Scalable Incident Detection via Natural Language Processing and Probabilistic Language Models.

Post marketing safety surveillance depends in part on the ability to detect concerning clinical events at scale. Spontaneous reporting might be an effective component of safety surveillance, but it requires awareness and understanding among healthcare professionals to achieve its potential. Reliance on readily available structured data such as diagnostic codes risk under-coding and imprecision. Clinical textual data might bridge these gaps, and natural language processing (NLP) has been shown to aid in scalable phenotyping across healthcare records in multiple clinical domains. In this study, we developed and validated a novel incident phenotyping approach using unstructured clinical textual data agnostic to Electronic Health Record (EHR) and note type. It's based on a published, validated approach (PheRe) used to ascertain social determinants of health and suicidality across entire healthcare records. To demonstrate generalizability, we validated this approach on two separate phenotypes that share common challenges with respect to accurate ascertainment: 1) suicide attempt; 2) sleep-related behaviors. With samples of 89,428 records and 35,863 records for suicide attempt and sleep-related behaviors, respectively, we conducted silver standard (diagnostic coding) and gold standard (manual chart review) validation. We showed Area Under the Precision-Recall Curve of ∼ 0.77 (95% CI 0.75-0.78) for suicide attempt and AUPR ∼ 0.31 (95% CI 0.28-0.34) for sleep-related behaviors. We also evaluated performance by coded race and demonstrated differences in performance by race were dissimilar across phenotypes and require algorithmovigilance and debiasing prior to implementation.

Cardiovascular implantable electronic device lead safety: Harnessing real-world remote monitoring data for medical device evaluation.

BACKGROUND: Current methods to identify cardiovascular implantable electronic device lead failure include postapproval studies, which may be limited in scope, participant numbers, and attrition; studies relying on administrative codes, which lack specificity; and voluntary adverse event reporting, which cannot determine incidence or attribution to the lead. OBJECTIVE: The purpose of this study was to determine whether adjudicated remote monitoring (RM) data can address these limitations and augment lead safety evaluation. METHODS: Among 48,191 actively monitored patients with a cardiovascular implantable electronic device, we identified RM transmissions signifying incident lead abnormalities and, separately, identified all leads abandoned or extracted between April 1, 2019, and April 1, 2021. We queried electronic health record and Medicare fee-for-service claims data to determine whether patients had administrative codes for lead failure. We verified lead failure through manual electronic health record review. RESULTS: Of the 48,191 patients, 1170 (2.4%) had incident lead abnormalities detected by RM. Of these, 409 patients had administrative codes for lead failure, and 233 of these 409 patients (57.0%) had structural lead failure verified through chart review. Of the 761 patients without administrative codes, 167 (21.9%) had structural lead failure verified through chart review. Thus, 400 patients with RM transmissions suggestive of lead abnormalities (34.2%) had structural lead failure. In addition, 200 patients without preceding abnormal RM transmissions had leads abandoned or extracted for structural failure, making the total lead failure cohort 600 patients (66.7% with RM abnormalities, 33.3% without). Patients with isolated right atrial or left ventricular lead failure were less likely to have lead replacement and administrative codes reflective of lead failure. CONCLUSION: RM may strengthen real-world assessment of lead failure, particularly for leads where patients do not undergo replacement.

Information Extraction From Electronic Health Records to Predict Readmission Following Acute Myocardial Infarction: Does Natural Language Processing Using Clinical Notes Improve Prediction of Readmission?

Background Social risk factors influence rehospitalization rates yet are challenging to incorporate into prediction models. Integration of social risk factors using natural language processing (NLP) and machine learning could improve risk prediction of 30-day readmission following an acute myocardial infarction. Methods and Results Patients were enrolled into derivation and validation cohorts. The derivation cohort included inpatient discharges from Vanderbilt University Medical Center between January 1, 2007, and December 31, 2016, with a primary diagnosis of acute myocardial infarction, who were discharged alive, and not transferred from another facility. The validation cohort included patients from Dartmouth-Hitchcock Health Center between April 2, 2011, and December 31, 2016, meeting the same eligibility criteria described above. Data from both sites were linked to Centers for Medicare & Medicaid Services administrative data to supplement 30-day hospital readmissions. Clinical notes from each cohort were extracted, and an NLP model was deployed, counting mentions of 7 social risk factors. Five machine learning models were run using clinical and NLP-derived variables. Model discrimination and calibration were assessed, and receiver operating characteristic comparison analyses were performed. The 30-day rehospitalization rates among the derivation (n=6165) and validation (n=4024) cohorts were 15.1% (n=934) and 10.2% (n=412), respectively. The derivation models demonstrated no statistical improvement in model performance with the addition of the selected NLP-derived social risk factors. Conclusions Social risk factors extracted using NLP did not significantly improve 30-day readmission prediction among hospitalized patients with acute myocardial infarction. Alternative methods are needed to capture social risk factors.

