-
American Journal of Epidemiology Feb 2024In epidemiology, collider stratification bias, the bias resulting from conditioning on a common effect of two causes, is oftentimes considered a type of selection bias,...
In epidemiology, collider stratification bias, the bias resulting from conditioning on a common effect of two causes, is oftentimes considered a type of selection bias, regardless of the conditioning methods employed. In this commentary, we distinguish between two types of collider stratification bias: collider restriction bias due to restricting to one level of a collider (or a descendant of a collider) and collider adjustment bias through inclusion of a collider (or a descendant of a collider) in a regression model. We argue that categorizing collider adjustment bias as a form of selection bias may lead to semantic confusion, as adjustment for a collider in a regression model does not involve selecting a sample for analysis. Instead, we propose that collider adjustment bias can be better viewed as a type of overadjustment bias. We further provide two distinct causal diagram structures to distinguish collider restriction bias and collider adjustment bias. We hope that such a terminological distinction can facilitate easier and clearer communication.
Topics: Humans; Selection Bias; Bias; Causality
PubMed: 37939152
DOI: 10.1093/aje/kwad213 -
Annals of Human Genetics Sep 2023A variant in the mucin 5B gene (MUC5B) is strongly associated with the risk of idiopathic pulmonary fibrosis. However, the same variant is associated with increased... (Meta-Analysis)
Meta-Analysis
A variant in the mucin 5B gene (MUC5B) is strongly associated with the risk of idiopathic pulmonary fibrosis. However, the same variant is associated with increased survival time. Previous work suggested that this may be explained by index event bias, with the true effect being to decrease survival. Here, we reassessed this claim using more recent methods and datasets. We found that the statistical assumptions of the previous analysis did not hold, and instead, we applied recent methods of corrected weighted least squares, MR-RAPS and Slope-hunter to both the previous data and an updated consortium meta-analysis. However, these analyses did not yield robust evidence for increased or decreased survival. In simulations of a true effect of decreased survival, we did not observe any realistic scenario in which index event bias led to an observed effect of increased survival. We therefore regard as unsafe the claim that MUC5B has a true effect of decreased survival. Alternative explanations should be sought to explain the observed association with increased survival.
Topics: Humans; Mucin-5B; Genetic Predisposition to Disease; Idiopathic Pulmonary Fibrosis
PubMed: 37537942
DOI: 10.1111/ahg.12522 -
Haematologica Dec 2023Monoclonal gammopathy of undetermined significance (MGUS) is an asymptomatic precursor condition that precedes multiple myeloma and related disorders but has also been...
Disease associations with monoclonal gammopathy of undetermined significance can only be evaluated using screened cohorts: results from the population-based iStopMM study.
Monoclonal gammopathy of undetermined significance (MGUS) is an asymptomatic precursor condition that precedes multiple myeloma and related disorders but has also been associated with other medical conditions. Since systematic screening is not recommended, MGUS is typically diagnosed due to underlying diseases and most cases are not diagnosed. Most previous studies on MGUS disease associations have been based on clinical cohorts, possibly resulting in selection bias. Here we estimate this selection bias by comparing clinically diagnosed and screened individuals with MGUS with regards to demographics, laboratory features, and comorbidities. A total of 75,422 participants in the Iceland Screens, Treats, or Prevents Multiple Myeloma (iStopMM) study were screened for MGUS by serum protein electrophoresis, immunofixation and free light chain assay (clinicaltrials gov. Identifier: NCT03327597). We identified 3,352 individuals with MGUS, whereof 240 had previously been clinically diagnosed (clinical MGUS), and crosslinked our data with large, nationwide registries for information on comorbidities. Those with clinical MGUS were more likely to have at least one comorbidity (odds ratio=2.24; 95% confidence interval: 1.30-4.19), and on average had more comorbidities than the screened MGUS group (3.23 vs. 2.36, mean difference 0.68; 95% confidence interval: 0.46-0.90). They were also more likely to have rheumatological disease, neurological disease, chronic kidney disease, liver disease, heart failure, or endocrine disorders. These findings indicate that individuals with clinical MGUS have more comorbidities than the general MGUS population and that previous studies have been affected by significant selection bias. Our findings highlight the importance of screening data when studying biological and epidemiological implications of MGUS.
