Person:

Normand, Sharon-Lise

Introduction: During the past decade, the volume of percutaneous coronary intervention (PCI) in China has risen by more than 20-fold. Yet little is known about patterns of care and outcomes across hospitals, regions and time during this period of rising cardiovascular disease and dynamic change in the Chinese healthcare system. Methods and analysis Using the China PEACE (Patient-centered Evaluative Assessment of Cardiac Events) research network, the Retrospective Study of Coronary Catheterisation and Percutaneous Coronary Intervention (China PEACE-Retrospective CathPCI Study) will examine a nationally representative sample of 11 900 patients who underwent coronary catheterisation or PCI at 55 Chinese hospitals during 2001, 2006 and 2011. We selected patients and study sites using a two-stage cluster sampling design with simple random sampling stratified within economical-geographical strata. A central coordinating centre will monitor data quality at the stages of case ascertainment, medical record abstraction and data management. We will examine patient characteristics, diagnostic testing patterns, procedural treatments and in-hospital outcomes, including death, complications of treatment and costs of hospitalisation. We will additionally characterise variation in treatments and outcomes by patient characteristics, hospital, region and study year. Ethics and dissemination The China PEACE collaboration is designed to translate research into improved care for patients. The study protocol was approved by the central ethics committee at the China National Center for Cardiovascular Diseases (NCCD) and collaborating hospitals. Findings will be shared with participating hospitals, policymakers and the academic community to promote quality monitoring, quality improvement and the efficient allocation and use of coronary catheterisation and PCI in China. Registration details http://www.clinicaltrials.gov (NCT01624896).

Objective: To characterize the absolute risks for older patients of readmission to hospital and death in the year after hospitalization for heart failure, acute myocardial infarction, or pneumonia. Design: Retrospective cohort study. Setting: 4767 hospitals caring for Medicare fee for service beneficiaries in the United States, 2008-10. Participants: More than 3 million Medicare fee for service beneficiaries, aged 65 years or more, surviving hospitalization for heart failure, acute myocardial infarction, or pneumonia. Main outcome measures Daily absolute risks of first readmission to hospital and death for one year after discharge. To illustrate risk trajectories, we identified the time required for risks of readmission to hospital and death to decline 50% from maximum values after discharge; the time required for risks to approach plateau periods of minimal day to day change, defined as 95% reductions in daily changes in risk from maximum daily declines after discharge; and the extent to which risks are higher among patients recently discharged from hospital compared with the general elderly population. Results: Within one year of hospital discharge, readmission to hospital and death, respectively, occurred following 67.4% and 35.8% of hospitalizations for heart failure, 49.9% and 25.1% for acute myocardial infarction, and 55.6% and 31.1% for pneumonia. Risk of first readmission had declined 50% by day 38 after hospitalization for heart failure, day 13 after hospitalization for acute myocardial infarction, and day 25 after hospitalization for pneumonia; risk of death declined 50% by day 11, 6, and 10, respectively. Daily change in risk of first readmission to hospital declined 95% by day 45, 38, and 45; daily change in risk of death declined 95% by day 21, 19, and 21. After hospitalization for heart failure, acute myocardial infarction, or pneumonia, the magnitude of the relative risk for hospital admission over the first 90 days was 8, 6, and 6 times greater than that of the general older population; the relative risk of death was 11, 8, and 10 times greater. Conclusions: Risk declines slowly for older patients after hospitalization for heart failure, acute myocardial infarction, or pneumonia and is increased for months. Specific risk trajectories vary by discharge diagnosis and outcome. Patients should remain vigilant for deterioration in health for an extended time after discharge. Health providers can use knowledge of absolute risks and their changes over time to better align interventions designed to reduce adverse outcomes after discharge with the highest risk periods for patients.

Background: Machine learning methods may complement traditional analytic methods for medical device surveillance. Methods and results Using data from the National Cardiovascular Data Registry for implantable cardioverter–defibrillators (ICDs) linked to Medicare administrative claims for longitudinal follow-up, we applied three statistical approaches to safety-signal detection for commonly used dual-chamber ICDs that used two propensity score (PS) models: one specified by subject-matter experts (PS-SME), and the other one by machine learning-based selection (PS-ML). The first approach used PS-SME and cumulative incidence (time-to-event), the second approach used PS-SME and cumulative risk (Data Extraction and Longitudinal Trend Analysis [DELTA]), and the third approach used PS-ML and cumulative risk (embedded feature selection). Safety-signal surveillance was conducted for eleven dual-chamber ICD models implanted at least 2,000 times over 3 years. Between 2006 and 2010, there were 71,948 Medicare fee-for-service beneficiaries who received dual-chamber ICDs. Cumulative device-specific unadjusted 3-year event rates varied for three surveyed safety signals: death from any cause, 12.8%–20.9%; nonfatal ICD-related adverse events, 19.3%–26.3%; and death from any cause or nonfatal ICD-related adverse event, 27.1%–37.6%. Agreement among safety signals detected/not detected between the time-to-event and DELTA approaches was 90.9% (360 of 396, k=0.068), between the time-to-event and embedded feature-selection approaches was 91.7% (363 of 396, k=−0.028), and between the DELTA and embedded feature selection approaches was 88.1% (349 of 396, k=−0.042). Conclusion: Three statistical approaches, including one machine learning method, identified important safety signals, but without exact agreement. Ensemble methods may be needed to detect all safety signals for further evaluation during medical device surveillance.