Adaptation of an NLP system to a new healthcare environment to identify social determinants of health.

Social determinants of health (SDoH) are increasingly important factors for population health, healthcare outcomes, and care delivery. However, many of these factors are not reliably captured within structured electronic health record (EHR) data. In this work, we evaluated and adapted a previously published NLP tool to include additional social risk factors for deployment at Vanderbilt University Medical Center in an Acute Myocardial Infarction cohort. We developed a transformation of the SDoH outputs of the tool into the OMOP common data model (CDM) for re-use across many potential use cases, yielding performance measures across 8 SDoH classes of precision 0.83 recall 0.74 and F-measure of 0.78.

Association of Wearable Device Use With Pulse Rate and Health Care Use in Adults With Atrial Fibrillation.

Importance: Increasingly, individuals with atrial fibrillation (AF) use wearable devices (hereafter wearables) that measure pulse rate and detect arrhythmia. The associations of wearables with health outcomes and health care use are unknown. Objective: To characterize patients with AF who use wearables and compare pulse rate and health care use between individuals who use wearables and those who do not. Design, Setting, and Participants: This retrospective, propensity-matched cohort study included 90 days of follow-up of patients in a tertiary care, academic health system. Included patients were adults with at least 1 AF-specific International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code from 2017 through 2019. Electronic medical records were reviewed to identify 125 individuals who used wearables and had adequate pulse-rate follow-up who were then matched using propensity scores 4 to 1 with 500 individuals who did not use wearables. Data were analyzed from June 2020 through February 2021. Exposure: Using commercially available wearables with pulse rate or rhythm evaluation capabilities. Main Outcomes and Measures: Mean pulse rates from measures taken in the clinic or hospital and a composite health care use score were recorded. The composite outcome included evaluation and management, ablation, cardioversion, telephone encounters, and number of rate or rhythm control medication orders. Results: Among 16 320 patients with AF included in the analysis, 348 patients used wearables and 15 972 individuals did not use wearables. Prior to matching, patients using wearables were younger (mean [SD] age, 64.0 [13.0] years vs 70.0 [13.8] years; P < .001) and healthier (mean [SD] CHA2DS2-VASc [congestive heart failure, hypertension, age ≥ 65 years or 65-74 years, diabetes, prior stroke/transient ischemic attack, vascular disease, sex] score, 3.6 [2.0] vs 4.4 [2.0]; P < .001) compared with individuals not using wearables, with similar gender distribution (148 [42.5%] women vs 6722 women [42.1%]; P = .91). After matching, mean pulse rate was similar between 125 patients using wearables and 500 patients not using wearables (75.01 [95% CI, 72.74-77.27] vs 75.79 [95% CI, 74.68-76.90] beats per minute [bpm]; P = .54), whereas mean composite use score was higher among individuals using wearables (3.55 [95% CI, 3.31-3.80] vs 3.27 [95% CI, 3.14-3.40]; P = .04). Among measures in the composite outcome, there was a significant difference in use of ablation, occurring in 22 individuals who used wearables (17.6%) vs 37 individuals who did not use wearables (7.4%) (P = .001). In the regression analyses, mean composite use score was 0.28 points (95% CI, 0.01 to 0.56 points) higher among individuals using wearables compared with those not using wearables and mean pulse was similar, with a -0.79 bpm (95% CI -3.28 to 1.71 bpm) difference between the groups. Conclusions and Relevance: This study found that follow-up health care use among individuals with AF was increased among those who used wearables compared with those with similar pulse rates who did not use wearables. Given the increasing use of wearables by patients with AF, prospective, randomized, long-term evaluation of the associations of wearable technology with health outcomes and health care use is needed.

