Machine-Learning-Based In-Hospital Mortality Prediction for Transcatheter Mitral Valve Repair in the United States.
Ontology highlight
ABSTRACT: Background
Transcatheter mitral valve repair (TMVR) utilization has increased significantly in the United States over the last years. Yet, a risk-prediction tool for adverse events has not been developed. We aimed to generate a machine-learning-based algorithm to predict in-hospital mortality after TMVR.Methods
Patients who underwent TMVR from 2012 through 2015 were identified using the National Inpatient Sample database. The study population was randomly divided into a training set (n = 636) and a testing set (n = 213). Prediction models for in-hospital mortality were obtained using five supervised machine-learning classifiers.Results
A total of 849 TMVRs were analyzed in our study. The overall in-hospital mortality was 3.1%. A naïve Bayes (NB) model had the best discrimination for fifteen variables, with an area under the receiver-operating curve (AUC) of 0.83 (95% CI, 0.80-0.87), compared to 0.77 for logistic regression (95% CI, 0.58-0.95), 0.73 for an artificial neural network (95% CI, 0.55-0.91), and 0.67 for both a random forest and a support-vector machine (95% CI, 0.47-0.87). History of coronary artery disease, of chronic kidney disease, and smoking were the three most significant predictors of in-hospital mortality.Conclusions
We developed a robust machine-learning-derived model to predict in-hospital mortality in patients undergoing TMVR. This model is promising for decision-making and deserves further clinical validation.
SUBMITTER: Hernandez-Suarez DF
PROVIDER: S-EPMC7736498 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:ObjectivesThis study sought to develop and compare an array of machine learning methods to predict in-hospital mortality after transcatheter aortic valve replacement (TAVR) in the United States.BackgroundExisting risk prediction tools for in-hospital complications in patients undergoing TAVR have been designed using statistical modeling approaches and have certain limitations.MethodsPatient data were obtained from the National Inpatient Sample database from 2012 to 2015. The data were randomly divided into a development cohort (n = 7,615) and a validation cohort (n = 3,268). Logistic regression, artificial neural network, naive Bayes, and random forest machine learning algorithms were applied to obtain in-hospital mortality prediction models.ResultsA total of 10,883 TAVRs were analyzed in our study. The overall in-hospital mortality was 3.6%. Overall, prediction models' performance measured by area under the curve were good (>0.80). The best model was obtained by logistic regression (area under the curve: 0.92; 95% confidence interval: 0.89 to 0.95). Most obtained models plateaued after introducing 10 variables. Acute kidney injury was the main predictor of in-hospital mortality ranked with the highest mean importance in all the models. The National Inpatient Sample TAVR score showed the best discrimination among available TAVR prediction scores.ConclusionsMachine learning methods can generate robust models to predict in-hospital mortality for TAVR. The National Inpatient Sample TAVR score should be considered for prognosis and shared decision making in TAVR patients.
| S-EPMC6650265 | biostudies-literature
Project description:BackgroundClinical trials have demonstrated health status benefit of transcatheter mitral valve repair (TMVr) with MitraClip in patients with mitral valve regurgitation. Real-world site-level variability in health status outcomes for TMVr, and factors associated with this variability, are unknown.MethodsAll patients undergoing TMVr procedure with MitraClip between November 2013 and March 2019 in the Transcatheter Valve Therapy Registry were included. Health status was measured at baseline and 30 days with the Kansas City Cardiomyopathy Questionnaire (KCCQ) Overall Summary (OS) score. Site-level variability in 30-day change in KCCQ-OS was examined by calculating the median odds ratio from a hierarchical logistic regression model, with ≥20-point improvement as the dependent variable. To define the extent to which patient characteristics, procedural characteristics (residual mitral valve regurgitation, periprocedural bleeding), site volume, and patients' baseline health status accounted for variability in outcomes, the proportion of variability (R2) explained by sequentially adding these variables to the model was quantified.ResultsAcross 339 sites, 12 415 patients (mean age 79.0±9.5 years, 46.1%. females, 89.5% White) completed baseline and 30-day health status assessments. Mean KCCQ-OS score was 43.0±24.4 at baseline and 67.0±24.9 at 30-day follow-up. Across sites, the proportion of patients achieving a ≥20-point improvement in KCCQ-OS ranged from 12.5% to 100% and the adjusted median odds ratio was 1.58 (95% CI, 1.46-1.69). The greatest contribution to the variability in health status outcomes was from patients' baseline KCCQ-OS score (R2=25%) with <1% of the variability explained by patient and procedural characteristics, and annual site volume.ConclusionsThere is moderate variation across sites in their patients' achievement of health status benefits from TMVr, with patient's baseline health status accounting for the largest proportion of this variation. This underscores the importance of patient selection in supporting more consistent health status benefit from TMVr.
