Survival Bias Research Articles

Implementation of the trial emulation approach in medical research: a scoping review

BackgroundWhen conducting randomised controlled trials is impractical, an alternative is to carry out an observational study. However, making valid causal inferences from observational data is challenging because of the risk of several statistical biases. In 2016 Hernán and Robins put forward the ‘target trial framework’ as a guide to best design and analyse observational studies whilst preventing the most common biases. This framework consists of (1) clearly defining a causal question about an intervention, (2) specifying the protocol of the hypothetical trial, and (3) explaining how the observational data will be used to emulate it.MethodsThe aim of this scoping review was to identify and review all explicit attempts of trial emulation studies across all medical fields. Embase, Medline and Web of Science were searched for trial emulation studies published in English from database inception to February 25, 2021. The following information was extracted from studies that were deemed eligible for review: the subject area, the type of observational data that they leveraged, and the statistical methods they used to address the following biases: (A) confounding bias, (B) immortal time bias, and (C) selection bias.ResultsThe search resulted in 617 studies, 38 of which we deemed eligible for review. Of those 38 studies, most focused on cardiology, infectious diseases or oncology and the majority used electronic health records/electronic medical records data and cohort studies data. Different statistical methods were used to address confounding at baseline and selection bias, predominantly conditioning on the confounders (N = 18/49, 37%) and inverse probability of censoring weighting (N = 7/20, 35%) respectively. Different approaches were used to address immortal time bias, assigning individuals to treatment strategies at start of follow-up based on their data available at that specific time (N = 21, 55%), using the sequential trial emulations approach (N = 11, 29%) or the cloning approach (N = 6, 16%).ConclusionDifferent methods can be leveraged to address (A) confounding bias, (B) immortal time bias, and (C) selection bias. When working with observational data, and if possible, the ‘target trial’ framework should be used as it provides a structured conceptual approach to observational research.

Statins and Mortality in COPD: A Methodological Review of Observational Studies

Randomized controlled trials and observational studies have reported conflicting results on the potential beneficial effects of statins on mortality in patients with chronic obstructive pulmonary disease (COPD). We performed a systematic search of the literature to review all observational studies reporting relative risks of death with statin use in COPD, focusing on potential sources of bias. We identified 15 observational studies, out of 2835, of which 12 were affected by time-related and other biases and the remaining 3 by confounding bias. All 15 studies were also subject to confounding bias due to lack of adjustment for important COPD-related factors. The risk of death associated with statin use was reduced across all 15 studies (pooled relative risk (PRR) 0.66; 95% CI: 0.59-0.74). The reduction was observed in 7 studies with immortal time bias (PRR 0.62; 95%: 0.53-0.72), two with collider-stratification bias (PRR 0.60; 95% CI: 0.45-0.80), one with time-window bias (RR 0.61; 95% CI: 0.38-0.98), one with immeasurable time bias (RR 0.50; 95% CI: 0.40-0.62), and one with exposure misclassification (RR 0.86; 95% CI: 0.72-1.03). The three studies that avoided these biases were, however, affected by confounding bias resulting in a PRR of 0.77 (95% CI: 0.61-0.98). In conclusion, the observational studies investigating statin use and mortality in COPD are affected by major biases, many of which can result in spurious protective effects. Well-designed observational studies that carefully emulate randomized trials are needed to resolve this uncertainty regarding the potential beneficial benefits of statins on mortality in patients with COPD.

The evolution of the Milky Way’s thin disc radial metallicity gradient with K2 asteroseismic ages

ABSTRACT The radial metallicity distribution of the Milky Way’s disc is an important observational constraint for models of the formation and evolution of our Galaxy. It informs our understanding of the chemical enrichment of the Galactic disc and the dynamical processes therein, particularly radial migration. We investigate how the metallicity changes with guiding radius in the thin disc using a sample of red giant stars with robust astrometric, spectroscopic, and asteroseismic parameters. Our sample contains 668 stars with guiding radii 4 < Rg < 11 kpc and asteroseismic ages covering the whole history of the thin disc with precision ${\approx} 25 {{\, \rm per\ cent}}$. We use MCMC analysis to measure the gradient and its intrinsic spread in bins of age and construct a hierarchical Bayesian model to investigate the evolution of these parameters independently of the bins. We find a smooth evolution of the gradient from ≈−0.07 dex kpc−1 in the youngest stars to ≈−0.04 dex kpc−1 in stars older than 10 Gyr, with no break at intermediate ages. Our results are consistent with those based on asteroseismic ages from CoRoT, with that found in Cepheid variables for stars younger than 1 Gyr, and with open clusters for stars younger than 6 Gyr. For older stars we find a significantly lower metallicity in our sample than in the clusters, suggesting a survival bias favouring more metal-rich clusters. We also find that the chemical evolution model of Chiappini '09 is too metal poor in the early stages of disc formation. Our results provide strong new constraints for the growth and enrichment of the thin disc and radial migration, which will facilitate new tests of model conditions and physics.