Topics: Humans; Multiple Myeloma; Monoclonal Gammopathy of Undetermined Significance; Iceland; Paraproteinemias; Comorbidity; Disease Progression
PubMed: 37439374
DOI: 10.3324/haematol.2023.283191 -
Cancers Aug 2023We discuss the importance of the in vivo models in elucidating cancer biology, focusing on the patient-derived xenograft (PDX) models, which are classic and standard... (Review)
Review
We discuss the importance of the in vivo models in elucidating cancer biology, focusing on the patient-derived xenograft (PDX) models, which are classic and standard functional in vivo platforms for preclinical evaluation. We provide an overview of the most representative models, including cell-derived xenografts (CDX), tumor and metastatic cell-derived xenografts, and PDX models utilizing humanized mice (HM). The orthotopic models, which could reproduce the cancer environment and its progression, similar to human tumors, are particularly common. The standard procedures and rationales of gastric adenocarcinoma (GAC) orthotopic models are addressed. Despite the significant advantages of the PDX models, such as recapitulating key features of human tumors and enabling drug testing in the in vivo context, some challenges must be acknowledged, including loss of heterogeneity, selection bias, clonal evolution, stroma replacement, tumor micro-environment (TME) changes, host cell carryover and contaminations, human-to-host cell oncogenic transformation, human and host viral infections, as well as limitations for immunologic research. To compensate for these limitations, other mouse models, such as syngeneic and humanized mouse models, are currently utilized. Overall, the PDX models represent a powerful tool in cancer research, providing critical insights into tumor biology and potential therapeutic targets, but their limitations and challenges must be carefully considered for their effective use. Lastly, we present an intronic quantitative PCR (qPCR) method to authenticate, detect, and quantify human/murine cells in cell lines and PDX samples.
PubMed: 37686627
DOI: 10.3390/cancers15174352 -
Annals of Epidemiology Oct 2023Placental histopathology is a resource for investigating obesity-associated pregnancy conditions. However, studies oversample adverse pregnancies, biasing findings. We...
PURPOSE
Placental histopathology is a resource for investigating obesity-associated pregnancy conditions. However, studies oversample adverse pregnancies, biasing findings. We examine the association between prepregnancy obesity (risk factor for inflammation) and histologic placental inflammation (correlated with impaired infant neurodevelopment) and how selection bias may influence the association.
METHODS
Singleton term deliveries between 2008 and 2012 from the Magee Obstetric Maternal and Infant database were analyzed. Prepregnancy body mass index (BMI) was categorized as underweight, lean (referent), overweight, and obese. Outcomes were diagnoses of acute (acute chorioamnionitis and fetal inflammation) and chronic placental inflammation (chronic villitis). Risk ratios for associations between BMI and placental inflammation were estimated using selection bias approaches: complete case, exclusion of pregnancy complications, multiple imputation, and inverse probability weighting. E-values approximated how susceptible estimates were to residual selection bias.
RESULTS
Across methods, obesity was associated with an 8-15% lower risk of acute chorioamnionitis, a 7%-14% lower risk of acute fetal inflammation, and a 12%-30% higher risk of chronic villitis relative to lean women. E-values indicated modest residual selection bias could explain away associations, though few measured indications of placental evaluations met this threshold.
CONCLUSIONS
Obesity may contribute to placental inflammation, and we highlight robust methods to analyze clinical data susceptible to selection bias.
Topics: Female; Pregnancy; Humans; Placenta; Chorioamnionitis; Selection Bias; Obesity; Inflammation; Body Mass Index
PubMed: 37302673
DOI: 10.1016/j.annepidem.2023.06.003 -
Frontiers in Sports and Active Living 2023Device-based measurement in physical activity surveillance is increasing, but research design choices could increase the risk of self-selection bias and reactive...
BACKGROUND
Device-based measurement in physical activity surveillance is increasing, but research design choices could increase the risk of self-selection bias and reactive behaviour. The aim of this study is to compare the self-reported physical activity profiles of four different samples: participants in a large national survey, participants in a telephone-based survey of non-responders, participants in the large national survey who accepted the invitation to device-based measuring, and the same sample during the week of monitoring.
