Browsing by Subject "meta-analysis"

Now showing 1 - 15 of 15

Bayesian Causal Inferencee In Meta-Analysis
(2019-05) Zhou, Jincheng
While the randomized clinical trial (RCT) is the gold standard for investigating the effect of a medical intervention, noncompliance to assigned treatments can threaten a trial's validity. Noncompliance, if not appropriately controlled, can introduce substantial bias into the estimate of treatment effect. The complier average causal effect (CACE) approach provides a useful tool for addressing noncompliance, where CACE measures the effect of an intervention in the latent subgroup of the study population that complies with its assigned treatment (the compliers). Meta-analysis of RCTs has become a widely-used statistical technique to combine and contrast results from multiple independent studies. However, no existing methods can effectively deal with heterogeneous noncompliance in meta-analysis of RCTs. For example, the commonly used meta-analysis regression methods investigate the impact of study-level variables (e.g., mean age of the study population) on the study-specific treatment effect size by assuming the study-level covariates to be fixed. However, noncompliance rates generally differ between treatment groups within a study and are commonly considered as random rather than fixed post-randomization variables. In addition, noncompliance may dynamically interact with the primary outcome and thus affect the response to treatment. Thus, meta-regression methods are not suitable to controlling for noncompliance. This thesis focuses on developing Bayesian methods to estimate CACE in meta-analysis of RCTs with binary or ordinal outcomes. Bayesian hierarchical random effects models are developed to appropriately account for the inherent heterogeneity in treatment effect and noncompliance between studies and treatment groups. We first present a Bayesian hierarchical model to estimate the CACE where heterogeneous compliance rates are available for each study. Second, we extend our approach to deal with incomplete noncompliance when some RCTs do not report noncompliance data. The results are illustrated by a re-analysis of a meta-analysis comparing the effect of epidural analgesia in labor versus no or other analgesia in labor on the outcome cesarean section, where noncompliance varies substantially between studies. Simulations are performed to evaluate the performance of the proposed approach and to illustrate the importance of including appropriate random effects by showing the impact of over- and under-fitting. Furthermore, we develop an R package, BayesCACE, to provide user-friendly functions to implement CACE analysis for binary outcomes based on the proposed Bayesian hierarchical models. This package includes flexible functions for analyzing data from a single RCT and from a meta-analysis of multiple RCTs with either complete or incomplete noncompliance data. The package also provides various functions for generating forest, trace, posterior density, and auto-correlation plots, and to review noncompliance rates, visually assess the model, and obtain study-specific and overall CACEs.
Complete Data and Analysis for: Fungicide Effectiveness on Soybean Rust in the Southeastern United States 2004-2014: A Meta-Analysis
(2016-08-02) ArchMiller, Althea A; Delaney, Mary A; Delaney, Dennis P; Wilson, Alan E; Sikora, Edward J; althea.archmiller@gmail.com; ArchMiller, Althea A
Soybean rust is a concern to soybean growers and management of soybean rust primarily depends on disease scouting and the timely use of fungicides. The goal of this study was to evaluate the efficacy of fungicide-use through a quantitative meta-analysis of data compiled from published and unpublished soybean fungicide trials across the southern United States from 2004 to 2014. The data included in this repository includes the complete dataset as a comma-separated-value file and all Program R code necessary to replicate the data processing, analysis, and graphing.
The Effects of Early Numeracy Interventions for Students in Preschool and Early Elementary: A Meta-Analysis
(2017-12) Nelson, Gena
The purpose of this meta-analysis was to examine the effectiveness of early numeracy interventions for young students, including students with disabilities or those at-risk for mathematics difficulty (MD). This study evaluated preschool, kindergarten, and first-grade interventions on early numeracy content, instructional features, and methodological components that improved students’ mathematics achievement. A total of 33 studies met inclusion criteria for this meta-analysis, with 51 treatment groups. Excluding outliers, the average weighted effect size for numeracy interventions across 49 treatment groups was moderate (g = 0.63), and the 95% confidence interval did not include zero [0.50, 0.73]. Results indicated that early numeracy interventions that included preschool and kindergarten students produced larger treatment effects than interventions with first-grade participants; in addition, treatment effects were slightly higher on average for students identified as at-risk for MD according to low socio-economic status and performance greater than the 25th percentile on a mathematics screener, compared to students who were identified as typically achieving or at-risk for MD according to performance below the 25th percentile. The results of the final meta-regression model for the total sample of studies indicated that the following predictors accounted for the most between-studies variance: concrete-representational-abstract instructional framework, intervention duration, risk status of participants, and the inclusion of counting with one-to-one correspondence in the intervention content (Pseudo R2 = 75%). Directions for future research on conducting interventions are provided, and implications for educators implementing early numeracy interventions are discussed.