Baseline phenotype and 30-day outcomes of people tested for COVID-19: an international network cohort including >3.32 million people tested with real-time PCR and >219,000 tested positive for SARS-CoV-2 in South Korea, Spain and the United States.

Early identification of symptoms and comorbidities most predictive of COVID-19 is critical to identify infection, guide policies to effectively contain the pandemic, and improve health systems' response. Here, we characterised socio-demographics and comorbidity in 3,316,107persons tested and 219,072 persons tested positive for SARS-CoV-2 since January 2020, and their key health outcomes in the month following the first positive test. Routine care data from primary care electronic health records (EHR) from Spain, hospital EHR from the United States (US), and claims data from South Korea and the US were used. The majority of study participants were women aged 18-65 years old. Positive/tested ratio varied greatly geographically (2.2:100 to 31.2:100) and over time (from 50:100 in February-April to 6.8:100 in May-June). Fever, cough and dyspnoea were the most common symptoms at presentation. Between 4%-38% required admission and 1-10.5% died within a month from their first positive test. Observed disparity in testing practices led to variable baseline characteristics and outcomes, both nationally (US) and internationally. Our findings highlight the importance of large scale characterization of COVID-19 international cohorts to inform planning and resource allocation including testing as countries face a second wave.

Association of Cardiac Rehabilitation With Decreased Hospitalization and Mortality Risk After Cardiac Valve Surgery.

Importance: National guidelines recommend cardiac rehabilitation (CR) after cardiac valve surgery, and CR is covered by Medicare for this indication. However, few data exist regarding current CR enrollment after valve surgery. Objective: To characterize CR enrollment after cardiac valve surgery and its association with outcomes, including hospitalizations and mortality. Design, Setting, and Participants: This cohort study of patients undergoing valve surgery was conducted in calendar year 2014, with follow-up through 2015. The study included all fee-for-service Medicare beneficiaries undergoing open cardiac valve surgery in 2014. Patients identified by inpatient diagnosis codes for open aortic, mitral, tricuspid, and pulmonary valve surgery were included. Data analysis occurred from January 2018 to March 2019. Exposures: Logistic regression was used to evaluate sociodemographic and clinical factors associated with CR enrollment. Main Outcomes and Measures: We used Andersen-Gill models to evaluate the association of CR enrollment with 1-year hospitalization risk and Cox regression models to evaluate the association of CR enrollment with 1-year mortality risk. Results: A total of 41 369 Medicare beneficiaries (median [interquartile range] age, 73 [68-79] years; 16 935 [40.9%] female) underwent open valve surgery in the United States in 2014. Fewer than half of patients (17 855 [43.2%]) who had valve surgery enrolled in CR programs. Several racial/ethnic groups had lower odds of enrolling in CR programs after valve surgery compared with white patients, including Asian patients (odds ratio [OR], 0.36 [95% CI, 0.28-0.47]), black patients (OR, 0.60 [95% CI, 0.54-0.67]), and Hispanic patients (OR, 0.36 [95% CI, 0.28-0.46]). Patients undergoing concomitant coronary artery bypass grafting had higher odds of CR enrollment (OR, 1.26 [95% CI, 1.20-1.31]) than those without the concomitant coronary artery bypass graft procedure, as did patients in the Midwest census region (OR, 2.40 [95% CI, 2.28-2.54]) compared with those in the South (reference). Cardiac rehabilitation enrollment was associated with fewer hospitalizations within 1 year of discharge (hazard ratio, 0.66 [95% CI, 0.63-0.69] after multivariable adjustment). Enrollment was also associated with a 4.2% absolute decrease in 1-year mortality risk (hazard ratio, 0.39 [95% CI, 0.35-0.44] after multivariable adjustment). Conclusions and Relevance: Fewer than half of Medicare beneficiaries undergoing cardiac valve surgery enroll in CR programs, and there are marked racial/ethnic disparities among those that do. Cardiac rehabilitation is associated with decreased 1-year cumulative hospitalization and mortality risk after valve surgery. These results invite further study on barriers to CR enrollment in this population.

Congestive heart failure information extraction framework for automated treatment performance measures assessment.