| S-EPMC7738373 | biostudies-literature
Project description:This review outlines the first trial experience with transcatheter therapy for mitral regurgitation (MR), developed from the EVEREST II MitraClip trial in a trial population comprised predominantly of patients with degenerative mitral regurgitation (DMR). Subsequent experience with MitraClip and several other devices has been mostly in functional MR patients. At the same time, there has been ongoing experience with MitraClip in DMR, and a variety of other devices have been developed for catheter-based treatment of MR. Annuloplasty devices have been indicated for DMR, and the potential for transcatheter annuloplasty to be used, in conjunction with other catheter techniques, such as chordal replacement, as it is in standard mitral repair, is developing. Transcatheter mitral valve replacement will clearly have some role for MR of both functional and degenerative etiologies, when repair is not feasible or fails. This review will discuss the evidence base and future development of these mitral repair and replacement approaches for DMR.
| S-EPMC6288224 | biostudies-other
Project description:BACKGROUND:Mitral regurgitation is a heterogeneous disease. Determining which patients derive optimal outcomes from transcatheter edge-to-edge mitral valve repair (TMVR) remains challenging. We sought to determine whether baseline mitral valve anatomic characteristics are predictive of left atrial pressure (LAP) changes during TMVR with MitraClip. METHODS AND RESULTS:Consecutive patients with severe mitral regurgitation undergoing TMVR (n=112) underwent continuous intraprocedural LAP monitoring and retrospective echocardiographic analysis for specific mitral anatomic characteristics. Procedural success (optimal LAP reduction) was defined as ?40% reduction in left atrial V-wave pressure compared with baseline. Echocardiographic predictors of optimal LAP reduction and increased postprocedure mean diastolic gradient were evaluated. Mean age was 79±14 years, and 36 patients (32%) were women. Primary, mixed, and secondary mitral regurgitation were present in 78 patients (70%), 22 patients (20%), and 12 patients (10%), respectively. Baseline mean LAP and V-wave were 22±6 and 38±13 mm Hg; after TMVR, these decreased to 19±5 and 27±10 mm Hg, respectively (P<0.0001 for both). Independent predictors of optimal LAP reduction were the presence of a flail scallop, mitral regurgitation localized to a single scallop, and high-quality 3-dimensional echocardiographic imaging. Independent predictors of elevated postprocedure mean diastolic gradient were elevated preprocedure mean diastolic gradient, mitral annular calcification, and implantation of multiple clips. CONCLUSIONS:Mitral valve pathoanatomic features, including a flail leaflet and single jet, are predictive of optimal LAP reduction with TMVR. High-quality 3-dimensional imaging may help select patients with the highest likelihood of optimal hemodynamic results with TMVR. These data expand current knowledge about patient selection for TMVR and deserve further study in larger cohorts.
| S-EPMC5850157 | biostudies-literature
Project description:Mitral valve regurgitation (MR) is the most common valvular heart disease and current variables associated with MR recurrence are still controversial. We aim to develop a machine learning-based prognostic model to predict causes of mitral valve (MV) repair failure and MR recurrence. 1000 patients who underwent MV repair at our institution between 2008 and 2018 were enrolled. Patients were followed longitudinally for up to three years. Clinical and echocardiographic data were included in the analysis. Endpoints were MV repair surgical failure with consequent MV replacement or moderate/severe MR (>2+) recurrence at one-month and moderate/severe MR recurrence after three years. 817 patients (DS1) had an echocardiographic examination at one-month while 295 (DS2) also had one at three years. Data were randomly divided into training (DS1: n = 654; DS2: n = 206) and validation (DS1: n = 164; DS2 n = 89) cohorts. For intra-operative or early MV repair failure assessment, the best area under the curve (AUC) was 0.75 and the complexity of mitral valve prolapse was the main predictor. In predicting moderate/severe recurrent MR at three years, the best AUC was 0.92 and residual MR at six months was the most important predictor. Machine learning algorithms may improve prognosis after MV repair procedure, thus improving indications for correct candidate selection for MV surgical repair.
