Bayesian Multilevel Models Research Articles

Migratory birds have higher prevalence and richness of avian haemosporidian parasites than residents

Individuals of migratory species may be more likely to become infected by parasites because they cross different regions along their route, thereby being exposed to a wider range of parasites during their annual cycle. Conversely, migration may have a protective effect since migratory behaviour allows hosts to escape environments presenting a high risk of infection. Haemosporidians are one of the best studied, most prevalent and diverse groups of avian parasites, however the impact of avian host migration on infection by these parasites remains controversial. We tested whether migratory behaviour influenced the prevalence and richness of avian haemosporidian parasites among South American birds. We used a dataset comprising ~ 11,000 bird blood samples representing 260 bird species from 63 localities and Bayesian multi-level models to test the impact of migratory behaviour on prevalence and lineage richness of two avian haemosporidian genera (Plasmodium and Haemoproteus). We found that fully migratory species present higher parasite prevalence and higher richness of haemosporidian lineages. However, we found no difference between migratory and non-migratory species when evaluating prevalence separately for Plasmodium and Haemoproteus, or for the richness of Plasmodium lineages. Nevertheless, our results indicate that migratory behaviour is associated with an infection cost, namely a higher prevalence and greater variety of haemosporidian parasites.

Support provided by elderly in Italy: a hierarchical analysis of ego networks controlling for alter-overlapping.

Providing support outside the household can be considered an actual sign of an active social life for the elderly. Adopting an ego–network perspective, we study support Italian elders provide to kin or non–kin. More specifically, using Italian survey data, we build the ego–centered networks of social contacts elders entertain and the ego–networks of support elders provide to other non–cohabitant kin or non–kin. Since ego–network data are inherently multilevel, we use Bayesian multilevel models to analyze variation in support ties, controlling for the characteristics of elders and their contacts. This modeling strategy enables dealing with sparseness and alter–alter overlap in the ego support network data and to disentangle the effects related to the ego (the elder), the dyad ego–alter, the kind of support provided, as well as social contacts and contextual variables. The results suggest that the elderly in Italy who provide support outside their household — compared to all elders in the sample — are younger, healthier, more educated, and embedded in a more diversified ego–network of social contacts. The latter also conveys both the type and the recipient of the support, with the elderly who entertain few relationships with kin being more prone to provide aid to non–kin. Further, a “peer homophily” effect in directing elder support to a non–kin is also found.

Can Visualization Alleviate Dichotomous Thinking? Effects of Visual Representations on the Cliff Effect.

Common reporting styles for statistical results in scientific articles, such as p-values and confidence intervals (CI), have been reported to be prone to dichotomous interpretations, especially with respect to the null hypothesis significance testing framework. For example when the p-value is small enough or the CIs of the mean effects of a studied drug and a placebo are not overlapping, scientists tend to claim significant differences while often disregarding the magnitudes and absolute differences in the effect sizes. This type of reasoning has been shown to be potentially harmful to science. Techniques relying on the visual estimation of the strength of evidence have been recommended to reduce such dichotomous interpretations but their effectiveness has also been challenged. We ran two experiments on researchers with expertise in statistical analysis to compare several alternative representations of confidence intervals and used Bayesian multilevel models to estimate the effects of the representation styles on differences in researchers' subjective confidence in the results. We also asked the respondents' opinions and preferences in representation styles. Our results suggest that adding visual information to classic CI representation can decrease the tendency towards dichotomous interpretations - measured as the 'cliff effect': the sudden drop in confidence around p-value 0.05 - compared with classic CI visualization and textual representation of the CI with p-values. All data and analyses are publicly available at https://github.com/helske/statvis.

Calibration of surgical tools using multilevel modeling with LINEX loss function: Theory and experiment.