OBJECTIVE: This paper describes a new congestive heart failure (CHF) treatment performance measure information extraction system - CHIEF - developed as part of the Automated Data Acquisition for Heart Failure project, a Veterans Health Administration project aiming at improving the detection of patients not receiving recommended care for CHF. DESIGN: CHIEF is based on the Apache Unstructured Information Management Architecture framework, and uses a combination of rules, dictionaries, and machine learning methods to extract left ventricular function mentions and values, CHF medications, and documented reasons for a patient not receiving these medications. MEASUREMENTS: The training and evaluation of CHIEF were based on subsets of a reference standard of various clinical notes from 1083 Veterans Health Administration patients. Domain experts manually annotated these notes to create our reference standard. Metrics used included recall, precision, and the F 1 -measure. RESULTS: In general, CHIEF extracted CHF medications with high recall (>0.990) and good precision (0.960-0.978). Mentions of Left Ventricular Ejection Fraction were also extracted with high recall (0.978-0.986) and precision (0.986-0.994), and quantitative values of Left Ventricular Ejection Fraction were found with 0.910-0.945 recall and with high precision (0.939-0.976). Reasons for not prescribing CHF medications were more difficult to extract, only reaching fair accuracy with about 0.310-0.400 recall and 0.250-0.320 precision. CONCLUSION: This study demonstrated that applying natural language processing to unlock the rich and detailed clinical information found in clinical narrative text notes makes fast and scalable quality improvement approaches possible, eventually improving management and outpatient treatment of patients suffering from CHF.

Assessing the cost burden of United States FDA-mandated post-approval studies for medical devices.

Approved medical devices frequently undergo FDA mandated post-approval studies (PAS). However, there is uncertainty as to the value of PAS in assessing the safety of medical devices and the cost of these studies to the healthcare system is unknown. Since PAS costs are funded through device manufacturers who do not share the costs with regulators, we sought to estimate the total PAS costs through interviews with a panel of experts in medical device clinical trial design in order to design a general cost model for PAS which was then applied to the FDA PAS. A total of 277 PAS were initiated between 3/1/05 through 6/30/13 and demonstrated a median cost of $2.16 million per study and an overall cost of $1.22 billion over the 8.25 years of study. While these costs are funded through manufacturers, the ultimate cost is borne by the healthcare system through the medical device costs. Given concerns regarding the informational value of PAS, the resources used to support mandated PAS may be better allocated to other approaches to assure safety.

Use of multiple imputation method to improve estimation of missing baseline serum creatinine in acute kidney injury research.

BACKGROUND AND OBJECTIVES: Baseline creatinine (BCr) is frequently missing in AKI studies. Common surrogate estimates can misclassify AKI and adversely affect the study of related outcomes. This study examined whether multiple imputation improved accuracy of estimating missing BCr beyond current recommendations to apply assumed estimated GFR (eGFR) of 75 ml/min per 1.73 m(2) (eGFR 75). DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: From 41,114 unique adult admissions (13,003 with and 28,111 without BCr data) at Vanderbilt University Hospital between 2006 and 2008, a propensity score model was developed to predict likelihood of missing BCr. Propensity scoring identified 6502 patients with highest likelihood of missing BCr among 13,003 patients with known BCr to simulate a "missing" data scenario while preserving actual reference BCr. Within this cohort (n=6502), the ability of various multiple-imputation approaches to estimate BCr and classify AKI were compared with that of eGFR 75. RESULTS: All multiple-imputation methods except the basic one more closely approximated actual BCr than did eGFR 75. Total AKI misclassification was lower with multiple imputation (full multiple imputation + serum creatinine) (9.0%) than with eGFR 75 (12.3%; P<0.001). Improvements in misclassification were greater in patients with impaired kidney function (full multiple imputation + serum creatinine) (15.3%) versus eGFR 75 (40.5%; P<0.001). Multiple imputation improved specificity and positive predictive value for detecting AKI at the expense of modestly decreasing sensitivity relative to eGFR 75. CONCLUSIONS: Multiple imputation can improve accuracy in estimating missing BCr and reduce misclassification of AKI beyond currently proposed methods.

iDASH: integrating data for analysis, anonymization, and sharing.

iDASH (integrating data for analysis, anonymization, and sharing) is the newest National Center for Biomedical Computing funded by the NIH. It focuses on algorithms and tools for sharing data in a privacy-preserving manner. Foundational privacy technology research performed within iDASH is coupled with innovative engineering for collaborative tool development and data-sharing capabilities in a private Health Insurance Portability and Accountability Act (HIPAA)-certified cloud. Driving Biological Projects, which span different biological levels (from molecules to individuals to populations) and focus on various health conditions, help guide research and development within this Center. Furthermore, training and dissemination efforts connect the Center with its stakeholders and educate data owners and data consumers on how to share and use clinical and biological data. Through these various mechanisms, iDASH implements its goal of providing biomedical and behavioral researchers with access to data, software, and a high-performance computing environment, thus enabling them to generate and test new hypotheses.

