Length of Stay Prediction Models for Oral Cancer Surgery: Machine Learning, Statistical and ACS-NSQIP.

OBJECTIVE: Accurate prediction of hospital length of stay (LOS) following surgical management of oral cavity cancer (OCC) may be associated with improved patient counseling, hospital resource utilization and cost. The objective of this study was to compare the performance of statistical models, a machine learning (ML) model, and The American College of Surgeons National Surgical Quality Improvement Program's (ACS-NSQIP) calculator in predicting LOS following surgery for OCC. MATERIALS AND METHODS: A retrospective multicenter database study was performed at two major academic head and neck cancer centers. Patients with OCC who underwent major free flap reconstructive surgery between January 2008 and June 2019 surgery were selected. Data were pooled and split into training and validation datasets. Statistical and ML models were developed, and performance was evaluated by comparing predicted and actual LOS using correlation coefficient values and percent accuracy. RESULTS: Totally 837 patients were selected with mean patient age being 62.5 ± 11.7 [SD] years and 67% being male. The ML model demonstrated the best accuracy (validation correlation 0.48, 4-day accuracy 70%), compared with the statistical models: multivariate analysis (0.45, 67%) and least absolute shrinkage and selection operator (0.42, 70%). All were superior to the ACS-NSQIP calculator's performance (0.23, 59%). CONCLUSION: We developed statistical and ML models that predicted LOS following major free flap reconstructive surgery for OCC. Our models demonstrated superior predictive performance to the ACS-NSQIP calculator. The ML model identified several novel predictors of LOS. These models must be validated in other institutions before being used in clinical practice. LEVEL OF EVIDENCE: Level 3 Laryngoscope, 2024.

Weakly Supervised Learning for Textbook Question Answering.

Textbook Question Answering (TQA) is the task of answering diagram and non-diagram questions given large multi-modal contexts consisting of abundant text and diagrams. Deep text understandings and effective learning of diagram semantics are important for this task due to its specificity. In this paper, we propose a Weakly Supervised learning method for TQA (WSTQ), which regards the incompletely accurate results of essential intermediate procedures for this task as supervision to develop Text Matching (TM) and Relation Detection (RD) tasks and then employs the tasks to motivate itself to learn strong text comprehension and excellent diagram semantics respectively. Specifically, we apply the result of text retrieval to build positive as well as negative text pairs. In order to learn deep text understandings, we first pre-train the text understanding module of WSTQ on TM and then fine-tune it on TQA. We build positive as well as negative relation pairs by checking whether there is any overlap between the items/regions detected from diagrams using object detection. The RD task forces our method to learn the relationships between regions, which are crucial to express the diagram semantics. We train WSTQ on RD and TQA simultaneously, i.e., multitask learning, to obtain effective diagram semantics and then improve the TQA performance. Extensive experiments are carried out on CK12-QA and AI2D to verify the effectiveness of WSTQ. Experimental results show that our method achieves significant accuracy improvements of 5.02% and 4.12% on test splits of the above datasets respectively than the current state-of-the-art baseline. We have released our code on https://github.com/dr-majie/WSTQ.

Dipeptidyl peptidase-4 inhibitors and gallbladder or biliary disease in type 2 diabetes: systematic review and pairwise and network meta-analysis of randomised controlled trials.