Frequency of persistent tooth pain after root canal therapy: a systematic review and meta-analysis
(Elsevier, 2010-02) Nixdorf, DR; Moana-Filho, EJ; Law, AS; McGuire, LA; Hodges, JS; John, MT
Little is known about the frequency of persistent pain after endodontic procedures even though pain is a core patient-oriented outcome. We estimated the frequency of persistent pain, regardless of etiology, after endodontic treatment. METHODS: Persistent tooth pain was defined as pain present > or = 6 months after endodontic treatment. Endodontic procedures included in the review were pulpectomy, nonsurgical root canal treatment, surgical root canal treatment, and retreatment. Four databases were searched electronically complemented by hand searching. Two independent reviewers determined eligibility, abstracted data, and assessed study quality. A summary estimate of persistent all-cause tooth pain frequency was established by using a random-effects meta-analysis. Using subgroup analyses, we explored the influence of treatment approach (surgical/nonsurgical), longitudinal study design (prospective/retrospective), follow-up rate, follow-up duration, initial treatment versus retreatment, and quality of reporting (Strengthening the Reporting of Observational Studies in Epidemiology rankings) on the pain frequency estimate. RESULTS: Of 770 articles retrieved and reviewed, 26 met inclusion criteria. A total of 5,777 teeth were enrolled, and 2,996 had follow-up information regarding pain status. We identified 168 teeth with pain and derived a frequency of 5.3% (95% confidence interval, 3.5%-7.2%, p < 0.001) for persistent all-cause tooth pain. High and statistically significant heterogeneity among studies (I2 = 80%) was present. In subgroup analysis, prospective studies had a higher pain frequency (7.6%) than retrospectives studies did (0.9%). Quality of study reporting was identified as the most influential reason for study heterogeneity. CONCLUSIONS: The frequency of all-cause persistent tooth pain after endodontic procedures was estimated to be 5.3%, with higher report quality studies suggesting >7%.
The impact of self-affirmation on defensive processing of health messages: A meta-analysis
(2016-03) He, Xiaofei
This meta-analysis studies the effects of self-affirmation on cognitive, affective, and behavioral responses to threatening health messages. It analyzes how the effects vary as a function of three moderators: self-affirmation domains, health topics, and self-relevance levels. In addition, this analysis examines the role of emotions in the self-affirmation process. Effect sizes for 11 outcome variables were extracted from 55 studies and analyzed (N = 10,611). I performed fixed-effect and random-effects models to examine the main effect and moderating effects. Both models indicated small but statistically significant positive effects of self-affirmation in increased perceived message effectiveness, perceived susceptibility, response efficacy, and behavior. The results lend empirical support to self-affirmation as an effective intervention strategy. Moderator analyses with both fixed-effect and random-effects models revealed that self-affirmation was most effective (1) when we used the desirable traits self-affirmation domain; (2) when we exposed participants to messages of unhealthy behaviors cessation; and (3) among participants with low self-relevance. The two most commonly used self-affirmation domains (i.e., desirable traits and values), did not work equally well in reducing defensiveness. Moreover, these two domains were not effective in restoring self-integrity when applied to high-self-relevance populations, or to reducing defensive responses to messages of unhealthy behavior cessation. Meta-analytic review of the role that emotions play in the self-affirmation process shows that self-affirmation was effective in reducing negative emotions aroused by delivery of threatening health messages. However, the evidence of a mediating role of (positive) emotions in the self-affirmation process is scant.
