Final Multivariate Model Research Articles

FOLFIRI-ramucirumab (FOLFIRI-Ram) versus ramucirumab-paclitaxel (Ram-Pac) in the second line (2L) for patients with advanced upper gastrointestinal (UGI) cancer: A real-world propensity-score matched analysis of survival.

377 Background: The approach to 2L treatment of advanced UGI (gastric, esophageal, and gastroesophageal junction) cancers is guided primarily by the results of the RAINBOW trial, a global randomized phase III trial that demonstrated the benefit of Ram-Pac over paclitaxel. However, the onset of peripheral neuropathy limits its use in the palliative-intent setting. The RAMIRIS trial demonstrated the feasibility of 2L FOLFIRI-Ram compared to Ram-Pac, however it failed to meet the prespecified endpoint of ≥ 65% 6-month overall survival (OS). Despite this, FOLFIRI-Ram has emerged as an alternative 2L regimen and now endorsed in major society guidelines. Yet, comparative outcomes between FOLFIRI-Ram and Ram-Pac therapy in the real world are lacking. Methods: This study used the nationwide Flatiron Health electronic health record-derived deidentified database. During the study period (January 2011-June 2024), longitudinal patient-level data were derived from approximately 280 cancer clinics. Patients were eligible for inclusion if they had advanced unresectable or metastatic UGI cancer and received 2L treatment with either FOLFIRI-Ram or Ram-Pac. Demographics and lab values at the time of 2L initiation were extracted. Given anticipated imbalances in sample size, patients were matched 1:6 (FOLFIRI-Ram:Ram-Pac) for inclusion into the analytic dataset using a greedy match based on a logit model to predict propensity scores from key clinical and laboratory characteristics. The primary endpoint was OS from the start of 2L therapy, analyzed using the Kaplan-Meier method, log-rank test, and Cox proportional hazards model. A hybrid approach was used to construct a multivariate Cox model. Results: A total of 15,908 patients were included in the database, of whom 5189 received a 2L treatment. Of these, 631 patients received 2L Ram-Pac and 40 received 2L FOLFIRI-Ram. After matching, the final sample size was 240 patients who received Ram-Pac and 40 who received FOLFIRI-Ram. Baseline clinical and laboratory characteristics were well-balanced. The median OS from initiation of 2L FOLFIRI-Ram was 9.7 months (95% CI 6.9-12.3) and 7.6 months with 2L Ram-Pac (95% CI 6.3-9.2 months). The hazard ratio (HR) for death with FOLFIRI-Ram was 0.77 (95% CI 0.52-1.15, log-rank P=0.20) in reference to Ram-Pac. The adjusted HR death for FOLFIRI-Ram after adjustment for baseline albumin and white blood cell count was 0.87 (95% CI 0.58-1.30, p=0.49) in the final multivariate model. Conclusions: This real-world propensity-score matched analysis of patients with advanced UGI cancers receiving FOLFIRI-Ram or Ram-Pac supports the ongoing use of FOLFIRI-Ram in the 2L as an alternative to Ram-Pac. These data are consistent with results from the RAMIRIS study, however further prospective study is needed.

P-1889. Penicillin Allergy and Its Impact on Clinical Outcomes in OPAT Patients: A Retrospective Cohort Study

Abstract Background 10% of the U.S. population is labeled as allergic to penicillin, with higher rates among women and older adults. These individuals often experience worse clinical outcomes. Despite the extensive use of Outpatient Parenteral Antimicrobial Therapy (OPAT), data on the prevalence of penicillin allergy in this population and its impact on clinical outcomes are scarce. Our study aims to address this knowledge gap by assessing rates of penicillin allergy in OPAT patients and its associations with OPAT-related clinical outcomes. Table 1 Comparison of characteristics and complications among OPAT patients considering penicillin allergy status Methods We conducted a retrospective cohort study at Tufts Medical Center, including all patients aged ≥18 years discharged on OPAT between April 2022 and October 2022. Data on demographics, penicillin allergy status, complications, and outcomes were extracted through chart reviews. The primary outcomes were: (1) rates of penicillin allergy, and (2) association between penicillin allergy and OPAT complication rate (line- or antibiotic-related), OPAT extension, 30-day ED visits, and readmission rates. We used a multivariable logistic regression model to examine these associations, adjusting for age, sex, race, ethnicity, and Charlson comorbidity index. Two-sided p values were considered statistically significant at .05. Table 2 Association of documented penicillin allergy and complications related to OPAT Results A total of 231 patients received OPAT; 39 (17.9%) had documented penicillin allergy. Females and white patients had higher rates of penicillin allergy compared to their counterparts; 24.4% vs 12.4% and 19.1% vs 9.4%, respectively (Table 1). In the final multivariable logistic regression model, penicillin allergy status was not associated with OPAT extension (adjusted odds ratios[aOR] 0.83; 95% confidence interval[CI] 0.36-1.89), OPAT-related complications (aOR 1.04; 95%CI 0.42-2.57), 30-day ED visits (aOR 0.38, 95%CI 0.05-3.13), and 30-day readmissions (aOR 0.24; 95%CI 0.03-1.87) (Table 2). Conclusion Our findings suggest that the rates of PCN allergy were higher than in hospital settings (∼15%), but these do not significantly influence the OPAT clinical outcomes. However, neither group met the required 248 subjects to reliably detect a 10% increase in complication rates, limiting the significance of the findings. Further research with a larger sample size is needed to validate these findings. Disclosures Shira Doron, MD, Medtronic: Expert Testimony|Sunovion: Advisor/Consultant|Vertex: Speaker