Evaluation of cataract formation in fish exposed to environmental radiation at Chernobyl and Fukushima

Recent studies apparently finding deleterious effects of radiation exposure on cataract formation in birds and voles living near Chernobyl represent a major challenge to current radiation protection regulations. This study conducted an integrated assessment of radiation exposure on cataractogenesis using the most advanced technologies available to assess the cataract status of lenses extracted from fish caught at both Chernobyl in Ukraine and Fukushima in Japan. It was hypothesised that these novel data would reveal positive correlations between radiation dose and early indicators of cataract formation.The structure, function and optical properties of lenses were analysed from atomic to millimetre length scales. We measured the short-range order of the lens crystallin proteins using Small Angle X-Ray Scattering (SAXS) at both the SPring-8 and DIAMOND synchrotrons, the profile of the graded refractive index generated by these proteins, the epithelial cell density and organisation and finally the focal length of each lens.The results showed no evidence of a difference between the focal length, the epithelial cell densities, the refractive indices, the interference functions and the short-range order of crystallin proteins (X-ray diffraction patterns) in lens from fish exposed to different radiation doses. It could be argued that animals in the natural environment which developed cataract would be more likely, for example, to suffer predation leading to survivor bias. But the cross-length scale study presented here, by evaluating small scale molecular and cellular changes in the lens (pre-cataract formation) significantly mitigates against this issue.

CRISPR/Cas9 Essential Gene Editing in Drosophila.

Since the addition of the CRISPR/Cas9 technology to the genetic engineering toolbox, the problems of low efficiency and off-target effects hamper its widespread use in all fields of life sciences. Furthermore, essential gene knockout usually results in failure and it is often not obvious whether the gene of interest is an essential one. Here, we report on a new strategy to improve the CRISPR/Cas9 genome editing, which is based on the idea that editing efficiency is tightly linked to how essential the gene to be modified is. The more essential the gene, the less the efficiency of the editing and the larger the number of off-targets, due to the survivorship bias. Considering this, we generated deletions of three essential genes in Drosophila: trf2, top2, and mep-1, using fly strains with previous target gene overexpression ("pre-rescued" genetic background).

Predation bias of Ordovician predators on trilobites

Although prey selection has been commonly documented in extant animals, evidence of prey selection in deep time is rare by comparison. Here, we collected 147 broken sclerites of trilobites, almost all of which were caused by lethal predation, from the Upper Ordovician in NW China. The damage was mainly distributed in three larger trilobites ( Platyptychopyge , Birmanites and Ordosaspis ), and larger-bodied trilobites were more vulnerable to predation. One plausible explanation is that Late Ordovician predators in this area preferred larger trilobites, because of the greater nutrients and energy provided by larger prey; also, larger prey require less effort to capture because they are less capable of escaping from predation compared with smaller prey, which can more easily hide or escape. Another, lower possibility is survivor bias owing to smaller prey being completely eaten up. Moreover, the large-eyed trilobite Nileus , which has a similar size and morphology to Ordosaspis , had a theoretical preying rate less than 1/20 of that of Ordosaspis , indicating that the eye is an important sensory organ in trilobites. The increased predation pressure may have also contributed to the increased eye diversity and visual system resolution of trilobites since the Ordovician. Supplementary material: Supplementary figures and tables are available at https://doi.org/10.6084/m9.figshare.c.6670609 Thematic collection: This article is part of the Chemical Evolution of the Mid-Paleozoic Earth System and Biotic Response collection available at: https://www.lyellcollection.org/topic/collections/chemical-evolution-of-the-mid-paleozoic-earth-system

Dysthyroidism during immune checkpoint inhibitors is associated with improved overall survival in adult cancers: data mining of 1385 electronic patient records

BackgroundDysthyroidism (DT) is a common toxicity of immune checkpoint inhibitors (ICIs) and prior work suggests that dysthyroidism (DT) might be associated with ICI efficacy.Patients and methodsConSoRe, a new generation data...

