scholarly journals How Well do Covariates Perform When Adjusting for Sampling Bias in COVID-19 Research? Insights from Multiverse Analyses

2021 ◽  
Author(s):  
Keven Joyal-Desmarais ◽  
Jovana Stojanovic ◽  
Eric Kennedy ◽  
Joanne Enticott ◽  
Vincent Gosselin Boucher ◽  
...  

Background. COVID-19 research has relied heavily on convenience-based observational samples, which—though often necessary—are susceptible to important sampling biases. We begin with a theoretical overview and introduction to the dynamics that underlie sampling bias. We then empirically examine sampling bias in COVID-19 surveys and evaluate the degree to which common statistical adjustments for demographic covariates successfully attenuate such bias. Methods. We analysed responses to identical questions from three convenience and three largely representative samples (total N = 13,731) collected in Canada within the International COVID-19 Awareness and Responses Evaluation Study (www.icarestudy.com). We compared samples on 11 behavioural and psychological outcomes (e.g., adherence to COVID-19 prevention measures, vaccine intentions) across three time points and employed multiverse-style analyses to examine how 512 combinations of demographic covariates (e.g., sex, age, education, income, ethnicity) impact sampling discrepancies on these outcomes. Results. Significant discrepancies emerged between samples on 73% of outcomes, such that participants in the convenience samples held more positive thoughts towards and engaged in more COVID-19 prevention behaviours. Covariates attenuated sampling differences in only 55% of cases and increased differences in 45%. No covariate performed reliably well. Conclusion. Our results suggest that convenience samples may hold more positive dispositions towards COVID-19 prevention behaviours being studied than would samples drawn using more representative means. Adjusting results for demographic covariates frequently increased rather than decreased bias, suggesting that researchers should be cautious when interpreting adjusted findings. Using multiverse-style analyses as extended sensitivity analyses is recommended.

Stroke ◽  
2015 ◽  
Vol 46 (suppl_1) ◽  
Author(s):  
Scott C Brown ◽  
Kefeng Wang ◽  
Chuanhui Dong ◽  
Mary B Farrell ◽  
Gary V Heller ◽  
...  

OBJECTIVE: Accreditation of cerebrovascular ultrasound laboratories by the Intersocietal Accreditation Commission (IAC) or equivalent bodies is supported by The Joint Commission certification of stroke centers. Limited information exists on the accreditation status and geographic distribution of these testing facilities in the US. The aims were to: (1) Identify the proportion of IAC accredited vascular testing facilities used by Medicare beneficiaries for outpatient cerebrovascular testing services; (2) Describe the geographical distribution of these facilities; and (3) Identify variation in the types and volumes of cerebrovascular testing procedures by accreditation status. METHODS: As a part of the VALUE (Vascular Accreditation, Location & Utilization Evaluation) Study, we examined the proportion of IAC accredited facilities that conducted cerebrovascular testing in a 5% CMS random Outpatient Limited Data Set (LDS) for the US in 2011 and investigated their geographical distribution using the Medicare Provider of Services (POS) file. RESULTS: Of the 7,864 total facilities billing Medicare for cerebrovascular testing procedures, only 22% (n=1,723) were IAC accredited. The percentage of facilities conducting cerebrovascular testing that were IAC accredited varied by region (Χ2[3]=400.4, p<0.0001), with 43%, 21%, 17% and 13% located in the Northeast, South, Midwest, and West, respectively. However, when examining the total number of cerebrovascular outpatient procedures conducted in 2011 (total n=38,646), 41% (15,729) were conducted in IAC accredited facilities. Moreover, when examining procedure type across all sites, 98% (38,011) of all cerebrovascular testing procedures conducted were carotid duplex, of which 41% (15,417) were conducted in IAC accredited facilities. In contrast, 1% (n=315) of all cerebrovascular procedures were transcranial (TCD), of which 56% (n=177) were conducted in IAC accredited facilities. CONCLUSIONS: The proportion of IAC accredited facilities conducting outpatient cerebrovascular testing is low and varies by region. The growing number of certified stroke centers should be accompanied by more accredited vascular testing facilities that could potentially improve quality of stroke care.