Association between socioeconomic status and nonurgent presentations to pediatric emergency departments: a retrospective study.

Access to healthcare remains a persistent challenge. Socially disadvantaged populations often encounter barriers to care and may frequently seek out emergency departments (EDs), including for nonurgent medical care. The objective of this study is to study the association between nonurgent presentations to pediatric EDs and patients' socioeconomic environment in an urban setting. A retrospective study of all visits to a pediatric ED in eight participating centers of the Paris metropolitan area (France) between 1 January 2017 and 31 December 2021 was carried out. Routinely collected data were analyzed. Socioeconomic status was evaluated using ecological variables defined at the municipality level. These variables were collected from public sources and included a social deprivation index, the accessibility to general practitioners, the proportion of single-parent families, and the proportion of immigrants. The primary endpoint was a nonurgent ED presentation, defined as being assigned one of the two lowest triage categories on a five-point scale. A multilevel logistic model assessed the association between nonurgent ED presentations and patients' characteristics, socioeconomic environment, and healthcare accessibility. Nonurgent visits accounted for 51.6% of the 1 499 108 visits during the study period. The admission rate was 2.1% for nonurgent presentations and 18.8% for urgent presentations. In the final multivariate model (n = 1 412 895 visits), after adjustment for sex, age, time of day, day of the week, month, and year, the risk of nonurgent presentation was significantly higher for children living in less advantaged areas and in areas where the rate of single-parent families was high. It was also higher for children living close to the ED. This extensive retrospective multicenter study emphasizes the increased risk of visiting EDs for nonurgent medical care among children from disadvantaged urban areas.

P1197 Progression to colorectal cancer or high-grade dysplasia in individuals with inflammatory bowel disease and low-grade dysplasia: A cohort study of risk factors and incidence

Abstract Background Individuals with inflammatory bowel disease (IBD) face an increased risk for colorectal cancer (CRC) and high grade dysplasia (HGD) compared with individuals without IBD.1,2 Low-grade dysplasia (LGD) is recognized as a key risk factor for HGD and CRC development in IBD.3 However, the risk for progression from LGD to HGD or CRC is still debated. Studies report diverging results and are often limited by small sample sizes and selection bias.4 The influence of diagnosis-year and medications is particularly underexplored. Therefore, we sought to utilize the Danish healthcare registers to examine the impact of candidate risk factors for progression from LGD to either HGD or CRC among Danish individuals with IBD. Methods We conducted a cohort study, including all Danish individuals diagnosed with IBD and LGD between 1977 and 2022. IBD with LGD was defined as individuals with a LGD diagnosis up to 180 days before or any time after their first IBD diagnosis, using the latest date as index date. Individuals with a colectomy, HGD, or CRC before the index date were excluded. We investigated the cumulative incidence of progression to HGD or CRC, treating death as a competing risk. A univariate Cox regression model was used to assess hazard ratios (HRs) for potential risk factors. Variables with a P-value of ≤ 0.1 were included in the final multivariable cox regression model. Results We included 7,082 individuals with IBD and LGD for further analysis. As illustrated in figure 1 the cumulative incidence of progression from LGD to HGD or CRC at 15 years post-index was 6.70% (95% CI: 5.93–7.54). Several risk factors were identified (table 1). Of particular interest was the influence of index-year and treatment with prednisone and anti-TNF-α. In the final model, we observed a reduction in progression risk over time, with a decreased risk for progression in the period 2018-2021 (HR: 0.32 [95% CI: 0.21-0.50]) and 2015-2018 (HR: 0.53 [95% CI: 0.34-0.69]) relative to 1977-2010. High-dose prednisone treatment was associated with progression to HDG or CRC (HR: 1.77 [95% CI: 1.23-2.57]), whereas anti-TNF-α treatment was not (HR: 0.81 [95% CI:0.38–1.74]). Conclusion Among Danish individuals with IBD the risk of progression from LGD to HGD or CRC declined over time. Treatment with High-dose prednisone was associated with an increased risk of progression, whereas anti-TNF-α treatment was not. We speculate that recent advancements in biological treatments of IBD, the implementation of 'treat-to-target' strategies, and improved detection and resection of LGD lesions, may partly explain declining progression risk.