Impact of conversion surgery after chemotherapy in patients with initially unresectable and recurrent biliary tract cancer.

Gemcitabine, cisplatin, and S-1 chemotherapy was superior to gemcitabine and cisplatin chemotherapy for progression-free survival and overall survival for unresectable and recurrent biliary tract cancer in a randomized phase III trial (KHBO1401). This study aimed to evaluate the outcome of conversion surgery after chemotherapy in biliary tract cancer patients (ancillary study, KHBO1401-3C). A total of 246 patients were enrolled in KHBO1401. We compared progression-free and overall survivals between the conversion surgery and non-conversion surgery groups. Eight patients (3.3%) underwent conversion surgery with chemotherapy, seven of whom were diagnosed with unresectable disease and one with recurrence. Six and two patients received gemcitabine, cisplatin, and S-1 chemotherapy as well as gemcitabine and cisplatin chemotherapy, respectively. Three patients in the conversion surgery group who received gemcitabine, cisplatin, and S-1 chemotherapy showed no disease progression and survived without postoperative chemotherapy. Preoperative carbohydrate antigen 19-9 (CA19-9) level was a prognostic factor for conversion surgery. After correcting for immortal time bias, 1-year progression-free survival rates in the conversion surgery and non-conversion surgery groups were 50.0% and 19.0%, respectively (hazard ratio 0.343, 95% confidence interval 0.286-0.843, p = 0.0092). One-year overall survival rates in the conversion surgery and non-conversion surgery groups were 87.5% and 56.0%, respectively (hazard ratio 0.222, 95% confidence interval 0.226-0.877, p = 0.0197). Conversion surgery might be an option for the treatment of unresectable and recurrent biliary tract cancer in patients with normal preoperative CA19-9 level.

Linear fixed-effects estimation with nonrepeated outcomes

We demonstrate that popular linear fixed-effects panel-data estimators are biased and inconsistent when applied in a discrete-time hazard setting, even if the data-generating process is consistent with the linear model. The bias is not just survival bias, but originates from the impossibility to transform the model such that the remaining disturbance term becomes conditional mean independent of the explanatory variables. The bias is hence present even in the absence of unobserved heterogeneity. We discuss instrumental variables estimation, using first-differences of the explanatory variables as instruments, as alternative estimation strategy. Monte Carlo simulations and an empirical application substantiate our theoretical results.

Preexisting social ties among Auschwitz prisoners support Holocaust survival

Survivor testimonies link survival in deadly POW camps, Gulags, and Nazi concentration camps to the formation of close friendships with other prisoners. To provide evidence free of survival bias on the importance of social ties for surviving the Holocaust, we study individual histories of 30 thousand Jewish prisoners who entered the Auschwitz-Birkenau concentration camp on transports from the Theresienstadt ghetto. We ask whether the availability of potential friends among fellow prisoners on a transport influenced the chances of surviving the Holocaust. Relying on multiple proxies of preexisting social networks and varying social-linkage composition of transports, we uncover a significant survival advantage to entering Auschwitz with a larger group of potential friends.

Study of the impact of cognitive biases on the process of managerial decision-making in the educational sphere

Study of the impact of cognitive biases on the process of managerial decision-making in the educational sphere

Is Fragmented Cancer Care Associated With Medical Expenditure? Nationwide Evidence From Patients With Lung Cancer Using National Insurance Claim Data.

Objectives: We aimed to investigate the association between fragmented cancer care in the early phase after cancer diagnosis and patient outcomes using national insurance claim data. Methods: We identified National Health Insurance beneficiaries diagnosed with lung cancer in South Korea from 2010 to 2014. We included 1,364 lung cancer patients with reduced immortal time bias and heterogeneity. We performed multiple regression analysis using a generalized estimate equation with a gamma distribution for medical expenditures. Results: Among the 1,364 patients with lung cancer, 12.8% had fragmented cancer care. Healthcare costs were higher in fragmented cancer care for both during diagnosis to 365days and diagnosis to 1,825days. Linear regression results showed that fragmented cancer care was associated with 1.162 times higher costs during the period from diagnosis to 365days and 1.163 times the cost for the period from diagnosis to 1,825days. Conclusion: We found fragmented cancer care is associated with higher medical expenditure. Future health policy should consider the limitation of patients' free will when opting for fragmented cancer care, as there are currently no control mechanisms.