Author(s):  
T. L. Dickson ◽  
F. A. Simonen

The United States Nuclear Regulatory Commission (USNRC) initiated a comprehensive project in 1999 to determine if improved technologies can provide a technical basis to reduce the conservatism in the current regulations for pressurized thermal shock (PTS) while continuing to provide reasonable assurance of adequate protection to public health and safety. A relaxation of PTS regulations could have profound implications for plant license renewal considerations. During the PTS re-evaluation study, an improved risk-informed computational methodology was developed that provides a more realistic characterization of PTS risk. This updated methodology was recently applied to three commercial PWRs. The results of this study provide encouragement that a technical basis can be established to support a relaxation of current PTS regulations. One significant model improvement applied in the PTS re-evaluation study was the development of flaw databases derived from the non-destructive and destructive examinations of material from cancelled reactor pressure vessels (RPV). Empirically-based statistical distributions derived from these databases and expert illicitation were used to postulate the number, size, and location of flaws in welded and base metal (plate and forging) regions of an RPV during probabilistic fracture mechanics (PFM) analyses of RPVs subjected to transient loading conditions such as PTS. However, limitations in the available flaw data have required assumptions to be made to complete the risk-based flaw models. Sensitivity analyses were performed to evaluate the impact of four flaw-related assumptions. Analyses addressed: 1) truncations of distributions to exclude flaws of extreme depth dimensions, 2) vessel-to-vessel differences in flaw data, 3) large flaws observed in weld repair regions, and 4) the basis for estimating the number of surface breaking flaws. None of the four alternate weld flaw models significantly impacted calculated vessel failure frequencies or invalidated the tentative conclusions derived from the USNRC PTS re-evaluation study.


2019 ◽  
Vol 40 (Supplement_1) ◽  
Author(s):  
W Gan ◽  
D Bennett ◽  
A Mahajan ◽  
H Du ◽  
Z Chen ◽  
...  

Abstract Background Observational studies have reported conflicting results about the associations of iron status with risk of cardio-metabolic diseases but such studies are constrained by confounding and reverse causality. Purpose To assess the causal relevance of iron status biomarkers (transferrin, serum iron, and ferritin) for risk of coronary artery diseases (CAD), ischaemic stroke (IS), and type 2 diabetes (T2D), using Mendelian randomization (MR). Methods Effect size estimates for genetic variants associated with iron status biomarkers were obtained from the Genetics of Iron Status consortium (transferrin saturation, serum iron, and ferritin: n=48,972). The corresponding effects of these variants on the risk of CAD, IS and T2D were obtained from a meta-analysis of unrelated participants of European ancestry in the UK Biobank (UKB), together with previously recruited participants in CARDIOGRAMplusC4D (total n=90,377 CAD cases), MEGASTROKE (total n=43,381 IS cases) and DIAGRAM (total n=74,124 T2D cases), respectively. The main analysis used a two-sample inverse-variance weighted MR, while the sensitivity analyses used weighted-median, weighted-mode, MR-PRESSO, and MR-Egger approaches. Results MR analysis demonstrated significant inverse association of each of the three genetically-instrumented iron status biomarker with risk of CAD (transferrin saturation OR=0.96 [95% CI: 0.92–0.99], p=0.02; serum iron OR=0.93 [0.89–0.97], p=0.001; and ferritin OR=0.86 (0.79–0.94), p=0.001, per 1 SD higher level). In contrast, these iron status biomarkers showed positive associations with risk of T2D (transferrin saturation OR=1.06 [1.01–1.11], p=0.01; serum iron OR=1.06 [0.99–1.13], p=0.07; and ferritin OR=1.12 [0.99–1.26], p=0.06, per 1 SD higher level). There was positive, but non-significant, association of IS with each of the iron status biomarker analysed. Sensitivity analyses using several different MR approaches yielded concordant results. Conclusions Among European adults, iron status appeared to have causal associations, but in opposite directions, with the risk of CHD and T2D. Our findings highlight the need for caution about strategies for advocating iron supplementation in individuals with normal haemoglobin levels for prevention of CAD. Acknowledgement/Funding British Heart Found, Medical Research Council, Wellcome Trust, NIHR Biomedical Research Centre, Oxford


2020 ◽  
Vol 117 (12) ◽  
pp. 6476-6483 ◽  
Author(s):  
Elli J. Theobald ◽  
Mariah J. Hill ◽  
Elisa Tran ◽  
Sweta Agrawal ◽  
E. Nicole Arroyo ◽  
...  