| S-EPMC8469985 | biostudies-literature
Project description:Background Transcatheter mitral valve repair (TMVr) is currently offered at selected centers that meet certain operator and institutional requirements. We sought to explore the hypothesis that the availability of TMVr is associated with improved outcomes of MV surgery. Methods and Results We used the Nationwide Readmissions Database to identify patients who underwent MV surgery at centers with or without TMVr capabilities between January 1 and December 31, 2017. The primary end point was in-hospital mortality. Secondary end points were postoperative complications, resource use, and 30-day readmissions. A total of 24 477 patients from 595 centers (446 TMVr, 149 non-TMVr) were included. There were modest but statistically significant differences in the prevalence of comorbidities between the groups. Patients at non-TMVr centers had higher unadjusted in-hospital mortality than those at TMVr centers (5.6% versus 3.6%, P<0.001). They also had higher rates of postoperative complications, longer hospitalizations, higher cost, and fewer home discharges but similar 30-day readmission rates. After propensity matching, mortality remained higher at non-TMVr centers (5.5% versus 4.0%, P<0.001). Rates of postoperative complications, prolonged hospitalizations, and nonhome discharges also remained higher. Postoperative mortality was consistently higher at non-TMVr centers in multiple risk-adjustment analyses incrementally accounting for differences in risk factors, surgical volume, availability of surgical repair, and excluding concomitant procedures. In the most comprehensive model, surgery at non-TMVr centers was associated with higher odds of death (odds ratio, 1.41; 95% CI, 1.14-1.73; P=0.002). Conclusions Mitral valve surgery at TMVr centers is associated with improved in-hospital outcomes compared with non-TMVr centers.
| S-EPMC8174333 | biostudies-literature
Project description:ImportanceThe US Centers for Medicare and Medicaid Services recently released an updated national coverage determination proposal for transcatheter aortic valve replacement (TAVR) that maintains a focus on hospital TAVR volume and percutaneous coronary intervention (PCI) volume, and the national coverage determination for transcatheter mitral valve repair (TMVr) also has PCI volume requirements. However, the associations between hospital PCI volume and TAVR and TMVr outcomes are unknown.ObjectiveTo investigate whether hospital inpatient PCI volume is associated with rates of risk-adjusted in-hospital mortality and 30-day hospital readmission after TAVR and TMVr.Design, setting, and participantsThis population-based cross-sectional study of the 2016 Nationwide Readmissions Database included procedures completed in hospitals with a minimum of 5 TAVR or 5 TMVr procedures between January 1, 2016, and November 30, 2016.ExposuresHospitals were divided into quartiles based on annual inpatient PCI volumes.Main outcomes and measuresPrimary outcomes were in-hospital mortality and 30-day readmission rates. The associations between hospital inpatient PCI quartile and outcomes were evaluated using Kruskal-Wallis tests. Risk adjustment for in-hospital mortality rates was done through inclusion of variables based on the Elixhauser comorbidity classification, and risk adjustment for 30-day readmission rates was done in accordance with the Hospital-Wide Readmission Measure methodology used by the Centers for Medicare and Medicaid Services for public reporting.ResultsThere were 283 hospitals that performed at least 5 TAVRs, with a median inpatient PCI volume of 386 (interquartile range, 299-571) procedures, and 125 hospitals that performed at least 5 TMVr procedures, with a median inpatient PCI volume of 451 (interquartile range, 326-651) procedures. There was no association between hospital inpatient PCI volume and median TAVR risk-standardized in-hospital mortality (median [IQR] rates: bottom quartile, 1.82% [1.77%-1.90%]; second quartile, 1.81% [1.76%-1.86%]; third quartile, 1.81% [1.75%-1.90%]; top quartile, 1.82% [1.76%-1.91%]; P = .75) or the 30-day readmission (median [IQR] rates: bottom quartile, 13.6% [13.2%-14.3%]; second quartile, 13.3% [12.7%-14.0%]; third quartile, 13.5% [12.7%-14.3%]; top quartile, 13.8% [12.8%-14.3%]; P = .10) rates. Similarly, there was no association between hospital inpatient PCI volume and median TMVr risk-standardized in-hospital mortality rates (median [IQR] rates: bottom quartile, 1.84% [1.47%-2.53%]; second quartile, 1.65% [1.21%-3.02%]; third quartile, 1.80% [1.52%-3.58%]; top quartile, 1.76% [1.33%-4.20%]; P = .71) or 30-day readmission rates (median [IQR] rates: bottom quartile, 13.4% [13.1%-13.6%]; second quartile, 13.1% [12.9%-13.5%]; third quartile, 13.1% [12.9%-13.5%]; top quartile, 13.3% [12.8%-13.6%]; P = .30).Conclusions and relevanceIn this study, there was no association between inpatient PCI volume and TAVR or TMVr outcomes. Further evidence is needed to support inclusion of PCI volume minimums in national coverage determination requirements for hospital TAVR and TMVr programs.