P0397 Nutritional status and the risk of proctectomy in patients with perianal fistulising Crohn’s disease

Abstract Background Perianal fistulas are the most common type of anal involvement in Crohn’s disease (CD), affecting approximately one-third of the patients.1 They are associated with a poor quality of life and can lead to significant disability.2 Despite combined medical and surgical treatments, many patients experience refractory disease, with up to 20% requiring proctectomy, which results in a permanent stoma.3 Consequently, investigating factors that may influence the outcomes of perianal disease is a critical area of research in CD. One such factor that remains to be explored is the impact of nutritional status on fistula healing and the overall prognosis of patients with perianal CD.4 This study aimed to evaluate how nutritional status influences the risk of undergoing proctectomy. Methods A retrospective cohort study was conducted with 64 patients with CD presenting chronic symptomatic anorectal fistulas. These patients underwent seton placement at two healthcare centres in Brasília, DF, Brazil, from January 2010 to December 2022. Data were extracted from the electronic patient records, including demographic and clinical characteristics. At the first hospital admission, nutritional risk screening was evaluated by applying the NRS-2002. The nutritional status was assessed retrospectively using the GLIM criteria. The prevalence of malnutrition and its association with the risk of proctectomy were analysed. Binary logistic regression with univariate analysis was used to screen independent risk factors for building a final multivariate model by including variables with clinical or statistical significance. Results Seventeen out of sixty-four patients were diagnosed with malnutrition (26.6%). The median follow-up was 53 months. The malnourished group had a higher prevalence of proctitis (100% vs 78.7%, p= 0.038) and lower mean albumin (34.7 vs 39.7 g/L, p= 0.002) and haemoglobin levels (10.9 vs 12.5 g/dL, p=0.003). Malnourished patients were younger at diagnosis (24.2 vs 30y, p=0.046) and presented lower BMI (19.7 vs 24.9 kg/m², p<0.001), higher Harvey-Bradshaw Index levels (10 vs 5, p<0.001) and rates of corticosteroid therapy (29.4% vs 4.3%, p= 0.012). The proctectomy rate was 47.1% in the malnourished group compared to 14.9% in non-malnourished patients (p=0.007). The binary logistic regression results indicated that malnutrition (OR=6.24; 95% CI= 1.27 - 30.57; p=0.024) was an independent risk factor for proctectomy. Conclusion Malnutrition, as assessed by GLIM criteria, was a significant risk factor for a proctectomy procedure in patients with CD presenting chronic symptomatic perianal fistulas. Future research should examine whether nutritional interventions can improve the prognosis of patients with perianal fistulising CD.

Micronutrient status and risk of Mycobacterium tuberculosis infection in Indonesian tuberculosis case contacts.

Certain micronutrient levels have been associated with the risk of developing TB disease. We explored the possible association of selected at-risk micronutrient levels with the development of Mycobacterium tuberculosis (M.tb) infection. This cohort study in Bandung, Indonesia, followed Interferon Gamma Release Assay (IGRA) negative TB case contacts with a repeat IGRA test at 3 mo. At baseline, blood was analysed for haemoglobin, 25-hydroxyvitamin D, retinol-binding protein, C-reactive protein, alpha-1-acid glycoprotein, serum transferrin receptor (sTfR), ferritin, zinc and selenium. Total body iron was calculated using ferritin and sTfR status. Associations between case contact micronutrient concentration and IGRA conversion were estimated using Poisson regression. Of 430 contacts, 115 (27%) underwent IGRA conversion. Ferritin concentration (adjusted for inflammation) was positively associated with risk of IGRA conversion (incidence rate ratio [IRR] for ferritin=1.17; 95% CI 1.01 to 1.35; p=0.03), but other select micronutrients were not. This association held for ferritin in the final multivariable model (IRR=1.27; 95% CI 1.09 to 1.47; p=0.002). The risk of developing M.tb infection, as defined by IGRA conversion, is associated with increasing ferritin. Interventions in TB case contacts to temporarily reduce iron levels, including considering withholding any iron supplementation, may be worthy of evaluation.

Anew harmony box supplemented with gonial angle and age based on agrowing Swiss population.