Quantifying the tool–tissue interaction forces in surgery can be used in the training process of novice surgeons to help them better handle surgical tools and avoid exerting excessive forces. A significant challenge concerns the development of proper statistical learning techniques to model the relationship between the true force exerted on the tissue and several outputs read from sensors mounted on the surgical tools. We propose a nonparametric bootstrap technique and a Bayesian multilevel modeling methodology to estimate the true forces. We use the linear exponential loss function to asymmetrically penalize the over and underestimation of the applied forces to the tissue. We incorporate the direction of the force as a group factor in our analysis. A weighted approach is used to account for the nonhomogeneity of read voltages from the surgical tool. Our proposed Bayesian multilevel models provide estimates that are more accurate than those under the maximum likelihood and restricted maximum likelihood approaches. Moreover, confidence bounds are much narrower and the biases and root mean squared errors are significantly smaller in our multilevel models with the linear exponential loss function.

A multi-level spatio-temporal analysis on prostate cancer outcomes

BackgroundGeographic and temporal variation in positive surgical margins (PSM) for prostate cancer after radical prostatectomy (RP) has been observed. However, it is unclear how much of this variation could be attributed to patient, surgeon, institution, or socioeconomic-related factors and the impact of PSM on death among localized prostate cancer patients. MethodsThis study aimed to assess the independent and relative contribution of the patient, surgeon, institution and area-level risk factors on geographic and temporal variation of PSM and evaluate the impact of PSM on five-year all-cause and prostate cancer-specific mortality among localized prostate cancer patients. Within the hierarchical-related regression approach, we utilised Bayesian spatial-temporal multi-level models to study individual and area-level predictors with the outcomes, while accounting for geographically structured and unstructured correlation and non-linear trends. ResultsIndividual-level data included 10,075 localized prostate cancer cases with RP reported to the Prostate Cancer Outcomes Registry Victoria between 2009 and 2018. Area-level data comprised socio-economic disadvantage and remoteness data at the local government area level in Victoria, Australia. 26 % of patients had PSM, and the rates varied across areas by years. This variation was mainly associated with NCCN risk, followed by RP techniques, surgical institution type, surgeon volume and socio-economic disadvantage. Intermediate (Odds ratio/OR = 1.21,95 % credible interval/Crl = 1.05−1.41), high/very-high risk groups (OR = 2.24,95 % Crl = 1.91−2.64) and public surgical institution (OR = 1.64, 95 % Crl = 1.46−1.84) were independently associated with a higher likelihood of PSM. Robot-assisted (OR = 0.61, 95 % Crl = 0.55−0.68), laparoscopic RP (OR = 0.76, 95 % Crl = 0.62−0.93), high-volume surgeon (OR = 0.84, 95 % Crl = 0.76−0.93) and socio-economically least disadvantaged status (OR = 0.78, 95 % Crl = 0.64−0.94) showed a lower likelihood of PSM. PSM was also independently associated with a higher five-year all-cause and prostate cancer-specific mortality. ConclusionAggressive tumour characteristics and RP techniques were the main contributors to the likelihood of PSM following RP. Reducing the prevalence of PSM will generally improve prostate cancer-specific and all-cause mortality.

Decrease in prevalence but increase in frequency of non-marijuana drug use following the onset of the COVID-19 pandemic in a large cohort of young men who have sex with men and young transgender women

BackgroundSubstantial concern exists regarding the impact of the COVID-19 pandemic on substance use behaviors. This is especially true for subpopulations like young men who have sex with men and young transgender women (YMSM-YTW) who report higher rates of substance use. This study examines changes in prevalence and frequency of marijuana and non-marijuana drug use among YMSM-YTW following the onset of the pandemic. MethodData for this analysis (n = 458 participants, 1356 observations) come from an ongoing longitudinal cohort study of YMSM-YTW. A series of Bayesian multilevel models were used to examine change in prevalence and frequency of use for marijuana and non-marijuana drugs. ResultsResults indicated no systematic change in prevalence or frequency of marijuana use. However, a decrease in non-marijuana drug use was observed (OR = 0.60, 95 % CrI: [0.37, 0.94]) following the onset of the pandemic. Furthermore, a small increase in the frequency of non-marijuana drug use was observed (OR = 1.79, 95 % CrI: [1.02, 3.21]) among individuals who used these substances. ConclusionsThese findings concur with a small number of studies identifying a decrease in drug use prevalence but increase in frequency among those who continue to use drugs. Despite the protective effect of lower drug use prevalence, higher frequency of use may lead to additional negative health outcomes of drug use, particularly among groups facing multiple health challenges such as YMSM-YTW. However, the pandemic likely has a unique impact on substance use behaviors across subpopulations.