OBJECTIVE: To examine the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases. DESIGN: Systematic review and pairwise and network meta-analysis. DATA SOURCES: PubMed, EMBASE, Web of Science, and CENTRAL from inception until 31 July 2021. ELIGIBILITY CRITERIA: Randomised controlled trials of adult patients with type 2 diabetes who received dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, and sodium-glucose cotransporter-2 inhibitors compared with placebo or other antidiabetes drugs. MAIN OUTCOME MEASURES: Composite of gallbladder or biliary diseases, cholecystitis, cholelithiasis, and biliary diseases. DATA EXTRACTION AND DATA SYNTHESIS: Two reviewers independently extracted the data and assessed the quality of the studies. The quality of the evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development and Evaluations framework (GRADE) approach. The meta-analysis used pooled odds ratios and 95% confidence intervals. RESULTS: A total of 82 randomised controlled trials with 104 833 participants were included in the pairwise meta-analysis. Compared with placebo or non-incretin drugs, dipeptidyl peptidase-4 inhibitors were significantly associated with an increased risk of the composite of gallbladder or biliary diseases (odds ratio 1.22 (95%confidence interval 1.04 to 1.43); risk difference 11 (2 to 21) more events per 10 000 person years) and cholecystitis (odds ratio 1.43 (1.14 to 1.79); risk difference 15 (5 to 27) more events per 10 000 person years) but not with the risk of cholelithiasis and biliary diseases. The associations tended to be observed in patients with a longer duration of dipeptidyl peptidase-4 inhibitor treatment. In the network meta-analysis of 184 trials, dipeptidyl peptidase-4 inhibitors increased the risk of the composite of gallbladder or biliary diseases and cholecystitis compared with sodium-glucose cotransporter-2 inhibitors but not compared with glucagon-like peptide-1 receptor agonists. CONCLUSIONS: Dipeptidyl peptidase-4 inhibitors increased the risk of cholecystitis in randomised controlled trials, especially with a longer treatment duration, which requires more attention from physicians in clinical practice. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42021271647.

Association of Glucagon-Like Peptide-1 Receptor Agonist Use With Risk of Gallbladder and Biliary Diseases: A Systematic Review and Meta-analysis of Randomized Clinical Trials.

Importance: Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) have been widely recommended for glucose control and cardiovascular risk reduction in patients with type 2 diabetes, and more recently, for weight loss. However, the associations of GLP-1 RAs with gallbladder or biliary diseases are controversial. Objective: To evaluate the association of GLP-1 RA treatment with gallbladder and biliary diseases and to explore risk factors for these associations. Data Sources: MEDLINE/PubMed, EMBASE, Web of Science, and Cochrane Library (inception to June 30, 2021), websites of clinical trial registries (July 10, 2021), and reference lists. There were no language restrictions. Study Selection: Randomized clinical trials (RCTs) comparing the use of GLP-1 RA drugs with placebo or with non-GLP-1 RA drugs in adults. Data Extraction and Synthesis: Two reviewers independently extracted data according to the PRISMA recommendations and assessed the quality of each study with the Cochrane Collaboration risk-of-bias tool. Pooled relative risks (RRs) were calculated using random or fixed-effects models, as appropriate. The quality of evidence for each outcome was assessed using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) framework. Main Outcomes and Measures: The primary outcome was the composite of gallbladder or biliary diseases. Secondary outcomes were biliary diseases, biliary cancer, cholecystectomy, cholecystitis, and cholelithiasis. Data analyses were performed from August 5, 2021, to September 3, 2021. Results: A total of 76 RCTs involving 103 371 patients (mean [SD] age, 57.8 (6.2) years; 41 868 [40.5%] women) were included. Among all included trials, randomization to GLP-1 RA treatment was associated with increased risks of gallbladder or biliary diseases (RR, 1.37; 95% CI, 1.23-1.52); specifically, cholelithiasis (RR, 1.27; 95% CI, 1.10-1.47), cholecystitis (RR, 1.36; 95% CI, 1.14-1.62), and biliary disease (RR, 1.55; 95% CI, 1.08-2.22). Use of GLP-1 RAs was also associated with increased risk of gallbladder or biliary diseases in trials for weight loss (n = 13; RR, 2.29; 95% CI, 1.64-3.18) and for type 2 diabetes or other diseases (n = 63; RR, 1.27; 95% CI, 1.14-1.43; P <.001 for interaction). Among all included trials, GLP-1 RA use was associated with higher risks of gallbladder or biliary diseases at higher doses (RR, 1.56; 95% CI, 1.36-1.78) compared with lower doses (RR, 0.99; 95% CI, 0.73-1.33; P = .006 for interaction) and with longer duration of use (RR, 1.40; 95% CI, 1.26-1.56) compared with shorter duration (RR, 0.79; 95% CI, 0.48-1.31; P = .03 for interaction). Conclusions and Relevance: This systematic review and meta-analysis of RCTs found that use of GLP-1 RAs was associated with increased risk of gallbladder or biliary diseases, especially when used at higher doses, for longer durations, and for weight loss. Trial Registration: PROSPERO Identifier: CRD42021271599.

