Predicting Penicillin Allergy: A United States Multicenter Retrospective Study.

BACKGROUND: Using the reaction history in logistic regression and machine learning (ML) models to predict penicillin allergy has been reported based on non-US data. OBJECTIVE: We developed ML positive penicillin allergy testing prediction models from multisite US data. METHODS: Retrospective data from 4 US-based hospitals were grouped into 4 datasets: enriched training (1:3 case-control matched cohort), enriched testing, nonenriched internal testing, and nonenriched external testing. ML algorithms were used for model development. We determined area under the curve (AUC) and applied the Shapley Additive exPlanations (SHAP) framework to interpret risk drivers. RESULTS: Of 4777 patients (mean age 60 [standard deviation: 17] years; 68% women, 91% White, and 86% non-Hispanic) evaluated for penicillin allergy labels, 513 (11%) had positive penicillin allergy testing. Model input variables were frequently missing: immediate or delayed onset (71%), signs or symptoms (13%), and treatment (31%). The gradient-boosted model was the strongest model with an AUC of 0.67 (95% confidence interval [CI]: 0.57-0.77), which improved to 0.87 (95% CI: 0.73-1) when only cases with complete data were used. Top SHAP drivers for positive testing were reactions within the last year and reactions requiring medical attention; female sex and reaction of hives/urticaria were also positive drivers. CONCLUSIONS: An ML prediction model for positive penicillin allergy skin testing using US-based retrospective data did not achieve performance strong enough for acceptance and adoption. The optimal ML prediction model for positive penicillin allergy testing was driven by time since reaction, seek medical attention, female sex, and hives/urticaria.

Incidence and Predictors of In-Hospital Mortality Among Diabetics Undergoing Transcatheter Aortic Valve Replacement.

Objectives The main goals of this study are to delineate the differences in demographics, comorbidities and hospital outcomes between diabetic and non-diabetic aortic stenosis (AS) patients, and next is to evaluate the predictors of in-hospital mortality in AS patients undergoing transcatheter aortic valve replacement (TAVR). Methods We conducted an observational cross-sectional study using the nationwide inpatient sample (NIS) and included 33,325 adult patients with a primary discharge diagnosis of AS who underwent TAVR during the hospitalization. This sample was further grouped by comorbid diabetic which include non-diabetics (N = 23,585) versus diabetic patients (N = 9,740). Among the hospital outcomes we included the length of stay (LOS) and total cost during hospitalization, and the all-cause in-hospital mortality. We used an independent logistic regression model adjusted for demographic confounders to measure the adjusted odds ratio (aOR) of association of comorbid medical conditions and in-hospital mortality risk in non-diabetic and diabetic groups. Results The most prevalent medical comorbidities among inpatients with diabetes were hypertension (85.1%), followed by renal failure (38.0%), chronic lung disease (37.1%), obesity (21.3%), and these values were significantly higher compared with the non-diabetic group. The in-hospitality mortality was higher among the non-diabetic group (4.7%) compared to the diabetic group (2.8%). There was no significant difference in mean length of stay and mean total cost between the diabetic and non-diabetic groups. In diabetic AS inpatients, stroke (aOR: 4.58, 95%CI: 2.23-9.42) and fluid/electrolyte disorders (aOR: 4.25, 95%CI: 3.29-5.48) had a statistically significant association with mortality risk when compared to the non-diabetic group. Among the non-diabetic AS inpatients, fluid/electrolyte disorders had the highest mortality risk (aOR: 2.48, 95% CI 2.17-2.83) followed by coagulopathy (aOR: 2.03; CI: 1.77-2.32), congestive heart failure (aOR: 1.67; CI: 1.40-1.98), and renal failure (aOR: 1.62; CI: 1.41-1.86). Meanwhile, hypertension and obesity had a statistically non-significant and negative association with in-hospital mortality in diabetic and non-diabetic groups. Conclusions Diabetic AS inpatients following TAVR had a higher mortality risk with comorbid fluid/electrolyte disorders and stroke. In-hospital mortality following TAVR was lower among the diabetics compared to non-diabetics, and it underscores diabetes as a surgical risk factor in patients with AS. So, TAVR may be the preferred approach for diabetic patients with AS along with cardiovascular risk factor modification, strict glycemic control and timely renal function follow-up.

Cannabis Use Disorders Lead to Hospitalizations for Peptic Ulcer Disease: Insights From a Nationwide Inpatient Sample Analysis.