The effect of reinforcement, fibre, and matrix on the forming behaviour of continuous fibre reinforced thermoplastic composites

Forming simulations are a cost-effective solution to mitigate process-induced defects. The models developed to simulate the forming process require material property data for the dominant deformation mechanisms: intra-ply shear, bending, and inter-ply friction. These mechanisms are considered independent, and material property data has to be derived from experimental data for each mechanism separately. However, it is known that the material response to the deformation mechanisms is correlated, as the choice of matrix, fibre, and reinforcement influences the response to all mechanisms. Over the past years a large variety of thermoplastic composites have been characterised, covering a broad field of applications in automotive and aerospace industry. This makes it possible to start correlating the forming behaviour of thermoplastic composites. In this study, the effect of the constituents of a composite on the forming behaviour is analysed. To this end, a Bayesian cross-classified multilevel model with varying intercepts was applied, and the effects found by the model were analysed. Correlations were found between the effect of the constituents and their properties. The study confirms that the matrix material is an important indicator for the forming behaviour.

Size segregation and seasonal patterns in rusty crayfish Faxonius rusticus distribution and abundance on northern Lake Michigan spawning reefs

Non-native rusty crayfish are abundant egg predators on spawning reef habitats for lake trout and coregonines in northern Lake Michigan. To better understand rusty crayfish life-history on these unique habitats, we conducted monitoring in 2012 and 2013 at four locations previously identified as spawning areas for native fish. With the aid of a graphical causal model, we conducted an exploratory statistical analysis using a Bayesian multilevel modeling approach with model selection based on information criteria to identify important environmental variables for predicting rusty crayfish distribution and abundance on spawning reefs. We also compared seasonal trends in relative abundance, inferred from catch-per-unit-effort calculations from trapping, to previously reported accounts from a smaller inland lake. The results from our modeling provide evidence of size-class segregation across subtle changes in habitat characteristics of spawning reefs. Specifically, we found evidence that the distribution of >30 mm rusty crayfish was only weakly related to rock density (#/m2) relative to juveniles and smaller size classes. We also observed highest relative abundances from minnow trap monitoring in mid-October when water temperatures averaged 13.9 °C, which is later in the year and at cooler temperatures than similar monitoring from smaller inland lakes has reported. We hypothesize that unique environmental conditions elicit novel life-history responses from rusty crayfish on Lake Michigan spawning reefs and discuss our findings in the context of native fish restoration in the Laurentian Great Lakes.

Bayesian multilevel single case models using ‘Stan’. A new tool to study single cases in neuropsychology

Single case studies continue to play an important role in neuropsychological research. However, the range of statistical tools specifically designed for single cases is still limited. The current gold standard is the Crawford's t-test, but it is crucial to note that this is limited to simple designs and it is not possible to make inferences relevant to support for the null hypothesis with it. The Bayesian Multilevel Single Case models (BMSC) provide a novel tool that grants the flexibility of linear mixed model designs. BMSC is also able to support both null and alternative hypotheses in complex experimental designs using the Bayesian framework. We compared the BMSC and Crawford's t-test in a simulation study involving a case of no-dissociation and a case of simple dissociation between a single case patient and a series of control groups of different sizes (N = 5, 15, or 30). We then showed how BMSC is useful in complex designs by means of an example using real data. The BMSC proved to be more reliable than the Crawford's test, in terms of first-type errors and more precise estimating the parameters. Notably, the BMSC model provides a comprehensive vision of the whole experimental design, interpolating a single model. It follows the recent trend which involves a shift in attention from p-values to other inferential indices and estimates.