Mid-Arm Muscle and Subcutaneous Fat Associated with All-Cause Mortality Independent of BMI: A Prospective Cohort Study.

OBJECTIVE: This study aimed to systematically evaluate the association between triceps skinfold (TSF) thickness (which indicates subcutaneous fat) mid-arm muscle circumference (MAMC; which reflects muscle mass), mid-upper arm circumference (MUAC), and all-cause mortality. METHODS: A total of 17,717 adults from the China Health and Nutrition Survey (1993-2015) were included. Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality. The joint effect of TSF thickness and MAMC was examined, and planned subgroup analyses were performed. RESULTS: The highest quartiles of TSF thickness, MAMC, and MUAC were significantly associated with low all-cause mortality, independent of BMI (TSF thickness: HR = 0.704 [95% CI: 0.575-0.862]; MAMC: HR = 0.729 [95% CI: 0.607-0.876]; MUAC: HR = 0.713 [95% CI: 0.583-0.872]). A 1-SD increase showed comparable risk reductions for TSF thickness and MAMC (14.6% and 14.0%), with 16.1% risk reductions in MUAC. There were positive additive interactions between TSF thickness and MAMC. The inverse association existed in young, middle-aged, and elderly participants (P-heterogeneity > 0.05). CONCLUSIONS: Mid-arm muscle and subcutaneous fat were inversely associated with all-cause mortality, independent of BMI, beyond the elderly population. Mid-arm muscle and subcutaneous fat made comparable contributions to and had positive joint effects on all-cause mortality.

Association of gastro-oesophageal reflux disease and quality of life in patients with chronic rhinosinusitis.

OBJECTIVE: We determined the influence of gastro-oesophageal reflux disease (GERD) on quality of life (QOL) before and after functional-endoscopic-sinus-surgery (FESS) for chronic rhinosinusitis (CRS). METHODS: Medically-recalcitrant CRS patients were recruited prior to FESS. GERD was diagnosed endoscopically. QOL was compared between patients with vs without GERD at baseline and one-year post-FESS. RESULTS: Of 187 CRS patients receiving FESS, 40 had GERD. Pre-operative QOL was significantly worse in CRS patients with vs without GERD. Pre-operative GERD health-related QOL (GERD-HRQL) and reflux symptom index (RSI) scores were both correlated with pre-operative SNOT-22 scores. Compared with non-GERD CRS patients, GERD patients demonstrated larger SNOT-22 improvements after FESS, such that post-operative SNOT-22 values were no longer significantly different between GERD and non-GERD groups. However, post-FESS, in patients with CRS without nasal polyps (unlike those with nasal polyps), the GERD (vs non-GERD) group suffered from greater sleep dysfunction and otologic/facial symptoms. CONCLUSIONS: Compared to CRS patients without GERD, those with GERD experienced poorer pre-operative QOL and greater QOL improvement after FESS.

Stratification and management of patients ineligible for lung cancer screening.

This study identifies participants ineligible for lung cancer screening with the greatest likelihood of future eligibility. Lung cancer risk in participants enrolled in longitudinal lung screening was assessed using the Prostate, Lung, Colorectal and Ovarian lung cancer risk calculator (PLCOm2012) at two timepoints: baseline (T1) and follow-up (T2). Separate analyses were performed on four PLCOm2012 eligibility thresholds (3.25%, 2.00%, 1.50%, and 1.00%); only participants with a T1 risk less than the threshold were included in that analysis. Cox-models identified T1 risk factors associated with screen-eligibility at T2. Three models, applying differing assumptions of participant behavior, predicted future eligibility and were benchmarked against the observed cohort. Nine hundred and fifty-six participants had a T1 risk <3.25%; at 2.00% n= 755; at 1.50% n= 652; at 1.00% n= 484. Lung cancer risk increased over time in most screen-ineligible participants. However, risk increased much faster in participants who became screen-eligible at T2 compared to those who remained screen-ineligible (median per-year increase of 0.35% versus 0.02%, when using a 3.25% threshold). Participants smoking for >30 years, current smokers, less educated participants, and those with chronic obstructive pulmonary disease (COPD) at T1 were significantly more likely to become screen-eligible. New diagnoses of COPD and/or non-lung cancers between T1 and T2 precipitated eligibility in a subset of participants. The prediction model that assumed health behaviors observed at T1 continued to T2 reasonably predicted changes in lung cancer risk. This prediction model and the identified baseline risk factors can identify screen-ineligible participants who should be closely followed for future eligibility.

