstrated improved long-term survival rates in women (P = 0.029). Insulin-dependent diabetes mellitus, postoperative atrial fibrillation and prolonged intensive care unit stay were associated with long-term deaths for both genders, whereas obesity, former myocardial infarction and preoperative atrial fibrillation were significant risk factors in men. Female patients showed no adverse outcomes after MIDCAB, although risk factors were gender-specific. Overall, MIDCAB demonstrated excellent short- and long-term results as a treatment for single-vessel disease in both genders.

BackgroundIn-stent restenosis (ISR) remains a challenging complication following percutaneous coronary intervention, and intravascular brachytherapy (IVBT) has proved to be an important treatment strategy. However, limited data exist on sex-specific outcomes following IVBT.MethodsThis retrospective, single-center cohort study included 223 patients (61 women, 162 men) treated with IVBT for ISR between 2014 and 2023. The primary end points were all-cause mortality, target lesion revascularization, and major adverse cardiovascular events. Secondary outcomes included technical success, myocardial infarction, cardiac death, and heart failure hospitalization. Multivariable Cox regression was used to adjust for clinical and procedural covariates.ResultsBaseline characteristics were largely similar between sexes, except for higher body surface area and diabetes prevalence in men. Procedural success rates did not differ between groups. However, female sex was independently associated with a higher risk of target lesion revascularization (adjusted HR, 1.80; 95% CI, 1.07-3.01; P = .026) and major adverse cardiovascular events (adjusted HR, 1.63; 95% CI, 1.09-2.45; P = .017). Women also had a higher risk of myocardial infarction (adjusted HR, 2.58; 95% CI, 1.29-5.19; P = .008), whereas no significant sex-based differences were observed for all-cause mortality or heart failure hospitalization.ConclusionsDespite comparable procedural outcomes, women undergoing IVBT for ISR experienced higher rates of adverse cardiovascular events. These findings underscore the need for sex-stratified risk assessment and further prospective research to understand and address sex-based differences in outcomes after IVBT.

BackgroundIn-hospital mortality in patients with ST-segment elevation myocardial infarction (STEMI) is higher in those with COVID-19 than in those without COVID-19. The factors that predispose to this mortality rate and their relative contribution are poorly understood. This study developed a risk score inclusive of clinical variables to predict in-hospital mortality in patients with COVID-19 and STEMI.MethodsBaseline demographic, clinical, and procedural data from patients in the North American COVID-19 Myocardial Infarction registry were extracted. Univariable logistic regression was performed using candidate predictor variables, and multivariable logistic regression was performed using backward stepwise selection to identify independent predictors of in-hospital mortality. Independent predictors were assigned a weighted integer, with the sum of the integers yielding the total risk score for each patient.ResultsIn-hospital mortality occurred in 118 of 425 (28%) patients. Eight variables present at the time of STEMI diagnosis (respiratory rate of >35 breaths/min, cardiogenic shock, oxygen saturation of <93%, age of >55 ​years, infiltrates on chest x-ray, kidney disease, diabetes, and dyspnea) were assigned a weighted integer. In-hospital mortality increased exponentially with increasing integer risk score (Cochran-Armitage χ2, P ​< ​.001), and the model demonstrated good discriminative power (c-statistic ​= ​0.81) and calibration (Hosmer-Lemeshow, P ​= ​.40). The increasing risk score was strongly associated with in-hospital mortality (3.6%-60% mortality for low-risk and very high–risk score categories, respectively).ConclusionsThe risk of in-hospital mortality in patients with COVID-19 and STEMI can be accurately predicted and discriminated using readily available clinical information.

Non-vitamin K antagonist oral anticoagulants (NOACs), compared with warfarin, have a favorable risk-benefit profile. However, in the RE-LY study in patients with atrial fibrillation (AF), the number of patients with MI was higher in the dabigatran group as compared to the warfarin group. Many meta-analyses showed that dabigatran treatment led to an increased risk of myocardial infarction (MI). Large real-world data (RWD) did not confirm an increase in the risk of MI during dabigatran treatment. In our meta-analysis we excluded RWD, and each of the four drugs was evaluated in two key-phase III randomized controlled trials: in patients with AF and patients with AF and chronic coronary syndrome or acute coronary syndrome treated with percutaneous coronary interventions. In each study, warfarin was the comparator for NOACs. In this homogeneous group of patients, dabigatran, in direct comparison with warfarin, significantly increased the risk of MI by about 30%. Moreover, the risk of MI was also significantly higher than the opposite effect of activated factor (F) X inhibitors (FXa inhibitors) vs. warfarin. In our network meta-analysis, considering individual NOACs in recommended doses, we found an increased risk of MI compared to warfarin only in patients treated with dabigatran 150 mg twice a day and, in particular, 110 mg twice a day. In this review we present evidence supporting our opinion that in patients with AF and coronary stenting, the choice of NOACs (direct FXa vs. thrombin inhibitors) is equally as important as choosing the optimal antiplatelet therapy (single or dual antiplatelet therapy).