| S-EPMC7042824 | biostudies-literature
Project description:Importance:Volume metrics may have relevance in the evaluation of valve center expertise. However, a paucity of data exists regarding the quantity, volume, and geographic location of mitral valve (MV) surgical centers in the United States and the proportion of underserved populations they treat. Objectives:To evaluate the hospital, patient, and procedural characteristics of mitral valve repair or replacement (MVRR) in the United States as a function of hospital procedure volume. Design, Setting, and Participants:This cross-sectional, multicenter observational study was conducted from July 2014 to June 2018. Patients in the Society of Thoracic Surgeons Adult Cardiac Surgery Database undergoing any surgical procedure involving MVRR in the United States were included. Main Outcomes and Measures:Volume distribution of MVRR by hospital and hospital referral region. Results:There were 165?405 MVRRs performed in 1082 centers during the study period, of which 86?488 (52.3%) were MV repairs. There were 575 centers (53.1%) that performed 25 or more MVRRs per year. The geographic distribution of centers performing 25 or more MVRRs per year differed from those performing fewer than 25 MVRRs per year. Of 304 designated hospital referral regions, 235 (77.3%) had at least 1 center performing 25 or more MVRRs per year, representing accessibility to 1 or more such centers for 296.4 million of 320.1 million US residents (92.6% of the US population; Midwest, 60.0 million of 68.0 million [88.4%]; South, 112.6 million of 122.6 million [91.9%]; West, 68.6 million of 72.9 million [94.1%]; and Northeast, 54.9 million of 56.6 million [97.1%]). Of 304 hospital referral regions, 168 (55.3%) had at least 1 center performing 40 or more MVRRs per year, representing accessibility to 1 or more such centers for 259.8 million of 317.90 million (81.7%) of the US population (Midwest, 50.5 million of 67.9 million [74.5%]; South, 94.5 million of 121.1 million [78.1%]; West, 64.0 million of 72.8 million [88.0%]; Northeast, 50.1 million of 56.3 million [90.2%]). More black and Hispanic patients received operations in centers performing 25 or more MVRRs per year (22?984) vs those performing fewer than 25 MVRRs per year (3227), yet the proportion was higher in lower-volume centers (22?984 of 148?385 [15.5%] vs 3227 of 17?020 [19.0%]; P?<?.001). In centers performing 25 or more MVRRs per year vs fewer than 25 MVRRs per year, there was a lower percentage of Medicare and Medicaid patients (47?920 of 148?385 [32.3%] vs 6183 of 17?020 [.3%]; P?<?.001) and patients from rural zip codes (21?208 of 148?385 [14.3%] vs 3146 of 17?020 [18.5%]; P?<?.001). Conclusions and Relevance:Fifty-three percent of all centers performed 25 or more MVRRs per year, and 92.6% of the US population lived in an hospital referral region with at least 1 such center. Disparities in race/ethnicity, rurality, and insurance status exist among patients being treated at centers with different volumes. These data indicate that efforts to centralize care based on volume metrics will need to balance access vs quality.
| S-EPMC6777237 | biostudies-literature
Project description:A patient with severe mitral regurgitation and chronic systolic heart failure taking inotropic support at home presents for transcatheter edge-to-edge mitral valve repair, complicated by torrential mitral regurgitation from damaged mitral leaflets requiring escalating mechanical circulatory support and ultimately expedited orthotopic heart transplantation. (Level of Difficulty: Intermediate.).
| S-EPMC8305671 | biostudies-literature
Project description:BackgroundBleeding in the context of cardiac catheterization is frequent and negatively impacts on short- and long-term patient outcome. We evaluated the clinical impact of in-hospital bleeding events after transcatheter mitral valve repair (TMVr) in the long-term follow- up.Methods586 consecutive patients treated with first-time TMVr were enrolled in this registry. In-hospital MVARC (Mitral Valve Academic Research Council) bleedings were assessed and patients were grouped according to the incidence of a bleeding event. Multivariate logistic regression was used to identify significant independent predictors of MVARC bleeding. This study received approval by local ethics committee.Results78 patients (13.3%) suffered from an MVARC bleeding event (Access site-related bleedings: 46.2%; GI tract bleeding: 35.9%; Other bleedings: 17.9%). Among these bleeding subgroups, neither relevant differences in baseline characteristics nor in severity of bleeding events were observed. Despite not being an independent predictor for overall death in the multivariate Cox regression analysis, MVARC bleeding was associated with prolonged hospital stay. The ORBIT bleeding score was the best match to predictors of any MVARC bleeding found in our cohort (c-score overall cohort: 0.68; c-score GI bleeding cohort: 0.72).ConclusionMVARC bleedings after TMVr are frequent findings but were only in half of the cases related to the access site. The ORBIT score could be useful for identification of patients at high risk for non-access site bleeding and especially GI bleeding.
| S-EPMC7797941 | biostudies-literature