We tested the hypothesis that underrepresented students in active-learning classrooms experience narrower achievement gaps than underrepresented students in traditional lecturing classrooms, averaged across all science, technology, engineering, and mathematics (STEM) fields and courses. We conducted a comprehensive search for both published and unpublished studies that compared the performance of underrepresented students to their overrepresented classmates in active-learning and traditional-lecturing treatments. This search resulted in data on student examination scores from 15 studies (9,238 total students) and data on student failure rates from 26 studies (44,606 total students). Bayesian regression analyses showed that on average, active learning reduced achievement gaps in examination scores by 33% and narrowed gaps in passing rates by 45%. The reported proportion of time that students spend on in-class activities was important, as only classes that implemented high-intensity active learning narrowed achievement gaps. Sensitivity analyses showed that the conclusions are robust to sampling bias and other issues. To explain the extensive variation in efficacy observed among studies, we propose the heads-and-hearts hypothesis, which holds that meaningful reductions in achievement gaps only occur when course designs combine deliberate practice with inclusive teaching. Our results support calls to replace traditional lecturing with evidence-based, active-learning course designs across the STEM disciplines and suggest that innovations in instructional strategies can increase equity in higher education.


2014 ◽  
Vol 2014 (1) ◽  
pp. 1113-1125
Author(s):  
Xiaolong Geng ◽  
Michel C. Boufadel

ABSTRACT In April 2010, the explosion of the Deepwater Horizon (DWH) drilling platform led to the release of nearly 4.9 million barrels of crude oil into the Gulf of Mexico. The oil was brought to the supratidal zone of beaches (landward of the high tide line) by waves during storms, and was buried during subsequent storms. The objective of this paper is to investigate the biodegradation of subsurface oil in a tidally influenced sand beach located at Bon Secour National Wildlife Refuge and polluted by the DWH oil spill. Two transects were installed perpendicular to the shoreline within the supratidal zone of the beach. One transect had four galvanized steel piezometer wells to measure the water level. The other transect had four stainless steel multiport sampling wells that were used to collect pore water samples below the beach surface. The samples were analyzed for dissolved oxygen (DO), nitrogen, and redox conditions. Sediment samples were also collected at different depths to measure residual oil concentrations and microbial biomass. As the biodegradation of hydrocarbons was of interest, a biological model based on Monod kinetics was developed and coupled to the transport model MARUN, which is a two dimensional (vertical slice) finite element model for water flow and solute transport in tidally influenced beaches. The resulting coupled model, BIOMARUN, was used to simulate the biodegradation of total n-alkanes and polycyclic aromatic hydrocarbons (PAHs) trapped as residual oil in the unsaturated zone. Model parameter estimates were constrained by published Monod kinetics parameters. The field measurements, such as the concentrations of the oil, microbial biomass, nitrogen, and DO, were used as inputs for the simulations. The biodegradation of alkanes and PAHs was predicted in the simulation, and sensitivity analyses were conducted to assess the effect of the model parameters on the modeling results. Simulation results indicated that n-alkanes and PAHs would be biodegraded by 80% after 2 ± 0.5 years and 3.5 ± 0.5 years, respectively.