The scope of the present study was to create anew harmony box by adding two diagnostically and clinically important cephalometric variables, the gonial and interincisal angles, while also considering the effect of sex and age for agrowing Swiss population. Ahealthy sample with an overjet and overbite between 2and 4 mm, and 1.5and 4.5 mm, respectively, of the Zurich Craniofacial Growth Study was considered. Pairwise correlations between the cephalometric angles were evaluated with the Pearson correlation coefficient(r). Regression models were built for each cephalometric variable serving as the dependent one. The Akaike Information Criterion and the Bayesian Information Criterion were used to structure and select the final multivariable regression model. Bland-Altman plots and the Lin's concordance correlation coefficient were used to assess inter- and intraexaminer agreement. The sample consisted of 301 individuals with amean age of 13.4 years. Measurement concordance was confirmed both within and between examiners. The strongest correlations were observed between the angles SNB and SNA (r = 0.81), ArGoMe and SN-ML (r = 0.57), and SN-ML and SNB (r = 0.56). The SNB angle qualified as the dependent variable in the multivariable regression that framed the newly provided harmony box, with the predictor variables age (p < 0.001) and the angles SNA (p < 0.001), SN-ML (p < 0.001), SN-NL (p = 0.005), NSBa (p = 0.001), and ArGoMe (p < 0.001). The interincisal angle did not increase the robustness of the model and was excluded (p > 0.05). Contrary to the interincisal angle, gonial angle and age qualified for inclusion in the new harmony box for individualized cephalometrics in asample of healthy schoolchildren from Zurich, Switzerland.

How Antiretroviral Drug Concentrations Could Be Affected by Oxidative Stress, Physical Capacities and Genetics: A Focus on Dolutegravir Treated Male PLWH.

High levels of reactive oxygen species (ROS) are present in people living with HIV (PLWH), produced by intense physical activity; in response, our body produces antioxidant molecules. ROS influence the expression of gene-encoding enzymes and transporters involved in drug biotransformation. In addition, pharmacogenetics can influence transporter activity, and thus drug exposure. Currently, no studies concerning this topic are present in the literature. The aim of this study was to investigate whether some antioxidant molecules, physical exercise, and genetic variants could affect dolutegravir (DTG) concentrations in PLWH, switching from triple to dual therapy. Thirty PLWH were recruited and analyzed at baseline (triple therapy), and 6 months after (dual therapy). Physical capacities were investigated using validated tools. Drug concentrations and oxidative stress biomarkers levels were evaluated through liquid chromatography coupled with tandem mass spectrometry, while genetic variants through real-time PCR. No statistical differences were suggested for drug concentrations, with the exception of intracellular DTG (p = 0.047). Statistically significant correlations between DTG plasma concentrations and white blood cells (p = 0.011; S = 0.480) and cytoplasmic N-acetyl-cysteine (p = 0.033; S = -0.419) were observed. Finally, white blood cells and BMI remained in the final multivariate regression model as predictors of DTG concentrations. This is the first study showing possible factors related to oxidative stress impacting DTG exposure.

Factors Associated With the Recovery of Left Ventricular Ejection Fraction in Patients With Anthracycline-Induced Left Ventricular Dysfunction.

Neurohormonal blocking drugs, like beta-blockers, angiotensin-converting enzyme inhibitors (ACEIs), and angiotensin receptor blockers (ARBs), are recommended for treating anthracycline-induced left ventricular dysfunction (AILVD). However, there is limited evidence supporting their benefit. Therefore, this study evaluated associations of neurohormonal blockers and other clinical factors with recovery of left ventricular ejection fraction (LVEF) in patients with AILVD. This retrospective chart review assessed patients treated with at least one dose of anthracycline, then had ≥10% LVEF reduction or post-anthracycline LVEF value <50%, and then had a follow-up LVEF measurement ≥90 days later. The primary endpoint was LVEF recovery (highest follow-up LVEF-lowest LVEF post-anthracycline). Variables from univariable tests with P < .1 were incorporated in a multiple linear regression model for independent factors significantly associated with LVEF recovery (P < .05). Out of 104 patients, 83% were female, 86% self-reported white race, 53% had breast cancer, median (IQR) age was 52 (22) years, and LVEF recovery was 14% (16%). The final multivariable model included 2 significant variables: beta-blocker dose after anthracycline exposure (every 25 mg increase in beta-blocker dose was associated with 5.0% increase in LVEF recovery; P = .0005) and the time between the start of the anthracycline and the lowest LVEF post-anthracycline (every 5-year increase in time was associated with 1.8% decrease in LVEF recovery; P = .0379). In patients with AILVD, a higher beta-blocker dose and earlier detection of LVEF reduction post-anthracycline were significantly and independently associated with improved LVEF recovery. These findings need to be validated in a larger, independent cohort.

Dermal Exposure to Agrochemicals as Risk Factor for Skin Cancer in Farmers and Ranchers in the US Central States.

