Quantifying the risk of prosthetic joint infections after invasive dental procedures and the effect of antibiotic prophylaxis.

BACKGROUND: Dentists face the expectations of orthopedic surgeons and patients with prosthetic joints to provide antibiotic prophylaxis (AP) before invasive dental procedures (IDPs) to reduce the risk of late periprosthetic joint infections (LPJIs), despite the lack of evidence associating IDPs with LPJIs, lack of evidence of AP efficacy, risk of AP-related adverse reactions, and potential for promoting antibiotic resistance. The authors aimed to identify any association between IDPs and LPJIs and whether AP reduces LPJI incidence after IDPs. METHOD: The authors performed a case-crossover analysis comparing IDP incidence in the 3 months immediately before LPJI hospital admission (case period) with the preceding 12-month control period for all LPJI hospital admissions with commercial or Medicare supplemental or Medicaid health care coverage and linked dental and prescription benefits data. RESULTS: Overall, 2,344 LPJI hospital admissions with dental and prescription records (n = 1,160 commercial or Medicare supplemental and n = 1,184 Medicaid) were identified. Patients underwent 4,614 dental procedures in the 15 months before LPJI admission, including 1,821 IDPs (of which 18.3% had AP). Our analysis identified no significant positive association between IDPs and subsequent development of LPJIs and no significant effect of AP in reducing LPJIs. CONCLUSIONS: The authors identified no significant association between IDPs and LPJIs and no effect of AP cover of IDPs in reducing the risk of LPJIs. PRACTICAL IMPLICATIONS: In the absence of benefit, the continued use of AP poses an unnecessary risk to patients from adverse drug reactions and to society from the potential of AP to promote development of antibiotic resistance. Dental AP use to prevent LPJIs should, therefore, cease.

Electronic phenotyping of health outcomes of interest using a linked claims-electronic health record database: Findings from a machine learning pilot project.

OBJECTIVE: Claims-based algorithms are used in the Food and Drug Administration Sentinel Active Risk Identification and Analysis System to identify occurrences of health outcomes of interest (HOIs) for medical product safety assessment. This project aimed to apply machine learning classification techniques to demonstrate the feasibility of developing a claims-based algorithm to predict an HOI in structured electronic health record (EHR) data. MATERIALS AND METHODS: We used the 2015-2019 IBM MarketScan Explorys Claims-EMR Data Set, linking administrative claims and EHR data at the patient level. We focused on a single HOI, rhabdomyolysis, defined by EHR laboratory test results. Using claims-based predictors, we applied machine learning techniques to predict the HOI: logistic regression, LASSO (least absolute shrinkage and selection operator), random forests, support vector machines, artificial neural nets, and an ensemble method (Super Learner). RESULTS: The study cohort included 32 956 patients and 39 499 encounters. Model performance (positive predictive value [PPV], sensitivity, specificity, area under the receiver-operating characteristic curve) varied considerably across techniques. The area under the receiver-operating characteristic curve exceeded 0.80 in most model variations. DISCUSSION: For the main Food and Drug Administration use case of assessing risk of rhabdomyolysis after drug use, a model with a high PPV is typically preferred. The Super Learner ensemble model without adjustment for class imbalance achieved a PPV of 75.6%, substantially better than a previously used human expert-developed model (PPV = 44.0%). CONCLUSIONS: It is feasible to use machine learning methods to predict an EHR-derived HOI with claims-based predictors. Modeling strategies can be adapted for intended uses, including surveillance, identification of cases for chart review, and outcomes research.

Prevalence of Metabolic Syndrome in an Employed Population as Determined by Analysis of Three Data Sources.

OBJECTIVE: The aim of this study was to compare estimates of the prevalence and incidence of metabolic syndrome (MetS) using various data sources. METHODS: We integrated health risk assessment (HRA), claims, and biometric screening data from Lockheed Martin Corporation. We measured the extent to which MetS risk factors measured using HRA and medical claims correlated with biometric screening data. RESULTS: Using biometric data, 24.9% of employees were identified as having MetS. Prevalence estimates were much lower using HRA data (6.8%) and claims (3.7%). Between 2012 and 2014, 10.4% of the sample newly acquired MetS. The number of MetS risk factors per employee was predictive of diabetes, heart disease, health care costs, and utilization. CONCLUSION: MetS is prevalent and associated with progression to disease. It is more easily tracked with biometric screening data than with HRA or claims data. Employers should consider efforts to manage and prevent this condition in their workforce.