Classifying early infant feeding status from clinical notes using natural language processing and machine learning.

The objective of this study is to develop and evaluate natural language processing (NLP) and machine learning models to predict infant feeding status from clinical notes in the Epic electronic health records system. The primary outcome was the classification of infant feeding status from clinical notes using Medical Subject Headings (MeSH) terms. Annotation of notes was completed using TeamTat to uniquely classify clinical notes according to infant feeding status. We trained 6 machine learning models to classify infant feeding status: logistic regression, random forest, XGBoost gradient descent, k-nearest neighbors, and support-vector classifier. Model comparison was evaluated based on overall accuracy, precision, recall, and F1 score. Our modeling corpus included an even number of clinical notes that was a balanced sample across each class. We manually reviewed 999 notes that represented 746 mother-infant dyads with a mean gestational age of 38.9 weeks and a mean maternal age of 26.6 years. The most frequent feeding status classification present for this study was exclusive breastfeeding [n = 183 (18.3%)], followed by exclusive formula bottle feeding [n = 146 (14.6%)], and exclusive feeding of expressed mother's milk [n = 102 (10.2%)], with mixed feeding being the least frequent [n = 23 (2.3%)]. Our final analysis evaluated the classification of clinical notes as breast, formula/bottle, and missing. The machine learning models were trained on these three classes after performing balancing and down sampling. The XGBoost model outperformed all others by achieving an accuracy of 90.1%, a macro-averaged precision of 90.3%, a macro-averaged recall of 90.1%, and a macro-averaged F1 score of 90.1%. Our results demonstrate that natural language processing can be applied to clinical notes stored in the electronic health records to classify infant feeding status. Early identification of breastfeeding status using NLP on unstructured electronic health records data can be used to inform precision public health interventions focused on improving lactation support for postpartum patients.

Cancer patients' satisfaction with telehealth during the COVID-19 pandemic.

OBJECTIVE: To examine factors associated with cancer patients' satisfaction using telehealth during COVID-19, including video conferencing platforms and secure messaging systems. METHOD: Patients with cancer participated in a cross-sectional, web-based survey was conducted with patients with cancer. The survey included questions about satisfaction with video-conferencing and secure messaging platforms to interact with clinicians during the COVID-19 pandemic. Logistic regression analyses were conducted to examine predictors of satisfaction for each telehealth platform. RESULTS: Participants generally reported positive satisfaction with each telehealth platform. Both platforms were commonly used to review medical results and discuss symptoms or treatment. Participants identifying as a man were most satisfied with their video-conferencing session, especially if they had a comfortable place to sit. Patients were more satisfied with secure messaging because they could ask a question without scheduling an appointment. DISCUSSION: When strategically used together, video-conferencing platforms and secure messaging may increase patient satisfaction in cancer care during the remainder of the pandemic and beyond. Attention must be paid to optimizing factors that promote satisfaction for each telehealth platform.

Hybrid Double Valve Replacement for Multivalvular Disease With Severe Mitral Annular Calcification.

Severe mitral annular calcification poses a significant challenge, particularly in higher risk, frail, elderly patients with multivalvular disease. Both surgical and transcatheter options exist but remain fraught with risks of paravalvular leak, atrioventricular groove disruption, patient prosthesis mismatch, and left ventricular outflow tract obstruction. We present an innovative hybrid surgical approach to manage severe aortic and mitral valve disease in a Jehovah's Witness patient with severe circumferential mitral annular calcification.

Endovascular selective hypothermia facilitates giant aneurysm clipping: illustrative case.

BACKGROUND: The authors present a case of selective hypothermia used for neuroprotection during clipping of a giant partially thrombosed middle cerebral artery (MCA) aneurysm. Although these cases have traditionally required deep hypothermic cardiac arrest, this case illustrates a novel and entirely endovascular solution that avoids cardiac standstill and whole-body cooling. OBSERVATIONS: This is, to the authors' knowledge, the first case in human surgery of a catheter-based selective hypothermic circuit used to facilitate MCA trapping for almost 30 minutes. Core temperatures never dropped below 34°C, and the patient recovered uneventfully and has been well for over 5 years. LESSONS: The technical nuances and physiological changes unique to selective hypothermia are discussed.

Starch or Saline After Cardiac Surgery: A Double-Blinded Randomized Controlled Trial.

BACKGROUND: Despite decades of investigation, the balance of clinical risks and benefits of fluid supplementation with starch remain unresolved. Patient-centered outcomes have not been well explored in a "real-world" trial in cardiac surgery. OBJECTIVE: We sought to compare a starch-based fluid strategy with a saline-based fluid strategy in the cardiac surgery patient. DESIGN: A pragmatic blinded randomized controlled trial comparing starch-based with saline-based fluid strategy. SETTING: A large tertiary academic center in London Ontario between September 2009 and February 2011. PARTICIPANTS: Patients undergoing planned, isolated coronary revascularization. MEASUREMENTS: Serum creatinine and patient weight were measured daily postoperatively. METHODS: Patients were randomized to receive 6% hydroxyethyl starch (Voluven) or saline for perioperative fluid requirements. Fluid administration was not protocolized. Co-primary outcomes were incidence of acute kidney injury (AKI) and maximum postoperative weight gain. Secondary outcomes included bleeding, transfusion, inotropic and ventilator support, and fluid utilization. RESULTS: The study was prematurely terminated due to resource limitations. A total of 69 patients (19% female, mean age = 65) were randomized. Using RIFLE criteria for AKI, "risk" occurred in 12 patients in each group (risk ratio [RR] = 1.0; 95% confidence interval [CI] = 0.5-1.9; P = 1.00), whereas "injury" occurred in 7 of 35 (20%) and 3 of 34 (9%) of patients in the starch and saline groups, respectively (RR = 2.3; 95% CI = 0.6-8.1; P = .31). Maximum weight gain, bleeding and blood product usage, and overall fluid requirement were similar between groups. LIMITATIONS: The study had to be prematurely terminated due to resource limitations which led to a small sample size which was not sufficiently powered to detect a difference in the primary outcomes. CONCLUSIONS: This pragmatic double-blinded randomized controlled trial revealed a number of interesting hypothesis-generating trends and confirmed the feasibility of undertaking a logistically complex trial in a pragmatic fashion.