Farm operators are at a high risk of developing skin cancer due to their occupational sun exposure. With the growing incidence of skin cancer, it is also important to evaluate other occupational risk factors. Farm operators confront numerous physical, chemical, and biological hazards in their work environment. This study investigated whether dermal exposures to pesticides/fertilizers, animals/livestock, detergents/disinfectants, and fuels/solvents/paints were associated with the risk of skin cancer in farm and ranch operators. Surveillance data from the Central States Center for Agricultural Safety and Health (CS-CASH) Farm and Ranch Health and Safety Surveys in 2018 and 2020 were used to explore the risk of skin cancer in farm operators in seven US central states. Farm production variables from the DTN Farm Market database were merged with survey responses. The associations of skin cancer and exposure variables were analyzed using descriptive statistics and regression modeling. The prevalence of skin cancer was 10% among 7943 operators. Univariable analyses showed that men had 1.62 times higher odds of skin cancer compared to women. The odds of skin cancer increased significantly with age. Livestock, fed cattle, cow-calf, and beef production increased the odds of skin cancer. Exposure to pesticides/fertilizers and fuels/solvents/paints also increased the odds of skin cancer compared to unexposed operators. In the final multivariable model, the associations of skin exposure to pesticides/fertilizers (odds ratio (OR) = 1.30, 95% CI: 1.08-1.56) and to fuels/solvents/paints (OR = 1.21, 95% CI: 1.01-1.45) remained statistically significant after adjusting for sex, age, and state. Having livestock also increased the odds of skin cancer (OR = 1.18, 95% CI: 1.00-1.38). Skin is a critical source of occupational exposures among farm operators. Increased odds of skin cancer in this study emphasizes the need for better protection against exposures to chemicals including pesticides/fertilizers, and fuels/solvents/paints.

Standing balance impairment in persons with type 2 diabetes is predicted by peripheral neuropathy and vestibulopathy.

The three main sensory complications of Type 2 Diabetes are diabetic peripheral neuropathy (DPN), diabetic retinopathy (DR), and diabetic vestibulopathy (DV). Current screening for these has been found to be insufficient, and balance testing may be a way to improve screening practices. This study aims to assess if the presence of sensory complications can be used to predict balance scores. 52 participants with T2D were recruited for this study. Participants anthropometric measures, demographic, socioeconomic, and information related to the participants T2D were recorded. Participants completed the modified Clinical Test of Sensory Integration in Balance (mCTSIB) and scores were recorded. Bivariate analysis was conducted on all variables related to mCTSIB total balance scores. A multivariable linear regression model was created using mCTSIB total scores as the outcome and T2D sensory complications as the predictor while controlling for sex and age. Bivariate analysis revealed significant (p < 0.05) associations between mCTSIB total scores and T2D sensory complications, education, employment, annual income, and age. The final multivariate linear regression model was found to be significant (Adj R2 = 0.45, p < 0.01). This model showed that those with DPN (β = 121.03 cm, CI = 77.71-164.35, p < 0.01) or DV (β: 60.65, CI = 5.17-116.13, p = 0.04) had significantly higher balance scores compared to those that did not have a sensory complication, adjusting for sex and age. Sensory complications of T2D have the potential to predict balance scores and provides the first evidence that balance assessments may also be able to screen for DPN and DV.

Patient-Reported Distress in Individuals With Head and Neck Cancer

Distress is common among patients with cancer, and evidence of disparities associated with distress has been mixed. Head and neck cancer (HNC) is one of the most emotionally distressing cancers and is also a highly disparate disease. However, it is unknown whether there are disparities associated with patient-reported distress in HNC. To examine racial and sociodemographic factors associated with clinically meaningful distress in a cohort of patients with HNC. This retrospective cohort study analyzed data from a single tertiary institution at a National Cancer Institute-designated comprehensive cancer center in the Southeastern US. Participants included adult patients from January 2017 to December 2022 with confirmed diagnoses of head and neck squamous cell carcinoma who received radiation therapy up to 5 weeks from initial diagnosis. The outcome of interest was clinically meaningful distress (score of at least 4 on the distress thermometer). The distress thermometer is a single-item, self-reported tool used to measure a person's level of psychological distress over the past week, represented on a scale from 0 (no distress) to 10 (extreme distress). Patients were included in the study if they had visits with clinicians in the radiation oncology department within 5 weeks of diagnosis. Sociodemographic factors were race and ethnicity, age, sex, marital status, and health insurance status. Clinical variables (stage of presentation, anatomical subsites, smoking, and alcohol history) and problem list domain items were included in the adjusted model. Using multivariable logistic regression analysis, odds of meaningful distress by sociodemographic factors were estimated, adjusting for clinical factors and problem list domains. A total of 507 patients met the inclusion criteria. The study population included 389 male patients (76.7%). The median (IQR) age of participants was 63 (56-71) years, with a racial distribution of 89 Black patients (17.6%), 385 White patients (75.9%), and 33 patients of other categories (6.9%), including Asian, Native American, multiracial, declined to respond, and unknown. Overall, 232 patients (45.8%) had meaningful distress. The median (IQR) DT score for the entire cohort was 3 (0-6). Black patients had the highest rate of physical concerns (n = 31 [34.8%]) compared to the other racial groups (78 White patients [20.3%] and 7 persons of other race [21.2%]). However, in the final multivariable model, race was not significantly associated with clinically meaningful distress (Black compared with White: adjusted odds ratio [aOR], 0.76 [95% CI, 0.45-1.28]; other compared with White: aOR, 0.85 [95% CI, 0.37-1.94]). Compared to those married, unmarried patients were significantly more likely to report distress (aOR, 1.61 [95% CI, 1.05-2.50]). Also, patients with emotional problems had double the odds of reporting clinically meaningful distress (aOR, 2.03 [95% CI, 1.02-4.08]). Other factors associated with significant clinical distress included tobacco use (aOR, 2.14 [95% CI, 1.02-4.50]) and reported practical problems (aOR, 2.08 [95% CI, 1.17-3.69]). Independent of race or sociodemographic factors, the results of this retrospective cohort study underscore the need for social and emotional support in mitigating distress and optimizing mental health care in this patient population. Further studies should explore distress trajectories across the HNC continuum and their impact on HNC outcomes.