METHODS
In October 2020, 163,133 Danish adults participated in a national survey and of those 39,480 signed up for device-based measurements. A balanced random sample ( = 3,750) was invited to wear an accelerometer of whom 1,525 accepted the invitation. Additionally, a short telephone-based survey on 829 non-responders to the national survey was conducted. Sociodemographic characteristics and self-reported weekly frequencies of physical activity across four domains are compared.
RESULTS
The participants in the national survey were older, more often female, and more often not working. Participants in the telephone-based survey were younger, more often doing unskilled work, and were more often active at home and at work. The participants in the device-based sample were more often active during transport and leisure in the national survey, and participants categorized in the most active category increased during the week of monitoring from 29.0% to 60.7% and from 58.5% to 81.7% for active transport and leisure activities, respectively.
CONCLUSION
Recruiting a population representative sample for device-based measurement of physical activity is challenging, and there is a substantial risk of sample selection bias and measurement reactivity. Further research in this area is needed if device-based measures should be considered for population physical activity surveillance.
PubMed: 37614413
DOI: 10.3389/fspor.2023.1236870 -
Journal of Clinical Epidemiology Dec 2023Recalled childhood adiposity is inversely associated with breast cancer observationally, including in Mendelian randomization (MR) studies. Breast cancer studies...
OBJECTIVES
Recalled childhood adiposity is inversely associated with breast cancer observationally, including in Mendelian randomization (MR) studies. Breast cancer studies recruited in adulthood only include survivors of childhood adiposity and breast cancer or a competing risk. We assessed recalled childhood adiposity on participant reported sibling and maternal breast cancer to ensure ascertainment of nonsurvivors.
STUDY DESIGN AND SETTING
We obtained independent strong genetic predictors of recalled childhood adiposity for women and their associations with participant reported own, sibling and maternal breast cancer from UK Biobank genome wide association studies.
RESULTS
Recalled childhood adiposity in women was inversely associated with own breast cancer using Mendelian randomization inverse variance weighting (odds ratio (OR) 0.66, 95% confidence interval (CI) 0.52-0.84) but less clearly related to participant reported sibling (OR 0.89, 95% CI 0.69-1.14) or maternal breast cancer (OR 0.84, 95% CI 0.67-1.05).
CONCLUSION
Weaker inverse associations of recalled childhood adiposity with breast cancer with more comprehensive ascertainment of cases before recruitment suggests the inverse association of recalled childhood adiposity with breast cancer could be partly selection bias from preferential selection of survivors. Greater consideration of survival bias in public health relevant causal inferences would be helpful.
Topics: Humans; Female; Breast Neoplasms; Adiposity; Selection Bias; Genome-Wide Association Study; Mendelian Randomization Analysis; Polymorphism, Single Nucleotide; Body Mass Index
PubMed: 37783402
DOI: 10.1016/j.jclinepi.2023.09.015 -
Obstetrics and Gynecology Nov 2023Postpartum hemorrhage , defined as a cumulative blood loss of 1,000 mL or more or blood loss associated with signs or symptoms of hypovolemia regardless of the route of...
Postpartum hemorrhage , defined as a cumulative blood loss of 1,000 mL or more or blood loss associated with signs or symptoms of hypovolemia regardless of the route of delivery, is the leading cause of preventable maternal death worldwide. The United States has one of the highest maternal mortality rates among developed countries, with about 14% of all maternal deaths associated with postpartum hemorrhage. Although postpartum hemorrhage has multiple causes, the most common is uterine atony-when the uterus fails to adequately contract after childbirth-accounting for 80% of all postpartum hemorrhages. When postpartum hemorrhage occurs despite preventive measures, therapeutic measures are used. Intrauterine hemorrhage-control devices are often the second-line therapy when medical management is unsuccessful. Despite its widespread use in current obstetric practice, the mechanism of intrauterine balloon tamponade, such as the Bakri balloon, is counterintuitive to the physiologic uterine contraction that occurs after delivery to control bleeding, and data on its effectiveness are mixed. Vacuum-induced hemorrhage control, such as with the Jada System, cleared by the U.S. Food and Drug Administration in 2020, is a novel modality for control of postpartum bleeding. It mimics postpartum physiology by applying low-level intrauterine negative pressure to facilitate uterine compressive forces, thereby constricting blood vessels to achieve hemostasis. Preliminary data from four studies are promising but are limited by a lack of control groups, selection bias, or modest sample sizes. The results of ongoing and planned randomized controlled trials will clarify the role of the Jada System for reducing morbidity from postpartum hemorrhage.