Intraindividual Variability in Personality Research: Considering Time, Measurement, and the Interpersonal Context
(2024-05) Nguyen, Le Phuong Linh
Personality research has established robust associations between traits and a variety of life outcomes. Nevertheless, much of the literature relies on the Big Five traits, which broadly encapsulate important patterns of psychological individual differences. As a result, this broad conceptualization often leads to weaker associations with outcomes within specific domains. The current dissertation offers a comprehensive examination of different ways to expand upon traditional trait research. This includes combining multiple levels of personality constructs both within and outside of the Big Five framework, multiple perspectives through self and informant reports, and multiple timescales from one-time trait measures to dynamic state fluctuations and longitudinal trait changes. The primary focus is on intraindividual variability, or how people change in their psychological processes across time, and its relevance within the highly influential life domain of romantic relationships. Study 1 presents a preregistered meta-analytic review across k = 88 independent samples (N = 20,813) of the association between personality traits at both the domain (Big Five) and metatrait (Stability and Plasticity) levels with affective variability. We found a positive association between affective variability and Plasticity as well as its underlying traits. However, the pattern of findings was mixed and valence-specific for the Stability traits, and this metatrait itself was negatively associated with variability in Positive Affect but positively so for variability in Negative Affect. Study 2 further investigated intraindividual variability in psychological processes by examining assortative mating patterns in 138 established romantic couples using experience sampling methods of personality and affective states across 35 time points during a 7-day period. We found evidence for assortative mating based on both baseline traits and dynamic states. However, there was more evidence for perceived similarities than actual similarities at baseline, and there was much more evidence for dynamic similarities on states than baseline similarities on traits. There was also evidence for a complementarity effect or negative between-partner correlation on Volatility. Study 3 expanded the timescale from dynamic short-term state fluctuations to longitudinal trajectories of trait change across multiple months and years, examining assortative mating patterns in two complementary samples of early dating couples (N = 184) and married or cohabiting couples (N = 168). We found evidence for assortative mating across various relationship-specific characteristics both at baseline and longitudinally, which were often stronger in magnitude than assortment on Big Five traits. Consistent with Study 2, couples often perceived each other to be more similar than their actual similarity indicated. Nonetheless, in line with the literature, there was little evidence to support the benefits of between-partner similarity for relationship quality, especially after controlling for actor and partner effects of both partners’ individual characteristics. Altogether, this research program provides a broad and thorough examination of intraindividual variability in general as well as in the interpersonal context, and in doing so contributes to both the substantive body of literature and the methodological considerations needed when investigating these personality processes.
Meta-analytic data of interventions to improve test-taking effort and test performance on low-stakes educational assessments
(2019-06-04) Rios, Joseph A; jrios@umn.edu; Rios, Joseph A
The data come from a meta-analysis investigating the impact of interventions for improving test-taking effort and test performance on low-stakes educational assessments. The data were collected from 53 studies that (a) used a treatment-control group design, (b) administered a low-stakes group-based educational assessment, and (c) evaluated test-taking effort and/or test performance as outcomes. In total, there are 60 and 105 effect sizes for test-taking effort and test performance, respectively, along with study, sample, and moderator characteristics.
Online Supplement for 'Meta-Analytic and Empirical Estimates of the Resource Depletion Effect Size'
(2016-04-05) Yost, Tyler A.; yostx038@umn.edu; Yost, Tyler A.
The files included here are an online supplement to a dissertation titled 'Meta-Analytic and Empirical Estimates of the Resource Depletion Effect Size'. These files consist of R code to implement meta-analyses and Monte Carlo simulations, in addition to an Excel file containing a meta-analytic dataset.