Sample Size Requirements for Popular Classification Algorithms in Tabular Clinical Data: Empirical Study

Background The performance of a classification algorithm eventually reaches a point of diminishing returns, where the additional sample added does not improve the results. Thus, there is a need to determine an optimal sample size that maximizes performance while accounting for computational burden or budgetary concerns. Objective This study aimed to determine optimal sample sizes and the relationships between sample size and dataset-level characteristics over a variety of binary classification algorithms. Methods A total of 16 large open-source datasets were collected, each containing a binary clinical outcome. Furthermore, 4 machine learning algorithms were assessed: XGBoost (XGB), random forest (RF), logistic regression (LR), and neural networks (NNs). For each dataset, the cross-validated area under the curve (AUC) was calculated at increasing sample sizes, and learning curves were fit. Sample sizes needed to reach the observed full–dataset AUC minus 2 points (0.02) were calculated from the fitted learning curves and compared across the datasets and algorithms. Dataset–level characteristics, minority class proportion, full–dataset AUC, number of features, type of features, and degree of nonlinearity were examined. Negative binomial regression models were used to quantify relationships between these characteristics and expected sample sizes within each algorithm. A total of 4 multivariable models were constructed, which selected the best-fitting combination of dataset–level characteristics. Results Among the 16 datasets (full-dataset sample sizes ranging from 70,000-1,000,000), median sample sizes were 9960 (XGB), 3404 (RF), 696 (LR), and 12,298 (NN) to reach AUC stability. For all 4 algorithms, more balanced classes (multiplier: 0.93-0.96 for a 1% increase in minority class proportion) were associated with decreased sample size. Other characteristics varied in importance across algorithms—in general, more features, weaker features, and more complex relationships between the predictors and the response increased expected sample sizes. In multivariable analysis, the top selected predictors were minority class proportion among all 4 algorithms assessed, full–dataset AUC (XGB, RF, and NN), and dataset nonlinearity (XGB, RF, and NN). For LR, the top predictors were minority class proportion, percentage of strong linear features, and number of features. Final multivariable sample size models had high goodness-of-fit, with dataset–level predictors explaining a majority (66.5%-84.5%) of the total deviance in the data among all 4 models. Conclusions The sample sizes needed to reach AUC stability among 4 popular classification algorithms vary by dataset and method and are associated with dataset–level characteristics that can be influenced or estimated before the start of a research study.

Epidemiology of Latent Tuberculosis in Moroccan Healthcare Workers: Prevalence Patterns and Predisposing Factors.

Among occupational hazards in healthcare settings, latent tuberculosis infection (LTBI) ranks as a major concern, particularly threatening healthcare workers (HCWs) in nations grappling with intermediate to high tuberculosis (TB) rates. Our study was conducted in Morocco, a country characterized by widespread Bacillus Calmette-Guérin (BCG) vaccination and a moderate TB burden of 93 cases per 100,000 inhabitants in 2022. We examined both the prevalence of LTBI among Moroccan HCWs and its various risk factors. A cross-sectional study was conducted from August 2022 to October 2024 in two Moroccan hospitals. One hundred forty-seven HCWs were recruited and screened for LTBI using the QuantiFERON-TB Gold In-Tube (QFT-GIT) test. Data on demographics, occupational characteristics, and potential risk factors were collected through standardized questionnaires. Statistical analysis included univariate and multivariate logistic regression to identify factors associated with LTBI. The overall prevalence of LTBI was 32.65% (48/147). Multivariate analysis identified several independent risk factors: male gender (OR: 2.84; 95% CI: 1.54-5.22; p<0.001), age above 50 years (OR: 4.58; 95% CI: 1.50-13.90; p=0.007), smoking status (OR: 4.07; 95% CI: 1.63-10.14; p=0.003), and family history of TB (OR: 4.71; 95% CI: 1.33-16.65; p=0.016). Notably, neither specific work areas nor job categories were identified as significant risk factors in the final multivariate model. The observed LTBI prevalence among HCWs in Morocco demonstrates concordance with epidemiological data from comparable intermediate-burden nations. The elucidation of predisposing factors, with particular emphasis on non-occupational determinants, underscores the imperative for implementing systematic surveillance protocols and World Health Organization (WHO)-sanctioned infection prevention measures within healthcare facilities. These epidemiological findings provide an empirical foundation for the formulation and optimization of TB control strategies specifically tailored to the occupational health needs of Morocco's healthcare workforce.