Association of Factors Influencing Selection of Upfront Hematopoietic Cell Transplantation versus Nontransplantation Therapies in Myelofibrosis.

Despite the curative potential of allogeneic hematopoietic cell transplantation (HCT) for myelofibrosis (MF), a significant number of patients with MF do not undergo HCT. Factors influencing treatment preferences in these patients have not been well studied. This study was conducted to identify patient-, disease-, and donor-related factors influencing the decision regarding HCT in patients with MF. A secondary objective was to compare survival between patients who elected upfront HCT and those who opted for nontransplantation therapy. We conducted a retrospective chart review amongst patients meeting criteria for transplant indication, evaluating clinical characteristics, treatment preferences, and outcomes. Of the 183 study eligible patients age <70 years, 129 (70%) developed an HCT indication. Age >60 years was significantly associated with higher rates of HLA-typing refusal (13 of 72 versus 1 of 44; P = .02). Caucasian ethnicity was significantly associated with an increased rate of identifying well-matched donors compared with non-Caucasian ethnicity (75% versus 48%; P = .02). Of the 69 patients with well-matched donors, 34 (49%) preferred to not pursue upfront HCT despite an indication for transplantation. Patient preference for nontransplantation therapies was the most common reason for declining HCT. We did not find any difference in survival between patients pursuing upfront HCT and those opting for nontransplantation therapies, although more patients in the HCT arm were in remission at the last follow-up. Patients of Caucasian ethnicity were significantly more likely than non-Caucasian patients to identify a well-matched donor. Despite availability of a well-matched donor, a significant proportion of MF patients with an indication for transplantation do not pursue HCT. Patient age, donor type, and patient preference play major roles in the selection of upfront HCT. Although a survival difference was not observed between upfront HCT versus non-transplant therapy, more patients in the HCT arm were in remission at the last follow-up.

Non-operative management for oral cavity carcinoma: Definitive radiation therapy as a potential alternative treatment approach.

PURPOSE: To determine the outcomes of oral cavity squamous cell cancer (OSCC) patients treated with non-surgical approach i.e. definitive intensity-modulated radiation therapy (IMRT). METHODS: All OSCC patients treated radically with IMRT (without primary surgery) between 2005-2014 were reviewed in a prospectively collected database. OSCC patients treated with definitive RT received concurrent chemotherapy except for early stage patients or those who declined or were unfit for chemotherapy. The 5-year local, and regional, distant control rates, disease-free, overall, and cancer-specific survival, and late toxicity were analyzed. RESULTS: Among 1316 OSCC patients treated with curative-intent; 108 patients (8%) received non-operative management due to: medical inoperability (n = 14, 13%), surgical unresectability (n = 8, 7%), patient declined surgery (n = 15, 14%), attempted preservation of oral structure/function in view of required extensive surgery (n = 53, 49%) or extensive oropharyngeal involvement (n = 18, 17%). Sixty-eight (63%) were cT3-4, 38 (35%) were cN2-3, and 38 (35%) received concurrent chemotherapy. With a median follow-up of 52 months, the 5-year local, regional, distant control rate, disease-free, overall, and cancer-specific survival were 78%, 92%, 90%, 42%, 50%, and 76% respectively. Patients with cN2-3 had higher rate of 5-year distant metastasis (24% vs 3%, p = 0.001), with detrimental impact on DFS (p = 0.03) and OS (p < 0.02) on multivariable analysis. Grade ≥ 3 late toxicity was reported in 9% of patients (most common: grade 3 osteoradionecrosis in 6%). CONCLUSIONS: Non-operative management of OSCC resulted in a meaningful rate of locoregional control, and could be an alternative curative approach when primary surgery would be declined, unsuitable or unacceptably delayed.