Evaluation of an automated safety surveillance system using risk adjusted sequential probability ratio testing.

BACKGROUND: Automated adverse outcome surveillance tools and methods have potential utility in quality improvement and medical product surveillance activities. Their use for assessing hospital performance on the basis of patient outcomes has received little attention. We compared risk-adjusted sequential probability ratio testing (RA-SPRT) implemented in an automated tool to Massachusetts public reports of 30-day mortality after isolated coronary artery bypass graft surgery. METHODS: A total of 23,020 isolated adult coronary artery bypass surgery admissions performed in Massachusetts hospitals between January 1, 2002 and September 30, 2007 were retrospectively re-evaluated. The RA-SPRT method was implemented within an automated surveillance tool to identify hospital outliers in yearly increments. We used an overall type I error rate of 0.05, an overall type II error rate of 0.10, and a threshold that signaled if the odds of dying 30-days after surgery was at least twice than expected. Annual hospital outlier status, based on the state-reported classification, was considered the gold standard. An event was defined as at least one occurrence of a higher-than-expected hospital mortality rate during a given year. RESULTS: We examined a total of 83 hospital-year observations. The RA-SPRT method alerted 6 events among three hospitals for 30-day mortality compared with 5 events among two hospitals using the state public reports, yielding a sensitivity of 100% (5/5) and specificity of 98.8% (79/80). CONCLUSIONS: The automated RA-SPRT method performed well, detecting all of the true institutional outliers with a small false positive alerting rate. Such a system could provide confidential automated notification to local institutions in advance of public reporting providing opportunities for earlier quality improvement interventions.

Automated identification of postoperative complications within an electronic medical record using natural language processing.

CONTEXT: Currently most automated methods to identify patient safety occurrences rely on administrative data codes; however, free-text searches of electronic medical records could represent an additional surveillance approach. OBJECTIVE: To evaluate a natural language processing search-approach to identify postoperative surgical complications within a comprehensive electronic medical record. DESIGN, SETTING, AND PATIENTS: Cross-sectional study involving 2974 patients undergoing inpatient surgical procedures at 6 Veterans Health Administration (VHA) medical centers from 1999 to 2006. MAIN OUTCOME MEASURES: Postoperative occurrences of acute renal failure requiring dialysis, deep vein thrombosis, pulmonary embolism, sepsis, pneumonia, or myocardial infarction identified through medical record review as part of the VA Surgical Quality Improvement Program. We determined the sensitivity and specificity of the natural language processing approach to identify these complications and compared its performance with patient safety indicators that use discharge coding information. RESULTS: The proportion of postoperative events for each sample was 2% (39 of 1924) for acute renal failure requiring dialysis, 0.7% (18 of 2327) for pulmonary embolism, 1% (29 of 2327) for deep vein thrombosis, 7% (61 of 866) for sepsis, 16% (222 of 1405) for pneumonia, and 2% (35 of 1822) for myocardial infarction. Natural language processing correctly identified 82% (95% confidence interval [CI], 67%-91%) of acute renal failure cases compared with 38% (95% CI, 25%-54%) for patient safety indicators. Similar results were obtained for venous thromboembolism (59%, 95% CI, 44%-72% vs 46%, 95% CI, 32%-60%), pneumonia (64%, 95% CI, 58%-70% vs 5%, 95% CI, 3%-9%), sepsis (89%, 95% CI, 78%-94% vs 34%, 95% CI, 24%-47%), and postoperative myocardial infarction (91%, 95% CI, 78%-97%) vs 89%, 95% CI, 74%-96%). Both natural language processing and patient safety indicators were highly specific for these diagnoses. CONCLUSION: Among patients undergoing inpatient surgical procedures at VA medical centers, natural language processing analysis of electronic medical records to identify postoperative complications had higher sensitivity and lower specificity compared with patient safety indicators based on discharge coding.