Immortal Time Bias With Time-Varying Exposures in Environmental Epidemiology: A Case Study in Lung Cancer Survival.

Immortal time bias is a well-recognized bias in clinical epidemiology but is rarely discussed in environmental epidemiology. Under the target trial framework, this bias is formally conceptualized as a misalignment between the start of study follow-up (time 0) and treatment assignment. This misalignment can occur when attained duration of follow-up is encoded into treatment assignment using minimums, maximums, or averages. The bias can be exacerbated in the presence of time trends commonly found in environmental exposures. Using lung cancer cases from the California Cancer Registry (2000-2010) linked with estimated concentrations of particulate matter less than or equal to 2.5 μm in aerodynamic diameter (PM2.5), we replicated previous studies that averaged PM2.5 exposure over follow-up in a time-to-event model. We compared this approach with one that ensures alignment between time 0 and treatment assignment, a discrete-time approach. In the former approach, the estimated overall hazard ratio for a 5-μg/m3 increase in PM2.5 was 1.38 (95% confidence interval: 1.36, 1.40). Under the discrete-time approach, the estimated pooled odds ratio was 0.99 (95% confidence interval: 0.98, 1.00). We conclude that the strong estimated effect in the former approach was likely driven by immortal time bias, due to misalignment at time 0. Our findings highlight the importance of appropriately conceptualizing a time-varying environmental exposure under the target trial framework to avoid introducing preventable systematic errors.

Adenocarcinoma of the Bladder: Assessment of Survival Advantage Associated With Radical Cystectomy and Comparison With Urothelial Bladder Cancer

To evaluate the association between radical cystectomy (RC) and cancer-specific mortality (CSM) in patients diagnosed with adenocarcinoma of the bladder (ACB). Moreover, to directly compare the survival advantage of RC between ACB vs. urothelial bladder cancer (UBC). Non-metastatic muscle-invasive ACB and UBC patients were identified within Surveillance, Epidemiology, and End Results database (SEER 2000-2018). All analyses were stratified between RC vs. no-RC, in either organ-confined (OC: T2N0M0) or non-organ-confined (NOC: T3-4N0M0 or TanyN1-3M0) stages. Propensity score matching (PSM), cumulative incidence plots, competing risks regression (CRR) analyses, and 3 months' landmark analyses were performed. Overall, 1,005 ACB and 47,741 UBC patients were identified, of whom 475 (47%) and 19,499 (41%) were treated with RC, respectively. After PSM, comparison between RC vs. no-RC applied to 127 vs. 127 OC-ACB, 7,611 vs. 7,611 OC-UBC, 143 vs. 143 NOC-ACB, and 4,664 vs. 4,664 NOC-UBC patients. 36-month CSM rates in RC vs. no-RC patients were 14 vs. 44% in OC-ACB, 18 vs. 39% in OC-UBC, 49 vs. 66% in NOC-ACB, and 44 vs. 56% in NOC-UBC patients. In CRR analyses, the effect of RC on CSM yielded a hazard ratio of 0.37 in OC-ACB, of 0.45 in OC-UBC, of 0.65 in NOC-ACB and of 0.68 in NOC-UBC patients (all P values<0.001). Landmark analyses virtually perfectly replicated the results. In ACB, regardless of stage, RC is associated with lower CSM. The magnitude of this survival advantage was greater in ACB than in UBC, even after control for immortal time bias.

Unread, yet preserved: A case study on survival of the 19th-century printed poetry

Distant reading promises access to "the great unread", which should allow scholars to rethink the history of literature. However, the rise in volume of data does not guarantee the understanding of a corpus and its relation to the literary population. This article discusses how a "complete" corpus of the 19th-century poetry books in Russian might be collected with account for historical data and potential survivorship bias. Even if bibliographical sources cannot provide a complete list of books printed in a given period, the amount of "incompleteness" can be directly estimated with the unseen species models. The estimation of survival ratios for printed poetry shows differences in the loss rate across different types of sources: with conventional editions, like books and anthologies, are well-preserved, while booklets and pamphlets are the largest expected source of loss. These findings allow us to estimate what an "exhaustive" corpus can look like and define the features of "the unread" and "unseen" inside it.

