Independent predictors from stepwise logistic regression may be nothing more than publishable P values

Relationship between admission selection tools and student attrition in the early years of medical school

Objectives

Placement in medical schools is highly sought after worldwide with fierce competition among applicants. However, some of the best students withdraw after being accepted to medical school. The aim of this study was to investigate early student attrition within the first 2 years of medical school and determine its relationship to admission selection tools.

Methods

Quantitative research was conducted at the College of Medicine and Health Sciences from 2016 until 2020, during which time routine admission data and students' examination results for the first 2 years were collected and analyzed.

Results

The attrition rate during the study period was 31.7%. High school and college written examination scores were significantly related to completing the premedical program (p = 0.001 and p = 0.002, respectively). Female students scored significantly higher in multiple mini interviews (MMIs) compared with male counterparts (p < 0.001). However, the difference in MMI score was not related to student attrition (p = 0.148).

Conclusion

The cause of early attrition is complex and cannot be attributed to a single factor.Undergraduate high school score and written admission examination results were statistically significant factors in relation to student attrition rate and low academic performance. The results of this study showed that the female students scored significantly higher in the multiple MMI tests compared to male students. However, MMI score alone was not significantly related to student attrition.

Novel Insights on Establishing Machine Learning-Based Stroke Prediction Models Among Hypertensive Adults

Background

Stroke is a major global health burden, and risk prediction is essential for the primary prevention of stroke. However, uncertainty remains about the optimal prediction model for analyzing stroke risk. In this study, we aim to determine the most effective stroke prediction method in a Chinese hypertensive population using machine learning and establish a general methodological pipeline for future analysis.

Methods

The training set included 70% of data (n = 14,491) from the China Stroke Primary Prevention Trial (CSPPT). Internal validation was processed with the rest 30% of CSPPT data (n = 6,211), and external validation was conducted using a nested case–control (NCC) dataset (n = 2,568). The primary outcome was the first stroke. Four received analysis methods were processed and compared: logistic regression (LR), stepwise logistic regression (SLR), extreme gradient boosting (XGBoost), and random forest (RF). Population characteristic data with inclusion and exclusion of laboratory variables were separately analyzed. Accuracy, sensitivity, specificity, kappa, and area under receiver operating characteristic curves (AUCs) were used to make model assessments with AUCs the top concern. Data balancing techniques, including random under-sampling (RUS) and synthetic minority over-sampling technique (SMOTE), were applied to process this unbalanced training set.

Results

The best model performance was observed in RUS-applied RF model with laboratory variables. Compared with null models (sensitivity = 0, specificity = 100, and mean AUCs = 0.643), data balancing techniques improved overall performance with RUS, demonstrating a more satisfactory effect in the current study (RUS: sensitivity = 63.9; specificity = 53.7; and mean AUCs = 0.624. Adding laboratory variables improved the performance of analysis methods. All results were reconfirmed in validation sets. The top 10 important variables were determined by the analysis method with the best performance.

Conclusion

Among the tested methods, the most effective stroke prediction model in targeted population is RUS-applied RF. From the insights, the current study revealed, we provided general frameworks for building machine learning-based prediction models.

Stressful Life Events and the Clinical Expression of Obsessive-Compulsive Disorder (OCD): An Exploratory Study

BACKGROUND

In obsessive-compulsive disorder (OCD), symptom content and severity appear to fluctuate over the course of the life cycle in accordance with stressful life events. The objective of this paper was to compare OCD patients with and without reported stressful life events (SLEs) in terms of the sociodemographics of patients and the clinical characteristics of OCD.

METHODS

This was a cross-sectional study involving 1001 patients with OCD. Data concerning SLEs were collected via the Yale OCD Natural History Questionnaire, while for OCD symptoms, the Dimensional Yale-Brown Obsessive-Compulsive Scale was used.