Personality and its Impacts across the Behavioral Sciences: A Quantitative Review of Meta-Analytic Findings
(2017-05) Wilmot, Michael
Personality has consequences. Following the emergence of and scholarly convergence around the Five-Factor Model (FFM), or Big Five, some 35 years ago, research interest in personality traits has exploded across the behavioral sciences. Meta-analyses reporting Big Five (i.e., Emotional Stability, Agreeableness, Conscientiousness, Extraversion, and Openness/Intellect) relations have so proliferated that a quantitative second-order review was needed. The purpose of this dissertation was to conduct such a review. Data were gathered from an exhaustive search (through July 2016) of 167 published Big Five meta-analyses, which reported empirical relations to 712 unique correlate, behavioral, and outcome variables. A multi-hurdle selection process was used to screen variables for study inclusion, and a content-based coding procedure was used to organize variables into a set of four theoretically meaningful “meta-categories”—Well-Being, Performance, Leadership, and Counterproductivity—which were examined in series of three studies. Study 1 used procedures from first-order and second-order psychometric meta-analysis to estimate univariate relations for the Big Five traits. Empirical effect size benchmarks for interpreting trait relations were also developed. Study 2 built on the prior study by estimating univariate relations for the two metatraits, Stability and Plasticity. Results represent the most comprehensive nomological network of metatrait relations in the literature, and provide evidence of their wide-ranging theoretical and empirical relevance (e.g., Stability was the strongest predictor of Counterproductivity variables, and Plasticity was the strongest predictor of Leadership variables). Finally, Study 3 examined multivariate effects of both Big Five and metatraits models. Dominance analysis was also used to examine traits’ relative contribution to overall prediction. Results indicate that both trait models contributed substantial variance to predicting variables that are consequential and fundamental to human interest, and that most of these variables were multiply determined by at least two or three traits. Together, these studies summarize and advance knowledge about personality and its impacts across the behavioral sciences.
Personality Traits and Cognitive Abilities Relations Database
(2023-10-26) Stanek, Kevin C.; Ones, Deniz S.; stane040@umn.edu; Stanek, Kevin C.
This database consists of personality-ability correlations that were amassed for a set of meta-analyses. Contributing materials spanned journal articles, theses, personal communications, archival datasets, conference presentations, and others sources. Further details about how these data were curated, cleaned, and organized as well as results of the meta-analyses based on these data are presented in Of Anchors & Sails: Personality-Ability Trait Constellations (2023).
Quantitative Methods for Evidence Building in Clinical Pharmacology and Pharmaceutical Outcomes Research
(2021-05) Margraf, David
A variety of methods are employed to build evidence in pharmacology and pharmaceutical outcomes research. Descriptive and inferential statistics are used to describe the data and generalize findings to populations. Regression models, propensity score adjustment, and meta-analysis extend upon the quantitative approach to building evidence. Topic areas in this dissertation include demonstrating the application of these methods to a comparison of three-factor prothrombin complex concentrate versus four-factor prothrombin complex concentrate for emergent warfarin reversal via a propensity score adjusted retrospective cohort study and a systematic review and meta-analysis to address clinical problems and improve health outcomes. Also presented are the pharmacokinetics of intravenous N-acetylcysteine, Cysteine, and Glutathione and the effect of N-acetylcysteine as a reducing agent in Parkinson’s disease and Gaucher disease. While quantitative methods help us explore, explain, and generalize from data, it is imperative to consider the clinical relevance of the findings. We found that four-factor prothrombin complex concentrate is preferred for emergent warfarin reversal. This is a finding is useful in real-world patient care. Also, increased N-acetylcysteine plasma concentrations and Glutathione redox ratio are related, which could be used to optimize dosing in future studies. These examples are described in detail as examples of applications of quantitative methods.
R Code and Output Supporting: Computational reproducibility in The Wildlife Society's flagship journals
(2019-06-05) ArchMiller, Althea A; Johnson, Andrew D; Nolan, Jane; Edwards, Margaret; Elliot, Lisa H; Ferguson, Jake M; Iannarilli, Fabiola; Velez, Juliana; Vitense, Kelsey; Johnson, Douglas H; Fieberg, John R; ALTHEA.ARCHMILLER@GMAIL.COM; ArchMiller, Althea A
The goal of this study was to gauge the level of computational reproducibility, which is the ability to reach the same results using the same data and analysis methods, in the field of wildlife sciences. We randomly selected 80 papers published in the Journal of Wildlife Management and Wildlife Society Bulletin between 1 June 2016 and 1 June 2018. Of those for which we could obtain data, we attempted to reproduce their quantitative results using the original methods and data. The dataset shared in this repository is the de-identified results of our review, and the code provided here produces the results and figures in our published manuscript.