2021 ◽  
Author(s):  
◽  
Denise M. Kirby

<p>Mindfulness is defined as a fundamental way of being, a way of relating to all of one‟s experiences whether positive, negative, or neutral, with an attentive, curious, and nonevaluative mindset (Kabat-Zinn, 2003; Williams, Teasdale, Segal, & Kabat-Zinn, 2007). The basic premise underlying mindfulness is that accepting moment-to-moment experiences as they arise, with an open-minded disposition, helps protect against psychological symptoms such as anxiety, depression, stress, and adverse ruminative thoughts (Baer, 2003; Hofmann, Sawyer, Witt, & Oh, 2010). In addition, mindfulness is also thought to enhance psychological health and wellbeing, such as increases in emotional intelligence, body image, and positive mental health (Bohlmeijer, ten Klooster, Fledderus, Veehof, & Baer, 2011; Stewart, 2004). Extant mindfulness literature has largely focused on the outcomes that are associated with mindfulness-based interventions designed to enhance an individual‟s state of mindfulness, though, less is known about naturally occurring dispositional mindfulness. According to some researchers, mindfulness manifesting as a trait (or individual difference characteristic) is an inherent human capacity that varies between individuals (Brown & Ryan, 2003; Kabat-Zinn, 1990). To examine the complexities of trait mindfulness, it is necessary to have a comprehensive understanding of the operationalisation of the construct. To date, however, methods for assessing mindfulness have received little attention in psychological research (Baer, Smith, Hopkins, Krietemeyer, & Toney, 2006). Across a series of three empirical studies (Chapters 2 through 4), this thesis aimed to explore the naturally occurring construct of trait mindfulness using two measures designed to capture this multifaceted phenomenon – the Five-Facet Mindfulness Questionnaire (FFMQ; Baer et al., 2006) and the Trait Toronto Mindfulness Scale (Trait TMS; Davis, Lau, & Cairns, 2009). Study 1 (Chapter 2) began with a psychometric analysis of the FFMQ, and then explored how the tendency to be mindful related to both positive and negative psychological outcomes. Contrary to expectations, a confirmatory factor analysis of the FFMQ in this sample yielded a 2-factor model of mindfulness (Nonjudging Awareness and Describing) as opposed to the five factors proposed by Baer et al. (2006). Further analyses revealed that the factor of Nonjudging Awareness appeared to be the most potent aspect of mindfulness that positively predicted happiness, life satisfaction, and adaptive coping (positive psychological outcomes) and negatively predicted depression, anxiety, and maladaptive coping (negative psychological outcomes). In Study 2 (Chapter 3), both the FFMQ and the Trait TMS were compared and tested for factor structure across three points in time on a community sample of individuals (N = 319). Both measurement invariance and convergent validity were explored, with results supporting expectations. First, results yielded a 5-factor structure of the FFMQ and a 2-factor structure of the Trait TMS. Second, factorial invariance on both mindfulness measures was evidenced across three time points. Third, results indicated moderate construct overlap between these two trait measures of mindfulness, i.e., they tapped a similar construct, as was expected. Study 3 (Chapter 4) sought to extend on the previous two studies‟ findings by 1) exploring whether trait mindfulness was positively predictive of positive psychological outcomes and negatively predictive of negative outcomes across time, and 2) comparing the psychological functioning of two groups – individuals with and without meditation experience – across time. Using the same sample from Study 2, trait mindfulness, as measured by the FFMQ and Trait TMS, supported hypotheses, though some predicted findings appeared to be manifested between the second and third time points only. Moreover, results indicated that meditation experience may protect against maladaptive psychological outcomes such as depression and anxiety, while increasing feelings of subjective happiness and life satisfaction (supporting traditional meditation ideology and current empirical literature). In combination, the present research supports and contributes novel perspectives to the existing body of mindfulness literature by suggesting that establishing ways to amplify the naturally occurring trait of mindfulness, although challenging, will prove fertile within Western clinical psychology, which looks to influence an individual‟s psychological wellbeing. Chapter 5 concludes this thesis with a discussion of the implications of the combined series of studies. Strengths and limitations of this research were considered, and future directions were proposed.</p>


2018 ◽  
Author(s):  
Timo Gnambs ◽  
Markus Appel

Many societies are on the brink of a robotic era. In the near future, various autonomous computer systems are expected to be part of many people’s daily lives. Because attitudes influence the adoption of new technologies, we studied the attitudes towards robots in the European Union between 2012 and 2017. Using representative samples from 27 countries (three waves, total N = 80,396), these analyses showed that, within five years, public opinions regarding robots exhibited a marked negative trend. Respondents became more cautious towards the use of robots. This tendency was particularly strong for robots at the workplace, which are, despite the drop, still more positively evaluated than robots performing surgeries or autonomous cars. Attitudes were more positive among men and people in white-collar jobs. Moreover, countries with a larger share of older citizens evaluated robotic assistance more favorably. In general, these results highlight increasing reservations towards autonomous robotic systems in Europe.