RESULTS

Of the 1001 OCD patients, 605 (60.5%) reported experiencing at least one SLE in their lifetime. Self-declared nonwhite skin color (odds ratio (OR) = 1.51), the presence of a sensory phenomenon (OR = 1.47), and comorbidity with post-traumatic stress disorder (PTSD) (OR = 2.38) were some of the logistic regression variables related to the reported SLEs with relevant statistical significance and risk (i.e., OR) values.

CONCLUSIONS

Our results indicate that SLEs may make Brazilian OCD patients vulnerable to the onset or exacerbation of obsessive-compulsive symptoms. The positive association of the occurrence of SLEs and sensory phenomena in this population could corroborate that environmental influences impact the neurobiology associated with OCD, and likely with other psychiatric disorders as well.

Design and baseline data from the Gratitude Research in Acute Coronary Events (GRACE) study

BACKGROUND

Positive psychological constructs, especially optimism, have been linked with superior cardiovascular health. However, there has been minimal study of positive constructs in patients with acute coronary syndrome (ACS), despite the prevalence and importance of this condition. Furthermore, few studies have examined multiple positive psychological constructs and multiple cardiac-related outcomes within the same cohort to determine specifically which positive construct may affect a particular cardiac outcome.

MATERIALS AND METHODS

The Gratitude Research in Acute Coronary Events (GRACE) study examines the association between optimism/gratitude 2weeks post-ACS and subsequent clinical outcomes. The primary outcome measure is physical activity at 6months, measured via accelerometer, and key secondary outcome measures include levels of prognostic biomarkers and rates of nonelective cardiac rehospitalization at 6months. These relationships will be analyzed using multivariable linear regression, controlling for sociodemographic, medical, and negative psychological factors; associations between baseline positive constructs and subsequent rehospitalizations will be assessed via Cox regression.

RESULTS

Overall, 164 participants enrolled and completed the baseline 2-week assessment; the cohort had a mean age of 61.5+/?10.5years and was 84% men; this was the first ACS for 58% of participants.

CONCLUSION

The GRACE study will determine whether optimism and gratitude are prospectively and independently associated with physical activity and other critical outcomes in the 6months following an ACS. If these constructs are associated with superior outcomes, this may highlight the importance of these constructs as independent prognostic factors post-ACS.

Factors affecting unused remaining volume of intravenous patient-controlled analgesia in patients following laparoscopic gynecologic surgery

PURPOSE

This study was undertaken to evaluate the factors affecting the unused remaining volume of intravenous patient-controlled analgesia (IV PCA) in patients who had undergone laparoscopic gynecologic surgery.

METHODS

We retrospectively collected patient records from pre-existing PCA log sheets from 98 patients. Surgical factors and IV PCA-related data including remaining volume, administration duration, early discontinuation (yes or no), and adverse reactions were recorded. Chi-square test, one-way analysis of variance, and multiple linear regression were applied for data analysis.

RESULTS

The average age of the 98 patients was 40.0 ± 8.24 years. The incidence of postoperative nausea and vomiting (PONV) and early discontinuation were not statistically significant among the different surgical groups (p = .540 and p = .338, respectively). Twenty-eight patients wanted discontinuation of IV PCA and the remaining volume was 33.6 ± 7.8 mL (range 20-55 mL). The significant determinants of remaining volume were whether IV PCA was discontinued due to PONV and duration of surgery (p < .001). The surgical duration was inversely correlated with the remaining volume.

CONCLUSION

Early discontinuation of IV PCA due to PONV is a major contributing factor to wastage of medicine. Prevention and treatment of PONV is needed to encourage patients to maintain PCA use for pain control.

Predicting major adverse cardiac events in spine fusion patients: is the revised cardiac risk index sufficient?

STUDY DESIGN

Observational cohort study.

OBJECTIVE

To determine the accuracy of the Revised Cardiac Risk Index (RCRI) in predicting major adverse cardiac events in patients undergoing spine fusion surgery of 3 levels or more.