Objectives In this study, we aimed to explore the independent association between cannabis use disorders (CUD) and peptic ulcer disease (PUD)-related hospitalization, and then to delineate the demographic differences among PUD inpatients with versus without CUD. Methodology We conducted a cross-sectional study using the Nationwide Inpatient Sample of 50,444,133 patients. We then subgrouped them into PUD and non-PUD cohorts. We compared non-PUD and PUD cohorts using bivariate analysis to delineate the differences in demographics and comorbid risk factors (chronic lung disease, chronic kidney disease, liver disease, diabetes, chronic nonsteroidal anti-inflammatory drug use, tobacco abuse, and alcohol abuse). We used logistic regression analysis to measure the odds ratio (OR) of the association between CUD and PUD-related hospitalization. Results The prevalence of PUD was 0.14% (N = 70,898) among the total inpatient population. It was more prevalent in whites (65%) and males were at higher odds (OR: 1.11; P < 0.001) of being hospitalized for PUD. After controlling for potential comorbid risk factors and demographic confounders, the odds of association between CUD and PUD-related hospitalization were statistically significant (OR: 1.18; P < 0.001). Conclusions CUD was associated with a modest but significant increase of 18% in the likelihood of hospitalization for PUD. With the legalization of cannabis use and its increasing and problematic consumption, it is imperative to understand the impact of cannabis use on the physical health of patients and the related gastrointestinal problems.

Persistent Cannabis Abuse and Risk for Hospitalization for Acute Pancreatitis: A Cross-Sectional Study in United States Hospitals.

Objectives To explore the independent association between cannabis abuse and subsequent hospitalizations for acute pancreatitis (AP) and delineate the demographic differences among AP in patients with and without persistent cannabis abuse. Methods We conducted a retrospective cross-sectional study using the nationwide inpatient sample and included 50,444,133 patients (age 18-50 years) with a primary discharge diagnosis for medical illnesses and further grouped by presence of AP (N = 666,248). We used the logistic regression model to measure the odds ratio (OR) of the association between cannabis abuse and hospitalization for AP and adjusted it for demographic confounders and comorbid risk factors. Results Cannabis abuse significantly increases the odds for AP-related hospitalization (OR 2.12, P <0.001). When the regression model was controlled for potential risk factors (gall stones, cystic fibrosis, hypertriglyceridemia, hypercalcemia, hyperparathyroidism, abdominal surgeries, tobacco abuse, and alcohol abuse), cannabis abuse did not increase the odds for AP-related hospitalization (OR 0.72, P <0.001) due to the significant effect caused by gallstones (OR 30.98, P <0.001) and alcohol abuse (OR 12.69, P <0.001). AP inpatients with cannabis abuse were younger compared to non-cannabis abusers (mean age, 35.7 vs. 37.9 years), and majorly male (70.9% vs. 53.8%). AP was considerably more prevalent in whites (60.6%), followed by blacks (18.3%) and Hispanics (15.2%). Conclusion Cannabis abuse increased the unadjusted odds for AP-related hospitalization by two times, but after controlling for potential risk factors the adjusted odds of association significantly reduced. Cannabis-induced AP can be treated if a problematic recreational cannabis use pattern is discontinued at an earlier stage. Therefore, awareness campaigns and early supportive therapy among cannabis abusers might help diagnose and treat the comorbidity and improve the quality of life.

Demographic Pattern and Mortality Risk Factors for Prescription Opioid Overdose Hospitalizations: Results From Nationwide Inpatient Sample Analysis.

Objectives To explore the demographic patterns of hospitalizations related to prescription opioid overdose (POD) and evaluate the mortality risk of association in POD inpatients. Methodology We conducted a cross-sectional study using the Nationwide Inpatient Sample of 184,711 POD inpatients. A binomial logistic regression model was used to evaluate the odds ratio (OR) of association for mortality risk due to comorbidities (substance use disorders (SUD) and medical complications) in POD inpatients. Results POD inpatients were majorly females (54.1%), older adults aged 51-75 years (48.5%), whites (81.5%), and from lower household income quartet (32.8%). The most prevalent comorbid SUD among POD inpatients was alcohol (15.7%), followed by cannabis (5.7%), cocaine (4.2%), and amphetamine (1.8%). Comorbid alcohol use disorders had a minimally increased association with mortality but were not statistically significant (OR = 1.036; P = 0.438). POD in patients with cardiac arrest had the highest risk of mortality (OR = 103.423; P < 0.001), followed by shock (OR = 15.367; P < 0.001), coma (OR = 13.427; P < 0.001), and respiratory failure (OR = 12.051; P < 0.001). Conclusions Our study indicates that the hospitalizations related to POD were more prevalent among females, elders between 51 and 75 years of age, whites, and those in the lower household income quartet. The prevalence of prescription opioid use and the hospitalization related to POD remains a significant public health issue. POD inpatients with medical complications were at a higher risk of mortality than with comorbid SUD.