Topics: Pregnancy; Female; Humans; Postpartum Hemorrhage; Uterine Balloon Tamponade; Uterus; Postpartum Period; Parturition; Treatment Outcome
PubMed: 37797338
DOI: 10.1097/AOG.0000000000005403 -
Maternal & Child Nutrition Oct 2023Breastfeeding has been consistently associated with higher intelligence since childhood. However, this relation could be confounded due to maternal selection bias. We...
Breastfeeding has been consistently associated with higher intelligence since childhood. However, this relation could be confounded due to maternal selection bias. We estimated the association between predominant breastfeeding and intelligence in school-age children considering potential selection bias and we simulated the intelligence gap reduction between low versus higher socioeconomic status children by increasing breastfeeding. We analysed predominant breastfeeding practices (breastmilk and water-based liquids) of children 0-3 years included in the Mexican Family Life Survey (MxFLS-1). Intelligence was estimated as the z-score of the abbreviated Raven score, measured at 6-12 years in the MxFLS-2 or MxFLS-3. We predicted breastfeeding duration among children with censored data with a Poisson model. We used the Heckman selection model to assess the association between breastfeeding and intelligence, correcting for selection bias and stratified by socioeconomic status. Results show after controlling for selection bias, a 1-month increase in predominant breastfeeding duration was associated with a 0.02 SD increase in the Raven z-score (p < 0.05). The children who were predominantly breastfed for 4-6 months versus <1 month had 0.16 SD higher Raven z-score (p < 0.05). No associations were found using multiple linear regression models. Among low socioeconomic status children, increasing predominantly breastfeeding duration to 6 months would increase their mean Raven z-score from -0.14 to -0.07 SD and reduce by 12.5% the intelligence gap with high socioeconomic status children. In conclusion, predominant breastfeeding duration was significantly associated with childhood intelligence after controlling for maternal selection bias. Increased breastfeeding duration may reduce poverty-driven intelligence inequities.
Topics: Female; Child; Humans; Infant; Breast Feeding; Mexico; Child Development; Intelligence; Milk, Human
PubMed: 37218453
DOI: 10.1111/mcn.13534 -
IEEE Transactions on Medical Imaging Aug 2023Semi-supervised learning (SSL) has demonstrated remarkable advances on medical image classification, by harvesting beneficial knowledge from abundant unlabeled samples....
Semi-supervised learning (SSL) has demonstrated remarkable advances on medical image classification, by harvesting beneficial knowledge from abundant unlabeled samples. The pseudo labeling dominates current SSL approaches, however, it suffers from intrinsic biases within the process. In this paper, we retrospect the pseudo labeling and identify three hierarchical biases: perception bias, selection bias and confirmation bias, at feature extraction, pseudo label selection and momentum optimization stages, respectively. In this regard, we propose a HierArchical BIas miTigation (HABIT) framework to amend these biases, which consists of three customized modules including Mutual Reconciliation Network (MRNet), Recalibrated Feature Compensation (RFC) and Consistency-aware Momentum Heredity (CMH). Firstly, in the feature extraction, MRNet is devised to jointly utilize convolution and permutator-based paths with a mutual information transfer module to exchanges features and reconcile spatial perception bias for better representations. To address pseudo label selection bias, RFC adaptively recalibrates the strong and weak augmented distributions to be a rational discrepancy and augments features for minority categories to achieve the balanced training. Finally, in the momentum optimization stage, in order to reduce the confirmation bias, CMH models the consistency among different sample augmentations into network updating process to improve the dependability of the model. Extensive experiments on three semi-supervised medical image classification datasets demonstrate that HABIT mitigates three biases and achieves state-of-the-art performance. Our codes are available at https://github.com/CityU-AIM-Group/HABIT.
Topics: Bias; Motion; Supervised Machine Learning
PubMed: 37027665
DOI: 10.1109/TMI.2023.3247440