The trends and associated adverse maternal and perinatal outcomes of labour neuraxial analgesia among vaginal deliveries in China between 2012 and 2019: a real-world observational evidence

BackgroundThere is a lack of national report of the labour neuraxial analgesia (NA) rates in China in recent years, especially after the national promotion policy. The adverse maternal and perinatal outcomes associated with NA in China are also unknown. The aim of this study is to estimate the trends of NA rates from 2012 to 2019, to evaluate the effect of national policy on promoting NA and to identify the association between NA and adverse outcomes in China.MethodsWe used the individual data from China’s National Maternal Near Miss Surveillance System (NMNMSS) between 2012 and 2019, covering 438 hospitals from 326 urban districts or rural counties in 30 provinces across China. The analysis was restricted to singleton pregnant women who underwent vaginal delivery at or after 28 completed weeks of gestation. We estimate the trends of NA rates between 2012 and 2019, both at the national and provincial levels using Bayesian multilevel model. We also estimated the effect of the national pilot policy launched in 2018 using interrupted time-series analysis and identified the association between NA and adverse outcomes using modified Poisson regression combined with propensity score analysis.ResultsOver the study period, 620,851 of 6,023,046 women underwent vaginal delivery with NA. The estimated national NA rates increased from 8.4% in 2012 to 16.7% in 2019. Most provinces experienced the same rapid rise during this period. The national pilot policy accelerated the rise of the rates. No differences were observed between women with NA and without any analgesia in the incidence of uterine atony, placental retention, intrapartum stillbirths and 1- and 5-min Apgar scores lower than 7. However, women with NA had higher incidences of genital tract trauma (adjusted relative risk (aRR) 1.53, 95% confidence interval (CI) 1.04–2.26) and maternal near miss (aRR 1.35, 95% CI 1.08–1.69), only in hospitals which were not covered by the national pilot policy and usually lack of sufficient equipment and personnel.ConclusionsThe national policy can effectively increase the NA rate. However, as genital tract trauma and maternal near miss may increase in low-resource hospitals, but not in high-resource hospitals, further study is required to identify the reasons.

Metabolome-Wide, Phylogenetically Controlled Comparison Indicates Higher Phenolic Diversity in Tropical Tree Species.

Tropical plants are expected to have a higher variety of defensive traits, such as a more diverse array of secondary metabolic compounds in response to greater pressures of antagonistic interactions, than their temperate counterparts. We test this hypothesis using advanced metabolomics linked to a novel stoichiometric compound classification to analyze the complete foliar metabolomes of four tropical and four temperate tree species, which were selected so that each subset contained the same amount of phylogenetic diversity and evenness. We then built Bayesian phylogenetic multilevel models to test for tropical–temperate differences in metabolite diversity for the entire metabolome and for four major families of secondary compounds. We found strong evidence supporting that the leaves of tropical tree species have a higher phenolic diversity. The functionally closer group of polyphenolics also showed moderate evidence of higher diversity in tropical species, but there were no differences either for the entire metabolome or for the other major families of compounds analyzed. This supports the interpretation that this tropical–temperate contrast must be related to the functional role of phenolics and polyphenolics.

Detecting variability: A study on the application of bayesian multilevel modelling to archaeological data. Evidence from the Neolithic Adriatic and the Bronze Age Aegean

The detection and interpretation of variability in archaeological data has been a long-standing effort in the field. This paper aims to introduce the application of Bayesian multilevel modelling as a tool for the detection of variability at levels within nested archaeological data. Model structure, ways of construction, and the potential of using variability information to enhance archaeological interpretations is presented. This is demonstrated through the analysis of two case study datasets: Neolithic pottery finds from Mala (Nova) Pećina cave excavations in Croatia and stone finds from the Bronze Age site of Akrotiri, Thera, Greece. This is followed by a discussion of the multilevel model results and the possible interpretations that can be derived from them. Finally, propositions are made on how these and other models can be extended.

Risk of Workforce Exit due to Disability: State Differences in 2003–2016

Abstract A better understanding of trends in workforce exit due to disability and how these trends vary across states and subgroups can help federal and state policymakers identify both individual-level and state-level factors associated with increased risk of workforce exit due to disability. Using national survey data and Bayesian multilevel modeling techniques, we produce yearly estimates of trends in the risk of workforce exit due to disability for states and demographic subgroups. These estimates are more stable and have narrower uncertainty intervals than estimates produced using classical statistical methods. We identify Current Population Survey respondents as being “newly at-risk” of exiting the workforce due to disability if they reported being employed in one month and out of the labor force because of a disability in the next month, and we refer to their share of the working-age population as the “at-risk rate.” We find that age, education, race, and gender are important factors for the at-risk rate, in decreasing order. Holding demographics constant across states and time reduces the cross-state variation in the at-risk rate but does little to reduce variability over time. State at-risk rates are typically higher than application rates for the Social Security Administration’s disability programs, but the relationship between these rates varies considerably by state. Our preliminary exploration of the reasons for cross-state variation in this relationship suggests that differences across states may be due to differences in their industrial composition, job opportunities, and safety net structure.