Risk Factors for Abdominal Aortic Aneurysm and Larger Infrarenal Aortic Diameters in a General Population
(2018-07) Yao, Lu
Abdominal aortic aneurysms (AAAs) comprise an important public health issue, which could be reduced by primary prevention. Identifying AAA risk factors is critical for developing effective preventive strategies. Previous epidemiologic studies have suggested that some risk factors for atherosclerotic cardiovascular disease are also associated with increased risk of incident AAAs, including advanced age, male gender, white race, greater height, smoking, hypertension, dyslipidemia, and some biomarkers related to inflammation and hemostasis. Some observational studies showed an inverse relationship between diabetes and AAA; while others did not show an association. The inverse relationship between diabetes and AAA is considered counterintuitive in the context of diabetes being a risk factor for various cardiovascular diseases. To better understand the etiology of AAA, further investigation on the relation between atherosclerosis and AAA is warranted. Also, the relation between diabetes and AAA needs to be studied further. With the exception of screening studies where AAAs were defined commonly as a maximum infrarenal aortic diameter (IAD) ≥ 3 cm, in most existing epidemiologic studies, AAAs were obtained through medical records and death certificates. This approach ascertains clinical AAAs that were either symptomatic or at least clinically detected. However, large screening studies have suggested that most AAAs are asymptomatic, even though aortic size often expands rapidly and many asymptomatic AAAs may eventually become symptomatic. Furthermore, an increased IAD between 2.3 and 3 cm has been associated with higher risk of future AAA and other cardiovascular events. Thus, examining the determinants of elevated IADs (i.e. IAD ≥ 2.2) among individuals without clinical or asymptomatic AAAs is potentially important to the prevention of AAAs. Manuscript 1 examined the associations of carotid atherosclerosis and stiffness with later AAAs in ARIC. We used carotid intima-media thickness (1987-1992) and atherosclerotic plaque (1987-1989) as indices of carotid atherosclerosis, and used carotid Beta Index (1990-1992) to represent carotid distensibility. We identified 542 incident, clinical AAAs during follow-up through 2011 using hospital discharge codes, Medicare outpatient diagnoses, or death certificates during 22.5 years of follow-up. After multivariable adjustment, the presence of carotid atherosclerotic plaque at baseline was associated with 1.31 (95% CI: 1.10 - 1.57; P: 0.003) times higher risk of clinical AAA. Greater carotid intima-media thickness and Beta Index were also associated with clinical AAA with a dose-response across quartiles (P trend for both: 0.006; hazard ratios [95% CI] for the highest vs. lowest quartiles: 1.55 [1.13 - 2.11] and 1.68 [1.16 - 2.43], respectively). The results suggest that indices of greater carotid atherosclerosis and lower carotid distensibility are markers of increased AAA risk. Manuscript 2 explored risk factors for an elevated IAD (IAD ≥ 2.2 cm) in the absence of AAA in 5620 ARIC participants who attended an abdominal ultrasound screening in 2011-2013. We assessed a variety of risk factors and created derived variables to capture their long-term cumulative effects (over 1987-2013). In the model with adjustment for AAA risk factors, men (vs. women) had 2.50 (95% CI: 1.90, 3.28) times higher odds of having an elevated IAD, and participants with long-term diabetes (vs. non-diabetics) had 0.52 (0.35, 0.77) times lower odds. Height, waist circumference and smoking pack-years were positively associated with elevated IADs [ORs (95% CI) for the highest vs. lowest quintiles of each risk factor: 1.93 (1.36, 2.75), 1.67 (1.28, 2.19) and 1.62 (1.26, 2.08), respectively]. Other factors were not associated with elevated IAD. In summary, male sex, smoking, greater height, larger waist circumference and not having diabetes were associated with elevated IAD among persons without an AAA. The findings highlight the potential for primary prevention of AAA through control of these factors. Manuscript 3 represents a meta-analysis of prospective cohort studies and case-control studies to examine further the relation between diabetes and AAAs. We searched for English literature from online database search (MEDLINE (1966-), EMBASE and Web of Science) plus a manual examination of references in selected articles as of Feb 2018, and included a total of 12 cohorts with 11,410 AAAs in 2,665,121 adult participants and 4 case-control studies with 1,065 AAAs and 12,074 controls who met pre-determined eligibility criteria in the meta-analyses. A DerSimonian and Laird random effects model pooled association estimates and their 95% confidence intervals from studies. Diabetes was inversely associated with the risk of AAA (pooled relative risk: 0.56; 95% confidence interval: 0.50 - 0.63). Results were similar in the subgroup analyses by sex (male/female), setting (population/clinical), and study design (cohort/case-control). In summary, in contrast with diabetes being a risk factor for most cardiovascular diseases, diabetes appears to be strongly and inversely associated with the risk of AAA. In summary, my dissertation studies filled a gap of literature and further assessed AAA etiology by completing the three manuscripts. The three studies have potential to improve understanding of the etiology and early prevention of AAAs at the population level. Findings from my dissertation studies may offer a strategy to clinically identify high-risk individuals.