SUMMARY OF BACKGROUND DATA

Preoperative cardiac testing is extensively guided by the RCRI, which was developed and validated in thoracic, abdominal, and orthopedic surgical patients. Because multilevel spine fusion surgery is often associated with major transfusion, we hypothesize that the RCRI may not accurately characterize the risk of cardiovascular morbidity in these patients.

METHODS

After institutional review board approval, perioperative data were collected from 547 patients who underwent 3 or more levels of spinal fusion with instrumentation. Postoperative cardiac morbidity was defined as any combination of the following: arrhythmia requiring medical treatment, myocardial infarction (either by electrocardiographic changes or troponin elevation), or the occurrence of demand ischemia. The surgical complexity was categorized as anterior surgery only, posterior cervical and/or thoracic fusion, posterior lumbar fusion, or any surgery that included transpedicular osteotomies. Logistic regression analysis was performed to determine RCRI performance.

RESULTS

The RCRI performed no better than chance (area under the curve = 0.54) in identifying the 49 patients (9%) who experienced cardiac morbidity.

CONCLUSION

The RCRI did not predict cardiac morbidity in our patients undergoing major spine fusion surgery, despite being extensively validated in low-risk noncardiac surgical patients. Preoperative testing and optimization decisions, previously based on the RCRI, may need to be revised to include more frequent functional cardiac imaging and more aggressive implementation of pharmacologic modalities that may mitigate cardiac morbidity, similar to the preoperative evaluation for major vascular surgery.

LEVEL OF EVIDENCE

3.

Compound imaging technology and echogenic needle design: effects on needle visibility and tissue imaging

INTRODUCTION

Needle visualization in ultrasound-guided regional anesthesia can be improved by using needles of echogenic design with higher rate of reflection of ultrasound waves. Imaging solutions such as compound imaging might further improve imaging of both needle and tissue; these effects have not yet been studied. We hypothesized that compound imaging would significantly improve needle visibility, regardless of the insertion angle or needle type used. The effects of compound imaging on needle artifacts and tissue imaging were also investigated.

METHODS

A total of 200 video clips of in-plane needle insertions were obtained in embalmed cadavers with a conventional needle and an echogenic needle at 5 different insertion angles, with both conventional B-mode ultrasound imaging and compound imaging technology. Visibility of the needle shaft and needle tip as well as the needle artifact rate were assessed by a blinded investigator on a 4-point ordinal scale. The effects on tissue image quality and speckle artifacts were also assessed. Stepwise linear regression was performed to differentiate effects on needle visibility scores.

RESULTS

Imaging of the needle shaft and tip was significantly enhanced when compound imaging technology was used (P < 0.0001). Use of echogenically designed needles or shallow needle insertion angles improved visibility of both shaft and tip (both P < 0.0001). With compound imaging, there are fewer needle artifacts, and tissue imaging quality and speckle artifact rate are significantly improved.

CONCLUSIONS

Compound imaging technology enhances needle imaging with both echogenic and conventional needles. Tissue imaging and speckle artifacts are also optimized. Echogenic needle design results in better needle visibility scores in both B-mode and compound imaging.

Predicting fusarium head blight epidemics with weather-driven pre- and post-anthesis logistic regression models

Our objective was to identify weather-based variables in pre- and post-anthesis time windows for predicting major Fusarium head blight (FHB) epidemics (defined as FHB severity ≥ 10%) in the United States. A binary indicator of major epidemics for 527 unique observations (31% of which were major epidemics) was linked to 380 predictor variables summarizing temperature, relative humidity, and rainfall in 5-, 7-, 10-, 14-, or 15-day-long windows either pre- or post-anthesis. Logistic regression models were built with a training data set (70% of the 527 observations) using the leaps-and-bounds algorithm, coupled with bootstrap variable and model selection methods. Misclassification rates were estimated on the training and remaining (test) data. The predictive performance of models with indicator variables for cultivar resistance, wheat type (spring or winter), and corn residue presence was improved by adding up to four weather-based predictors. Because weather variables were intercorrelated, no single model or subset of predictor variables was best based on accuracy, model fit, and complexity. Weather-based predictors in the 15 final empirical models selected were all derivatives of relative humidity or temperature, except for one rainfall-based predictor, suggesting that relative humidity was better at characterizing moisture effects on FHB than other variables. The average test misclassification rate of the final models was 19% lower than that of models currently used in a national FHB prediction system.