The Impact of COVID-19 on Parkinson’s Disease: A Case-Controlled Registry and Questionnaire Study on Clinical Markers and Patients’ Perceptions

Introduction. Parkinson’s disease (PD) is a neurodegenerative disease with motor and nonmotor symptoms. Coronavirus disease 2019 (COVID-19) is an infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Objectives. To explore how COVID-19 affects motor, nonmotor, and general health aspects of PD and to map how PD patients perceive their change in symptoms since falling ill with COVID-19. Method. The study was descriptive, case-controlled, and based on both registry and questionnaire data. At baseline, the controls were matched on age, sex, and disease severity. Information on the severity of the disease, nonmotor symptoms, motor symptoms, and general health was retrieved from the Swedish Registry for PD. Registry data from a COVID-19 group ( n = 45 ) and a control group ( n = 73 ), as well as questionnaires from a COVID-19 group ( n = 24 ) and a control group ( n = 42 ), were compared. Results. We did not find that SARS-CoV-2 infection affects any major aspect of nonmotor symptoms, motor symptoms, general health, and perception of change in PD patients’ post-COVID-19. Compared to controls, the COVID-19 group reported a more positive subjective experience of pain and quality of life and a perception of change post-COVID-19 regarding general motor function, sleep quality, and mood (all p &lt; 0.05 ). Conclusion. Although SARS-CoV-2 infection does not seem to affect PD symptoms in any major respect, the subjective experience of several aspects of life in PD patients might be slightly improved post-COVID-19 compared to a control group. The findings warrant further investigations due to the small sample size and possible survivorship bias.

Effect of statins on the recurrence of hepatocellular carcinoma after liver transplantation: An illusion revealed by exposure density sampling.

Statins have been reported to reduce overall death and hepatocellular carcinoma (HCC) recurrence in liver transplantation (LT) recipients. However, previous retrospective studies have significant flaws in immortal time bias. Using data from 658 patients who received LT for HCC, we matched 140 statin users with statin nonusers in a 1:2 ratio at the time of the first statin administration after LT using the exposure density sampling (EDS). The propensity score, calculated using baseline variables (including explant pathology), was used for EDS to equilibrate both groups. HCC recurrence and overall death were compared after adjusting for information at the time of sampling. Among statin users, the median time to statin start was 219 (IQR 98-570) days, and intensity of statins was mainly moderate (87.1%). Statin users and nonusers sampled using EDS showed well-balanced baseline characteristics, including detailed tumour pathology, and similar HCC recurrence with cumulative incidences of 11.3% and 11.8% at 5 years, respectively (p = .861). In multivariate Cox models (HR 1.04, p = .918) and subgroup analyses, statins did not affect HCC recurrence. Conversely, statin users showed a significantly lower risk of overall death than nonusers (HR 0.28, p < .001). There was no difference in the type and intensity of statin usage between statin users who experienced HCC recurrence and those who did not. Upon controlling immortal time bias by EDS, statins did not affect HCC recurrence but reduced mortality after LT. Statin usage is encouraged for survival benefits but not for preventing HCC recurrence in LT recipients.

Is research on hedge fund performance published selectively? A quantitative survey

AbstractWe examine whether estimates of hedge fund performance reported in prior empirical research are affected by publication bias. Using a sample of 1019 intercept terms from regressions of hedge fund returns on risk factors (the “alphas”) collected from 74 studies published between 2001 and 2021, we show that the selective publication of empirical results does not significantly contaminate inferences about hedge fund returns. Most of our monthly alpha estimates adjusted for the (small) bias fall within a relatively narrow range of 30–40 basis points, indicating positive abnormal returns of hedge funds: Hedge funds generate money for investors. Studies that explicitly control for potential biases in the underlying data (e.g., backfilling and survivorship biases) report lower but still positive alphas. Our results demonstrate that despite the prevalence of publication selection bias in many other research settings, publication may not be selective when there is no strong a priori theoretical prediction about the sign of the estimated coefficients.