Sources of Variance In Reading Comprehension Research: the Role of Measures and Interventions
(2020-05) Diggs, Calvary
The purpose of this study was to examine if differences in reading comprehension measures’ response formats were associated with differential outcomes for reading comprehension interventions. Specifically, this study used meta-analysis to evaluate the overall treatment effect of reading comprehension interventions, the association between a measure’s response format and measured intervention outcomes, and whether specific intervention effects varied based on the measure’s response format. A systematic review of the literature identified 66 published and unpublished research reports and studies conducted since 2000. All studies administered a reading comprehension intervention for students in the primary grades and measured the effects using a reading comprehension measure. Meta-analytic findings suggested an overall positive effect of reading comprehension interventions for both intervention to control group comparisons at posttest (Hedge’s g = 0.20) and pretest to posttest comparisons in the intervention group (Hedge’s g = 0.71). The response format of a reading comprehension measure, specifically retell/summary formats, was significantly associated with intervention outcomes, even after controlling for purposively selected variables. Findings also indicated that improving background knowledge and multicomponent interventions were significantly associated with performance on measures of reading comprehension with retell/summary response formats. The results of this study provide additional evidence that measures using the retell/summary response formats value reading comprehension differently, specifically in the context of interventions. Findings may also be used to caution against the interchangeable use of retell/summary formats with other measures of reading comprehension.
Statistical methods for multivariate meta-analysis of diagnostic tests
(2015-05) Ma, Xiaoye
Accurate diagnosis is often the first step towards the treatment and prevention of disease. Many quantitative comparisons of diagnostic tests have relied on meta-analyses, which are statistical methods to synthesize all available information in various clinical studies. In addition, in order to effectively compare the growing number of diagnostic tests for a specific disease, innovative and efficient statistical methods to simultaneously compare multiple diagnostic tests are urgently needed for physicians and patients to make better decisions. In the literature of meta-analysis of diagnostic tests (MA-DT), discussions have been focused on statistical models under two scenarios: (1) when the reference test can be considered a gold standard, and (2) when the reference test cannot be considered a gold standard. We present an overview of statistical methods for MA-DT in both scenarios. This dissertation covers both conventional and advanced multivariate approaches for the first scenario, and a latent class random effects model when the reference test itself is imperfect. As study design and populations vary, the definition of disease status or severity could differ across studies. A trivariate generalized linear mixed model (TGLMM) has been proposed to account for this situation; however, its application is limited to cohort studies. In practice, meta-analytic data is often a mixture of cohort and case-control studies. In addition, some diagnostic accuracy studies only select a subset of samples to be verified by the reference test, which is known as potential source of partial verification bias in single studies. The impact of this bias on a meta-analysis has not been investigated. We propose a novel hybrid Bayesian hierarchical model to combine cohort and case-control studies, and correct partial verification bias at the same time. A recent paper proposed an intent-to-diagnose approach to handle non-evaluable index test results, and discussed several alternative approaches. However, no simulation studies have been conducted to test the performance of the methods. We propose an extended TGLMM to handle non-evaluable index test results, and examine the performance of the intent-to-diagnose approach, the alternative approaches, and the proposed approach by extensive simulation studies. To compare the accuracy of multiple tests in a single study, three designs are commonly used: 1) the multiple test comparison design; 2) the randomized design; and 3) the non-comparative design. Existing MA-DT methods have been focused on evaluating the performance of a single test by comparing it with a reference test. The increasing number of available diagnostic instruments for a disease condition and the different study designs being used have generated the need to develop an efficient and flexible meta-analysis framework to combine all designs for simultaneous inference. We develop a missing data framework and a Bayesian hierarchical model for network meta-analysis of diagnostic tests (NMA-DT), and offer key advantages over traditional MA-DT methods.

University Digital Conservancy

Browse by Subject

Browsing by Subject "meta-analysis"