Background Survival after prostate cancer (PCa) is high for many patients, and it is now more likely they will die from cardiovascular disease than from PCa specific causes. However, patients with active malignancies have been excluded from trials to evaluate best treatments for myocardial infarction (MI). As people with PCa become a larger subpopulation of those experiencing a MI, clinicians need more comprehensive evidence regarding MI in people with PCa and if they are receiving optimal treatment. Purpose To identify the Norwegian population diagnosed with PCa that have experienced a MI after their cancer diagnosis using routinely collected registry data, and describe relevant patient and treatment characteristics, as well as identify areas of missing data. Methods We used data from Cancer Registry Norway to identify all people diagnosed with PCa between 2004 and 2019. Cancer data was linked with the Norwegian Myocardial Infarction Registry to identify PCa patients that were hospitalised for a MI as well as important features of the MI treatment period. Descriptive statistics were used to describe the number of people with PCa, important cancer specific features and variables related to the MI, as well as to explore missing data within collected variables. Results We identified 70646 people diagnosed with PCa in Norway between 2004 and 2019, of which, 2929 (4.1%) experienced a MI (3465 MIs in total). Most had a single MI (2528, 86.3%), around 10.7% (314) had two MIs, and 3.0% (87) had three or more. Of all the MIs, 20.7% (718 in 700 patients) were classified as ST elevated MIs and 76.2% (2641 in 2218 patients) as non-ST elevated MIs (Table 1). Around 9% of patients were missing a baseline prostate specific antigen value and over 20% did not have an identified cancer stage. Thrombolysis and percutaneous intervention had a high number of records with unknown status of these treatments (58.5% and 36.6%, respectively). Conclusions This population-wide registry-based study has demonstrated that linkage between Cancer Registry Norway and Norwegian Myocardial Infarction Registry provides a comprehensive dataset to further study the relationship between PCa and MI in the Norwegian population. This study benefits from data sourced from registries that have previously been found to be highly comprehensive and accurate. There was evidence of some variables with higher missing data, but we are confident that our planned further linkages with the Norwegian Cardiovascular Disease Registry and Norwegian Patient Registry, as well as using other clinical data, will help to overcome the current missing data issues. It is our plan to carry this research forward to provide a comprehensive, longitudinal analysis of how patients with PCa experience and are treated for an acute MI in Norway including PCa specific risk factors, comparison with current guidelines and if further guidelines should be created for people with PCa. Funding Acknowledgement Type of funding sources: Foundation. Main funding source(s): Norwegian Cancer SocietyNorwegian Institute of Public Health

<sec> <title>UNSTRUCTURED</title> The amount of data generated in clinical practice is increasing substantially, which has benefitted the use of real-world data (RWD) for real-world evidence (RWE) in biomedical research. While early RWE efforts focused on structured EHR and claims data, advances in analytical methods, such as artificial intelligence, are expected to further enhance the ability to obtain deeper insights from RWD. Most recently, real-world imaging data (RWiD) has emerged as a novel and valuable resource, enabled by improvements in imaging infrastructure, data standardization, and de-identification technologies. Medical imaging is essential at multiple stages of clinical care, ranging from screening to post treatment assessment and surveillance. Medical imaging has become a key component of patient management, as it augments clinical decision-making across many medical specialties. RWiD is the retrospective collection of routinely gathered clinical imaging data. Using only radiology reports results in limited information compared to datasets that contain the actual images. Radiology reports primarily focus on clinical decision making rather than research purposes. Therefore, the actual images add value to real world datasets. However, using RWiD is challenging due to complex de-identification and harmonization, as well as requirements for file storage, file transfer, and computation. This article describes the background, challenges, and opportunities of real world imaging data with a focus on life sciences and biopharmaceutical applications. </sec>