Independent contributors to overall quality of life in people with advanced cancer

Background:

The definition of health for people with cancer is not focused solely on the physiology of illness and the length of life remaining, but is also concerned with improving the well-being and the quality of the life (QOL) remaining to be lived. This study aimed to identify the constructs most associated with QOL in people with advanced cancer.

Methods:

Two hundred three persons with recent diagnoses of different advanced cancers were evaluated with 65 variables representing individual and environmental factors, biological factors, symptoms, function, general health perceptions and overall QOL at diagnosis. Three independent stepwise multiple linear regressions identified the most important contributors to overall QOL. R² ranking and effect sizes were estimated and averaged by construct.

Results:

The most important contributor of overall QOL for people recently diagnosed with advanced cancer was social support. It was followed by general health perceptions, energy, social function, psychological function and physical function.

Conclusions:

We used effect sizes to summarise multiple multivariate linear regressions for a more manageable and clinically interpretable picture. The findings emphasise the importance of incorporating the assessment and treatment of relevant symptoms, functions and social support in people recently diagnosed with advanced cancer as part of their clinical care.

Combinatorial network of transcriptional regulation and microRNA regulation in human cancer

BACKGROUND

Both transcriptional control and microRNA (miRNA) control are critical regulatory mechanisms for cells to direct their destinies. At present, the combinatorial regulatory network composed of transcriptional regulations and post-transcriptional regulations is often constructed through a forward engineering strategy that is based solely on searching of transcriptional factor binding sites or miRNA seed regions in the putative target sequences. If the reverse engineering strategy is integrated with the forward engineering strategy, a more accurate and more specific combinatorial regulatory network will be obtained.

RESULTS

In this work, utilizing both sequence-matching information and parallel expression datasets of miRNAs and mRNAs, we integrated forward engineering with reverse engineering strategies and as a result built a hypothetical combinatorial gene regulatory network in human cancer. The credibility of the regulatory relationships in the network was validated by random permutation procedures and supported by authoritative experimental evidence-based databases. The global and local architecture properties of the combinatorial regulatory network were explored, and the most important tumor-regulating miRNAs and TFs were highlighted from a topological point of view.

CONCLUSIONS

By integrating the forward engineering and reverse engineering strategies, we manage to sketch a genome-scale combinatorial gene regulatory network in human cancer, which includes transcriptional regulations and miRNA regulations, allowing systematic study of cancer gene regulation. Our work establishes a pipeline that can be extended to reveal conditional combinatorial regulatory landscapes correlating to specific cellular contexts.

Reporting recommendations for tumor marker prognostic studies (REMARK): explanation and elaboration

BACKGROUND

The Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) checklist consists of 20 items to report for published tumor marker prognostic studies. It was developed to address widespread deficiencies in the reporting of such studies. In this paper we expand on the REMARK checklist to enhance its use and effectiveness through better understanding of the intent of each item and why the information is important to report.

METHODS

REMARK recommends including a transparent and full description of research goals and hypotheses, subject selection, specimen and assay considerations, marker measurement methods, statistical design and analysis, and study results. Each checklist item is explained and accompanied by published examples of good reporting, and relevant empirical evidence of the quality of reporting. We give prominence to discussion of the 'REMARK profile', a suggested tabular format for summarizing key study details.

SUMMARY

The paper provides a comprehensive overview to educate on good reporting and provide a valuable reference for the many issues to consider when designing, conducting, and analyzing tumor marker studies and prognostic studies in medicine in general. To encourage dissemination of the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK): Explanation and Elaboration, this article has also been published in PLoS Medicine.

Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK): explanation and elaboration

The REMARK (Reporting Recommendations for Tumor Marker Prognostic Studies) guideline includes a checklist which aims to improve the reporting of these types of studies. Here, we expand on the REMARK checklist to enhance its use and effectiveness through better understanding of the intent of each item and why the information is important to report. Each checklist item of the REMARK guideline is explained in detail and accompanied by published examples of good reporting. The paper provides a comprehensive overview to educate on good reporting and provide a valuable reference of issues to consider when designing, conducting, and analyzing tumor marker studies and prognostic studies in medicine in general.

Model selection in medical research: a simulation study comparing Bayesian model averaging and stepwise regression

Background

Automatic variable selection methods are usually discouraged in medical research although we believe they might be valuable for studies where subject matter knowledge is limited. Bayesian model averaging may be useful for model selection but only limited attempts to compare it to stepwise regression have been published. We therefore performed a simulation study to compare stepwise regression with Bayesian model averaging.

Methods

We simulated data corresponding to five different data generating processes and thirty different values of the effect size (the parameter estimate divided by its standard error). Each data generating process contained twenty explanatory variables in total and had between zero and two true predictors. Three data generating processes were built of uncorrelated predictor variables while two had a mixture of correlated and uncorrelated variables. We fitted linear regression models to the simulated data. We used Bayesian model averaging and stepwise regression respectively as model selection procedures and compared the estimated selection probabilities.

Results

The estimated probability of not selecting a redundant variable was between 0.99 and 1 for Bayesian model averaging while approximately 0.95 for stepwise regression when the redundant variable was not correlated with a true predictor. These probabilities did not depend on the effect size of the true predictor. In the case of correlation between a redundant variable and a true predictor, the probability of not selecting a redundant variable was 0.95 to 1 for Bayesian model averaging while for stepwise regression it was between 0.7 and 0.9, depending on the effect size of the true predictor. The probability of selecting a true predictor increased as the effect size of the true predictor increased and leveled out at between 0.9 and 1 for stepwise regression, while it leveled out at 1 for Bayesian model averaging.

Conclusions

Our simulation study showed that under the given conditions, Bayesian model averaging had a higher probability of not selecting a redundant variable than stepwise regression and had a similar probability of selecting a true predictor. Medical researchers building regression models with limited subject matter knowledge could thus benefit from using Bayesian model averaging.

Cardiovascular problems in noncardiac surgery

PURPOSE OF REVIEW

Perioperative cardiac complications remain a major area of concern as our surgical population increases in volume, age and frequency of comorbidity. A variety of strategies can be used to optimize patients and potentially reduce the incidence of these serious complications.

RECENT FINDINGS

Recent literature suggests a trend towards less invasive testing for detection and quantification of coronary artery disease and greater interest in pharmacologic 'cardioprotection' using beta-blockers, statins and other agents targeting heart rate control and other mechanisms (e.g. reducing inflammatory responses). The recent Perioperative Ischemic Evaluation study has substantially altered this approach at least towards widespread application to lower/intermediate risk cohorts. Considerable attention has been focused on ensuring optimal standardized perioperative management of patients with a recent percutaneous coronary intervention, particularly those with an intracoronary stent. Widespread surveillance of postoperative troponin release and increasing recognition of the prognostic potential of elevated preoperative brain natriuretic peptides point towards changing strategies for long-term risk stratification.

SUMMARY

The complexity of a particular patient's physiologic responses to a wide variety of surgical procedures, which are undergoing constant technological refinement generally associated with lesser degrees of invasivity and stress make calculation of patients' perioperative risk very challenging. At the present time, adequate information is available for the clinician to screen patients with high-risk preoperative predictors, delay elective surgery for patients with recent intracoronary stents and continue chronic beta-blockade in appropriate patients. New large-scale database and subanalyses of major trials (e.g. Perioperative Ischemic Evaluation and Coronary Artery Revascularization Prophylaxis) should provide additional information to minimize perioperative cardiac risk.