CONTEXTE: L'équilibre entre les avantages et les risques cliniques d'une supplémentation en fluides à base d'amidon n'est toujours pas établi malgré des décennies d'études. Les résultats des patients subissant une chirurgie cardiaque n'ont pas été explorés suffisamment dans le cadre d'un essai concret. OBJECTIF: Comparer deux stratégies de supplémentation liquidienne, une solution à base d'amidon et une solution saline, chez des patients subissant une chirurgie cardiaque. TYPE D'ÉTUDE: Un essai pragmatique, contrôlé, à répartition aléatoire et mené en double insu comparant deux stratégies de supplémentation liquidienne une solution à base d'amidon et une solution saline. CADRE: Un grand centre universitaire de soins tertiaires de London (Ontario) entre septembre 2009 et février 2011. SUJETS: Des patients subissant une revascularisation coronarienne planifiée et isolée. MESURES: La créatinine sérique et le poids du patient ont été mesurés quotidiennement à la suite de l'intervention. MÉTHODOLOGIE: Les patients ont été répartis aléatoirement pour recevoir du Voluven (solution d'amidon hydroxyéthylé à 6 %) ou une solution saline pour les fluidiques périopératoires. L'administration ne s'est pas faite selon un protocole établi. L'incidence d'insuffisance rénale aiguë (IRA) et un gain pondéral maximal après l'intervention constituaient les deux principaux résultats mesurés. Les résultats secondaires incluaient une hémorragie, l'utilisation de transfusion sanguine, d'inotrope, d'assistance respiratoire et l'administration de fluides. RÉSULTATS: L'étude a été interrompue prématurément par manque de ressources. Les 69 patients (19 % de femmes) répartis aléatoirement avaient en moyenne 65 ans. La classification RIFLE avait permis de détecter un « risque ¼ d'IRA chez 12 patients dans chacun des groupes (RC: 1,0; IC 95 %: 0,5-1,9; p=1,00) et une « insuffisance ¼ chez 7 patients sur 35 (20 %) du groupe « amidon ¼ et 3 patients sur 34 (9 %) du groupe « saline ¼ (RC: 2,3; IC 95 %: 0,6-8,1; p=0,31). Le gain pondéral maximal, le nombre d'hémorragies, l'utilisation de produits sanguins et les besoins liquidiens étaient similaires dans les deux groupes. LIMITES: L'étude a été interrompue prématurément en raison d'un manque de ressources. Par conséquent, le faible échantillon de patients s'avère insuffisamment puissant pour détecter des différences significatives entre les deux groupes. CONCLUSIONS: Cette étude a mis en lumière quelques tendances permettant d'émettre des hypothèses intéressantes. L'étude a également confirmé la possibilité d'entreprendre un essai logistique complexe de manière pragmatique.

International Olympic Committee consensus statement on pain management in elite athletes.

Pain is a common problem among elite athletes and is frequently associated with sport injury. Both pain and injury interfere with the performance of elite athletes. There are currently no evidence-based or consensus-based guidelines for the management of pain in elite athletes. Typically, pain management consists of the provision of analgesics, rest and physical therapy. More appropriately, a treatment strategy should address all contributors to pain including underlying pathophysiology, biomechanical abnormalities and psychosocial issues, and should employ therapies providing optimal benefit and minimal harm. To advance the development of a more standardised, evidence-informed approach to pain management in elite athletes, an IOC Consensus Group critically evaluated the current state of the science and practice of pain management in sport and prepared recommendations for a more unified approach to this important topic.

Topical vancomycin applied on closure of the sternotomy wound does not prevent high levels of systemic vancomycin.

OBJECTIVE: Vancomycin is effective in reducing the risk of mediastinits and topical vancomycin has been hypothesised to give high local dose concentrations while avoiding high systemic levels, thus avoiding the risk of bacterial resistance to this second-line antibiotic. However, this theory has never been tested and the degree to which vancomycin is absorbed systemically is unknown. METHODS: Fourteen patients undergoing elective coronary artery bypass grafts (CABG) received 500mg of topical vancomycin prior to sternotomy closure. Serum samples were taken at 30, 60, 120, 180 and 720min post-operatively. In addition, samples were taken from the drain bottles and urine samples taken daily for 5 days. Vancomycin levels were measured by fluorescence polarisation immunoassay, using the reverse dilution method to give a detection limit of 0.8mg/l. RESULTS: Vancomycin was detected in almost all serum samples. Peak concentration was at 30min and the mean value was 2.96mg/l (range, 0.99-5.00mg/l). This mean fell to 1.32mg/l at 6h. Of the 500mg of vancomycin applied, a mean of only 8.8mg was found to have been lost into the drain bottles in the first 24h (range, 0.17-12.5mg). When 5 consecutive days of urine collection was achieved, a mean of 151mg of vancomycin was excreted (range, 40-195mg) and vancomycin was detectable in the urine till day 5. The mean concentration of vancomycin in the urine was maximal on day 1 and was 24.4mg/l (range, 4.49-44.98mg/l). CONCLUSIONS: Topical vancomycin causes significant systemic concentrations in the 6h post-surgery and can be detected in the urine for up to 5 days post-surgery.