Lost in data translation: A critical review of datasets on refugees

AbstractQuantitative data are the backbone of the international refugee regime. Academic researchers and international organizations have been drawing attention to the problems of refugee data that obscure refugee vulnerabilities. Despite these efforts, the limitations of quantitative refugee data, particularly with regard to gendered aspects of refugee vulnerabilities, have often been overlooked. This article aims to address this issue by examining the major limitations of quantitative refugee data. It focuses on the question of “What are the major limitations of quantitative refugee data regarding gendered aspects of refugee vulnerabilities?” By conducting a systematic review of 67 datasets from international organizations, nation‐states, and independent data collectors, it argues that there are five fundamental problems with the data: (1) lack of disaggregation, (2) lack of cross‐tabulation and inability to harmonize datasets, (3) inconsistent periodization frequency, (4) lack of sex and gender diversity measurement, and (5) survivor bias. This article highlights the importance of critically analysing the data that informs theories and decisions regarding refugee vulnerabilities.

#5533 DOES AV FISTULA FORMATION DELAY TIMING OF INITIATION OF KIDNEY REPLACEMENT THERAPY IN PATIENTS WITH STAGE 5 CHRONIC KIDNEY DISEASE?

Abstract Background and Aims Several retrospective observational studies have supported kidney protective effects of arteriovenous fistula (AVF) formation. However, these studies were limited by immortal time and selection biases. We investigated whether AVF formation delays the initiation of kidney replacement therapy (KRT) in patients with stage 5 CKD by applying target trial emulation methods, which do not suffer from these biases. Method We included adult patients who had an eGFR ≤15 mL/min/1.73m2, attended the ‘low clearance’ nephrology clinic in the West of Scotland between January 1st 2010 and May 1st 2022, and had no prior AVF or AV graft formation. Available data were obtained from the Strathclyde Electronic Renal Patient Record. The target trial would randomize patients to either receive an AVF immediately or to not receive an AVF. To emulate this trial, we matched each patient who underwent AVF formation to patients who had not undergone AVF formation but remained eligible to participate in the trial, and were matched in sex, age (within 5 years) and eGFR (within 0.5 ml/min/1.73m2). Inverse probability of treatment weighting was used to adjust for baseline confounders, including age, sex, comorbidities, medication use, serum and urine biochemical measurements (eGFR CKD-EPI, eGFR CKD-EPI slope for 6 months preceding trial, haemoglobin, C-reactive protein, albumin, phosphate, adjusted calcium, ferritin, urine protein:creatinine ratio), and blood pressure. The primary outcome was kidney replacement therapy. The eGFR slope closest to the time of AVF creation was estimated as a co-primary endpoint. We estimated hazard ratios using Cox regression and estimated restricted mean survival time (RMST) from the Kaplan-Meier curves. The eGFR slope co-primary endpoint was analysed with a mixed-effects model. Results Among 2,988 included patients (55% men; mean [SD] age 64 [15] years), AVF formation was associated with a higher risk of KRT (HR 1.45; CI 1.20–1.49, p &amp;lt;0.001, Figure 1) and a lower risk of death (HR 0.68; (0.64–0.80, p = 0.001). The AVF group had a lower KRT-free survival with an estimated RMST difference of 265 days (95% CI −331 to −199, p &amp;lt;0.001) and higher overall survival (RMST difference of 191 days; 95% CI 57 to 326, p = 0.005). Finally, we used a mixed-effects model to analyse the association between eGFR CKD-EPI, time, and AVF formation. AVF formation and time were both associated with a negative slope in the model (estimates −1.28; 95% CI −1.36 to −1.19, p&amp;lt;0.001 and −0.01; 95% CI −0.01 to −0.01, p&amp;lt;0.001, respectively, Figure 2). Their interaction was also associated with a negative slope (estimate −0.001; 95% CI −0.0022 to −0.0004, p &amp;lt;0.001), suggesting those undergoing AVF formation had a more rapid eGFR decline compared to the control group. Conclusion Unlike what was observed in previously published work, we did not identify a kidney protective effect of AVF formation. The estimated time of dialysis initiation remains the main determinant for timing of access surgery. These findings illustrate the usefulness of target trial emulation in approaching research questions where randomised controlled trials would be impractical.