2021 ◽  
Author(s):  
Bjørn E. Holstein ◽  
Mogens Trab Damsgaard ◽  
Katrine Rich Madsen ◽  
Trine Pagh Pedersen ◽  
Mette Toftager

Abstract Chronic backpain among adolescents is important because the prevalence and the burden of disability is high. Chronic backpain tracks into adulthood and is associated with several health problems. The objective was to study trends in the prevalence of chronic backpain among adolescents 1991-2018, to examine the association with socioeconomic status (SES) and whether this association changed over time. The study used data from eight comparable cross-sectional school-surveys of nationally representative samples of 11-15-year-olds in 1991, 1994, 1998, 2002, 2006, 2010, 2014 and 2018, the Danish arm of the international Health Behaviour in School-aged Children (HBSC) study. The participation rate was 88.0%, n=29,952. Chronic backpain was defined as self-reported backpain daily or several days a week during the last six months. The prevalence of chronic backpain was 11.1%, significantly increasing from 8.9% in 1991 to 11.7% in 2018. The OR for chronic backpain was 1.20 (1.10-1.31) in middle and 1.56 (1.41-1.73) in low compared to high OSC. Sensitivity analyses with two other cut-points for backpain frequency showed similar associations. Conclusion: Chronic backpain is common among adolescents and the prevalence increased from 1991 to 2918. The prevalence was highest in lower SES families. We recommend increased efforts to prevent chronic backpain.


2021 ◽  
Author(s):  
Martin Voracek

Frischenschlager, Haidinger, and Mitterauer (FHM hereafter) reported an evaluation study of predictors for success in Austrian medical undergraduates (Croat Med J. 2005;46:58-65). Since most of the international literature on medical education topics still comes from the USA and the UK, in principle this Central European contribution is much needed and to be welcomed. However, several obvious and grave study limitations make the data and findings of FHM markedly less useful than the authors suggest. Relatedly, in the current absence of other Austrian studies on that topic it is foreseeable that the FHM evidence could be used to inform policy decisions in regards to medical education issues in Austria and, possibly so, elsewhere in Central Europe. I strongly opine that, due to the study’s design deficiencies, this should be avoided. The main purpose of the present comment is therefore to elucidate the shortcomings of this study as well as to stimulate further discussion and research.


Author(s):  
Sylvia Aponte-Hao ◽  
Sabrina T. Wong ◽  
Manpreet Thandi ◽  
Paul Ronksley ◽  
Kerry McBrien ◽  
...  

IntroductionFrailty is a medical syndrome, commonly affecting people aged 65 years and over and is characterized by a greater risk of adverse outcomes following illness or injury. Electronic medical records contain a large amount of longitudinal data that can be used for primary care research. Machine learning can fully utilize this wide breadth of data for the detection of diseases and syndromes. The creation of a frailty case definition using machine learning may facilitate early intervention, inform advanced screening tests, and allow for surveillance. ObjectivesThe objective of this study was to develop a validated case definition of frailty for the primary care context, using machine learning. MethodsPhysicians participating in the Canadian Primary Care Sentinel Surveillance Network across Canada were asked to retrospectively identify the level of frailty present in a sample of their own patients (total n = 5,466), collected from 2015-2019. Frailty levels were dichotomized using a cut-off of 5. Extracted features included previously prescribed medications, billing codes, and other routinely collected primary care data. We used eight supervised machine learning algorithms, with performance assessed using a hold-out test set. A balanced training dataset was also created by oversampling. Sensitivity analyses considered two alternative dichotomization cut-offs. Model performance was evaluated using area under the receiver-operating characteristic curve, F1, accuracy, sensitivity, specificity, negative predictive value and positive predictive value. ResultsThe prevalence of frailty within our sample was 18.4%. Of the eight models developed to identify frail patients, an XGBoost model achieved the highest sensitivity (78.14%) and specificity (74.41%). The balanced training dataset did not improve classification performance. Sensitivity analyses did not show improved performance for cut-offs other than 5. ConclusionSupervised machine learning was able to create well performing classification models for frailty. Future research is needed to assess frailty inter-rater reliability, and link multiple data sources for frailty identification.


Sign in / Sign up

Export Citation Format

Share Document