Development of inpatient risk stratification models of acute kidney injury for use in electronic health records.

OBJECTIVE: Patients with hospital-acquired acute kidney injury (AKI) are at risk for increased mortality and further medical complications. Evaluating these patients with a prediction tool easily implemented within an electronic health record (EHR) would identify high-risk patients prior to the development of AKI and could prevent iatrogenically induced episodes of AKI and improve clinical management. METHODS: The authors used structured clinical data acquired from an EHR to identify patients with normal kidney function for admissions from 1 August 1999 to 31 July 2003. Using administrative, computerized provider order entry and laboratory test data, they developed a 3-level risk stratification model to predict each of 2 severity levels of in-hospital AKI as defined by RIFLE criteria. The severity levels were defined as 150% or 200% of baseline serum creatinine. Model discrimination and calibration were evaluated using 10-fold cross-validation. RESULTS: Cross-validation of the models resulted in area under the receiver operating characteristic (AUC) curves of 0.75 (150% elevation) and 0.78 (200% elevation). Both models were adequately calibrated as measured by the Hosmer-Lemeshow goodness-of-fit test chi-squared values of 9.7 (P = 0.29) and 12.7 (P = 0.12), respectively. CONCLUSIONS: The authors generated risk prediction models for hospital-acquired AKI using only commonly available electronic data. The models identify patients at high risk for AKI who might benefit from early intervention or increased monitoring.

Occurrence of adverse, often preventable, events in community hospitals involving nephrotoxic drugs or those excreted by the kidney.

Medication errors in patients with reduced creatinine clearance are harmful and costly; however, most studies have been conducted in large academic hospitals. As there are few studies regarding this issue in smaller community hospitals, we conducted a multicenter, retrospective cohort study in six community hospitals (100 to 300 beds) to assess the incidence and severity of adverse drug events (ADEs) in patients with reduced creatinine clearance. A chart review was performed on adult patients hospitalized during a 20-month study period with serum creatinine over 1.5 mg/dl who were exposed to drugs that are nephrotoxic or cleared by the kidney. Among 109,641 patients, 17,614 had reduced creatinine clearance, and in a random sample of 900 of these patients, there were 498 potential ADEs and 90 ADEs. Among these ADEs, 91% were preventable, 51% were serious, 44% were significant, and 4.5% were life threatening. Of the potential ADEs, 54% were serious, 44% were significant, 1.6% were life threatening, and 96.6% were not intercepted. All 82 preventable events could have been intercepted by renal dose checking. Our study shows that ADEs were common in patients with impaired kidney function in community hospitals, and many appear potentially preventable with renal dose checking.

Risk-adjusted sequential probability ratio test control chart methods for monitoring operator and institutional mortality rates in interventional cardiology.

BACKGROUND: The objective of this study was to evaluate risk-adjusted sequential probability ratio test control charts for the detection of significant discrepancies between institution or individual interventional cardiologist postprocedural mortality rates and national or local event rate expectations. METHODS: Eight thousand nine hundred forty-two percutaneous coronary interventional procedures were performed by 27 operators between January 1, 2002, and November 30, 2006. The institution-based evaluation included all procedures, and the individual-based evaluations included 8750 procedures performed by 18 operators who had each done at least 100 PCI procedures. Risk-adjusted sequential probability ratio test control charts were developed to assess whether the odds ratios (ORs) for death were >2.0 for alpha = beta = 0.10. The American College of Cardiology 1.1 prediction model was used to risk-adjust both the institution and individuals, and an additional local model was used for individuals. RESULTS: After national risk adjustment, the local institution did not show mortality of more than a 1.5 OR. Two operators had a >2.0 mortality OR after national risk adjustment, and one of those remained elevated after local risk adjustment. Of 18 operators, 10 had insufficient data to allow us to accept or reject the hypothesis of increased risk. CONCLUSIONS: The local institution performed within national expectations, but 1 operator was identified as having poor performance, which prompted an in-depth review of that operator's cases. The review revealed that the operator had an unusually high number of patients who presented with risk factors not included in the risk-adjustment models. This study highlights the utility of risk-adjusted sequential probability ratio test as a method for outcomes monitoring and quality control in interventional cardiology.