Incidence of splenic malignancy and hemangiosarcoma in dogs undergoing splenectomy surgery at a surgical specialty clinic: 182 cases (2017-2021).

The objectives of this study were to evaluate the risk and predictive factors of splenic malignancy and hemangiosarcoma in dogs undergoing splenectomy at a surgical specialty clinic. Medical records, hematologic results, surgical reports, and histopathologic results from 182 dogs that underwent splenectomy for the treatment of splenic masses or nodules were reviewed retrospectively. The majority of dogs (57.7%) had benign splenic diagnoses with no malignancy. Hemangiosarcoma was diagnosed in 32.4% of the dogs. A final multivariable model indicated that thrombocytopenia, anemia, and a smaller diameter of the largest splenic nodule were risk factors for hemangiosarcoma (P<0.001), and hemoperitoneum (P = 0.01) was an additional risk factor when nodule diameter was not evaluated. There were 91 dogs that had hemoperitoneum, and 60.4% of those dogs had malignant splenic lesions. Of the 33 dogs that underwent a splenectomy for incidentally identified splenic lesions, 93.9% had benign splenic lesions. Breed size was not a significant predictor of splenic malignancy risk; however, all 6 of the German shepherds included in the study had a hemangiosarcoma diagnosis. Overall prevalence of splenic malignancy including HSA may be overestimated in some canine populations.

Digital dermatitis in dairy cattle in Egypt: herd-level risk factors and Treponema spp. prevalence across lesion M-scores.

Digital dermatitis (DD) is an infectious disease of the digital skin of dairy cows that is associated with compromised animal welfare and significant economic losses. The hind feet of 16,098 dairy cows from 55 herds were examined in the milking parlor, and DD lesions identified were classified using the M-score system and swabbed for PCR testing. Swabs were also collected from hind feet with normal digital skin for comparison. Data on farm construction and lameness prevention measures were collected through direct observation and in-person interviews with farm managers. Herd-level risk factors for DD were investigated using mixed-effects logistic regression models, with Farm ID included as a random-effects variable. The average within-herd and herd prevalence of DD were 6.4% (95% CI: 4.9-8.0%) and 92.7% (n=51, 95% CI: 85.9-99.6%), respectively. Among the investigated DD lesions, DD-associated treponemes were identified at frequencies of 87% for Treponema spp., 87.5% for T. medium, 86.2% for T. phagedenis-like, and 78.3% for T. pedis. Additionally, Fusobacterium necrophorum and Dichelobacter nodosus were identified at frequencies of 37% and 15.4%, respectively. Samples from normal digital skin exhibited substantially lower frequencies for all bacterial species investigated. Risk factors significantly associated with DD in the final multivariable model included practice of preventive early lactation trimming (OR=0.45, 95% CI: 0.26-0.78) and the presence of an abrasive bedding on farms (OR=1.6, 95% CI: 1.01-2.6). This comprehensive molecular and epidemiological analysis enhances understanding of DD etiology in Egypt and contributes to the development of effective control programs to improve cow welfare and support the sustainability of milk production.

History of Breast Cancer in Patients with Oral Lichen Planus: A Case-Control Study.

Objectives: The purpose of this study was to determine the association between oral lichen planus (OLP) and the history of cancer outside of oral cavity and the predominance of its location. Methods: This case-control study included 600 OLP patients and 600 control subjects evaluated in the same section, matched for age and sex to the OLP patients. OLP patients were diagnosed clinically and histologically. Initially, the prevalence of the most frequent types of cancers was described. A Pearson chi-squared test model was used to determine the association of cancer history and OLP. It was considered statistically significant whether p value was ≤0.05. The final multivariate regression model was built after applying a backward selection method to the complete multivariate model considering the odds ratio (OR) with a 95% confidence interval (CI). Results: The history of cancer was significantly associated with OLP regardless of age, sex, tobacco and or alcohol use in both univariate [OR = 2.26 (95%CI: 1.26-4.24); p = 0.008] and multivariate analyses [OR = 2.21 (95%CI: 1.21-4.19); p = 0.012]. According to the location of cancer, there was an association between OLP and history of breast cancer [OR = 3.71 (95%CI = 1.03-13.38); p = 0.032]. Conclusions: This case-control study suggests a higher frequency of cancer, particularly breast cancer, among patients with OLP compared to the control group. Due to the study's design and sample limitations, these findings should be interpreted cautiously. Future longitudinal, multi-institutional studies with rigorous control for cancer history and other confounding factors are essential to further explore the association between OLP and cancer, particularly breast cancer.

Predicting long-term functional anti-VEGF treatment outcomes in neovascular AMD in a real-world setting.