Accumulating studies reported that several phosphodiesterases (PDEs) inhibitors might have cardiovascular benefits. This study aimed to explore the relationship between genetically-predicted PDEs and ischaemia heart disease via drug target Mendelian randomization (MR) approach, and then examine the effect of inhibitors of identified target on the outcomes by using real-world data. In the two-sample MR study, the expression of genes encoding PDEs was used to proxy the level of PDEs and available expression quantitative trait loci (eQTLs) for each target gene were identified as the genetic instruments. The outcomes included coronary heart disease (CHD) and myocardial infarction (MI). In the real-world study, a retrospective cohort was conducted to compare the incidence of outcomes between PDE5 inhibitors and alprostadil use by linking Swedish nationwide registers. MR analyses identified two types of PDEs, PDE5, and PDE8, genetically-predicted expression in blood of the encoded genes was significantly associated with the risk of CHD [odds ratio (OR)PDE5A=1.22,95% confidence interval (CI)=1.06-1.40; ORPDE8A=1.26,95% CI=1.07-1.49] and MI (ORPDE5A=1.27,95% CI=1.09-1.48; ORPDE8A=1.24,95% CI=1.04-1.48). Notably, the highest expression of PDE5A was observed in artery aorta, which was also positively related to CHD (OR=1.17,95% CI=1.05-1.32) and MI (OR=1.15,95% CI=1.02-1.30). Real-world study provided supportive evidence that as compared to alprostadil use, PDE5 inhibitors use significantly reduced the incidence of CHD (adjusted HR=0.70,95% CI=0.66-0.73) and MI (adjusted HR=0.79,95% CI=0.73-0.84). This study provided observational and genetic evidence about the protective role of PDE5 inhibition against ischaemic heart disease, indicating the potential of these drugs to be repurposed for ischemia heart disease prevention and treatment.

BackgroundObservational studies using large claims databases for diabetes patients have been increasingly conducted. While validation of outcomes is important in such studies, validation studies from Japan are still scarce and small in scale with questions remaining on the representativeness of their findings. We examined the positive predictive value (PPV) of outcomes that often develop in type 2 diabetes patients: cardiovascular outcomes including congestive heart failure (CHF), myocardial infarction (MI), stroke-related diseases, and renal outcomes including end stage renal disease (ESRD), and death using a large Japanese database containing administrative claims and electronic medical record (EMR) data.Patients and MethodsWe used patient-level administrative claims data from 2003 and EMR data from 1985 to the most recent data up to December 2018 provided by Real World Data Co., Ltd. The database consisted of data from over 200 hospitals including ≥12 million uniquely identifiable patients. Among patients who had ≥1 type 2 diabetes diagnosis in the EMR, those who had administrative claims for each outcome were identified, and then the PPV was calculated for each outcome using the EMR as the gold standard.ResultsThe numbers of patients identified for each outcome were 1,700 for MI, 2,027 for hemorrhagic stroke, 3,722 for ESRD, 4,723 for ischemic stroke, 5,404 for CHF, 6,678 for any type of stroke, and 10,815 for death. PPVs ranged from 67.4% for ESRD, 78.7% for MI, 80.3% for death, 85.7% for ischemic stroke, 88.9% for any type of stroke, 89.9% for hemorrhagic stroke, and 95.7% for CHF. A post hoc analysis showed PPV for ESRD as 83.8%.ConclusionThis large-scale validation study on diagnosis in administrative claims showed reasonable PPVs for the outcomes. We believe that the definitions of outcomes can be considered to be appropriate for future studies using Japanese administrative claims data.

Machine learning (ML) could improve risk stratification following myocardial infarction (MI). However, previous ML studies for risk prediction following MI did not incorporate comprehensive echocardiographic data. This study sought to use machine learning (ML) to integrate comprehensive echocardiographic and clinical data for the predicting all-cause mortality following MI. Retrospective study of consecutive patients admitted with MI to a tertiary referral hospital, with echocardiography performed within 24 h of admission. The cohort was randomly split into training (70%) and test (30%) sets. Two ML models (XGBoost and a neural network) were developed using echocardiographic and clinical data, and then compared with conventional logistic regression. The Shapley Additive exPlanations method was used for ML model interpretation. In the final study population of 3202 patients (mean age, 63.2 ± 12.5 years; 29.2% females), ST-elevation MI was present in 28.8%, and the mean cohort LVEF was 52.5 ± 11.2%. At a median follow-up of 4.5 years, there were 465 deaths. In the test set, XGBoost achieved the highest performance (AUC, 0.854), compared with logistic regression (AUC, 0.824; P = 0.006) and the neural network (AUC, 0.808; P = <0.001) for the prediction of death. In the XGBoost model, the highest-ranked predictors included age, renal function, echocardiographic left ventricular outflow velocity time integral, and diastolic parameters. Further, in nested ML models, the addition of echocardiographic parameters provided incremental value beyond clinical variables alone (AUC, 0.854 vs. 0.820; P = 0.002). ML integration of comprehensive echocardiographic data with clinical data could lead to improved prediction of survival following MI. Clinical implementation should be considered.