To test to what degree retinal fluid (RF) after the loading phase and at the end of year 1 predicts long-term functional outcomes in neovascular macular degeneration (nAMD), as do macular (MA) atrophy, treatment density and treatment interval extension. In this retrospective single-center cohort study, a consecutive series of eyes with treatment-naïve nAMD followed under a treat-and-extend (T&E) protocol followed over ≥2 years. Best-corrected visual acuity (BCVA), presence of retinal fluid (RF) and macular atrophy (MA) were registered along with central retinal thickness (CRT) and treatment density over time. The relationship between these variables was tested by regression analysis. A total of 433 eyes were followed for 4.9 ± 2.2 years. A series of univariate analyses were run to select the covariates for the final multivariate regression model. CRT after loading, time to dryness, intraretinal fluid and MA after one year were found to predict visual function over 2 to 5 years. A final regression model was adjusted for visual acuity (VA) at baseline and showed that CRT after loading was predictive only in the short term (2 years) and that MA had the greatest predictive value for VA after 2 to 5 years. Intraretinal fluid (IRF) significantly predicted VA only after 4 years. The final regression model explained 21 to 32% of the variation in VA. In this large retrospective cohort, the presence of MA after one year was the strongest predictor of VA after 2 to 5 years, explaining a vision loss of 13 to 20 letters. The presence of IRF and SRF at any point of time had a comparably weak predictive potential for the outcomes over 5 and more years.

Cabotegravir Maintains Protective Efficacy in the Setting of Bacterial STIs: A Secondary Analysis of HPTN 083.

Sexually transmitted infections (STIs) have been shown to facilitate HIV transmission and acquisition. HPTN 083, a global clinical trial, demonstrated superiority of long-acting cabotegravir (CAB-LA) versus daily oral tenofovir disoproxil fumarate/emtricitabine (TDF/FTC) for HIV prevention among transgender women and cisgender men who have sex with men. This analysis assessed whether CAB-LA maintained protective efficacy when bacterial STIs (syphilis, rectal/urethral gonorrhea and chlamydia) were present. STI events per 100 person-years (PY) were calculated, including by subgroups (age, race/ethnicity, gender, education, treatment arm, drug use, alcohol use, region/country, condom usage, partner number, marital status, baseline STI). Association between baseline factors and STI incidence was modeled using Poisson regression. Cox proportional hazards modeling with STI status as a time-varying covariate was used to evaluate potential interactions between STI status and the relative efficacy of CAB-LA vs. TDF/FTC. Among 3,859 participants, overall STI incidence rate was 50.7 infections/100PY. STIs were diagnosed in 1,562 (40.5%) participants; 79% of STIs occurred in 25% of the participants. STI incidence was not different by PrEP arm. In the final multivariable model, age, region, race, education level, marital status, and baseline STI were associated with incident STI (p<0.05). HIV incidence was lower with CAB-LA vs. TDF/FTC with or without STIs (hazard ratios 0.37 and 0.31, respectively), with no significant interaction between STIs and the HR for HIV incidence (p = 0.75). In a large PrEP trial with high STI incidence, CAB-LA maintained robust protective efficacy relative to TDF/FTC in the setting of bacterial STIs.

Risk factors of fatigue among community-dwelling older adults in Bahir Dar, Northwest Ethiopia: a community-based cross-sectional study.

Fatigue is defined as subjective fatigue and a decline in physical and mental activity that does not improve with rest. Fatigue among older adults could lead to future comorbidity, mortality, decreased social interaction, greater strain on families, decreased productivity, and a higher need for hospitalization and rehabilitation. However, no studies have been conducted in Africa, particularly in Ethiopia. Therefore, this study aimed to evaluate the prevalence and factors of fatigue among older adults. A community-based cross-sectional study of 605 older adults was carried out using a single-stage cluster sampling technique. The Chalder Fatigue Scale (CFS) was used to assess fatigue, and data were collected through an interview. The collected data were coded, cleaned, and entered into EpiData version 4.6 and exported to SPSS Version 25 for analysis. Bivariate and multivariate logistic regression analyses were performed. Variables in the final multivariate logistic regression model with a 95% confidence interval (CI) and a p-value of 0.05 were considered statistically significant. The prevalence of fatigue among older adults was 37.9% (95% CI, 34-41.90). Significant risk factors included older age [adjusted odds ratio (AOR) = 6.13, CI = 3.25-11.58], the presence of two or more comorbidities (AOR = 5.68, CI = 2.97-10.83), physical inactivity (AOR = 3.33, CI = 1.56-7.12), poor social support (AOR = 2.83, CI = 1.61-4.96), insomnia (AOR = 5.48, CI = 3.38-8.88), and depression (AOR = 2.65, CI = 1.60-4.36). The prevalence of fatigue among older adults was noticeable, and it was summarized as a public health issue among older adults in the study area. Our study findings revealed that older age, the presence of comorbidities, physical inactivity, poor social support, insomnia, and depression were all risk factors for fatigue among community-dwelling older adults.