90552 Evidence synthesis with reconstructed survival data

ABSTRACT IMPACT: A one-stage Bayesian multilevel model for meta-analysis integrating different survival data is introduced to complete the information synthesis without assuming proportional hazard. OBJECTIVES/GOALS: To develop a general modeling approach to perform efficient and robust meta-analyses using aggregated data (AD) for survival type endpoint and apply to a meta-analysis to examine the association between measurable residual disease (MRD) and disease-free survival (DFS) and overall survival (OS) in patients with acute myeloid leukemia (AML). METHODS/STUDY POPULATION: A Bayesian semi-parametric hierarchical model with a time-varying HR effect was presented. Three common types of survival information, including reconstructed survival data, the hazard ratio (HR) estimates with corresponding CIs and survival rates at specific time points, are synthesized such that all literature from the systematic review can contribute properly to the estimation and uncertainty quantification of the model parameters. The time-varying effects was modeled by assuming piecewise hazard risk and piecewise constant hazard ratio. The heterogeneity across studies was expressed by an additive random study effect and a random treatment-by-study interaction. The method was applied to a systematic review of 81 publications reporting on 11,151 AML patients. RESULTS/ANTICIPATED RESULTS: In simulation studies that the proportional hazard assumption is either valid or violated, the proposed method was efficient to achieve comparable performance to IPD meta-analysis, a gold standard approach, in estimating the survival rates, the restricted mean survival time at specific time points and median survival time with the point estimates close to the true values. When HR is not proportional over time, the proposed method was robust in estimating HR and significantly outperformed the classical random-effects meta-analysis. In the application to AML study, the average HR for achieving MRD negativity was 0.36 (95% CrI, 0.33-0.39) for OS and 0.37 (95% CrI, 0.34-0.40) for DFS. The association of MRD negativity with OS and DFS was significant and consistent across all subgroups. DISCUSSION/SIGNIFICANCE OF FINDINGS: The proposed novel method provided a flexible framework for meta-analysis of survival data, to accommodate various types of survival data in one model without assuming proportional hazards assumption. The findings of AML meta-analysis suggest that achievement of MRD negativity is associated with superior DFS and OS in patients with AML.

Projecting the course of COVID-19 in Turkey: A probabilistic modeling approach

Background/aimThe COVID-19 pandemic originated in Wuhan, China, in December 2019 and became one of the worst global health crises ever. While struggling with the unknown nature of this novel coronavirus, many researchers and groups attempted to project the progress of the pandemic using empirical or mechanistic models, each one having its drawbacks. The first confirmed cases were announced early in March, and since then, serious containment measures have taken place in Turkey.Materials and methodsHere, we present a different approach, a Bayesian negative binomial multilevel model with mixed effects, for the projection of the COVID-19 pandemic and we apply this model to the Turkish case. The model source code is available at https://github.com/kansil/covid-19. We predicted the confirmed daily cases and cumulative numbers from June 6th to June 26th with 80%, 95%, and 99% prediction intervals (PI).ResultsOur projections showed that if we continued to comply with the measures and no drastic changes were seen in diagnosis or management protocols, the epidemic curve would tend to decrease in this time interval. Also, the predictive validity analysis suggests that the proposed model projections should have a PI around 95% for the first 12 days of the projections.ConclusionWe expect that drastic changes in the course of COVID-19 in Turkey will cause the model to suffer in predictive validity, and this can be used to monitor the epidemic. We hope that the discussion on these projections and the limitations of the epidemiological forecasting will be beneficial to the medical community, and policy makers.

Observational study of agreement between attending and trainee physicians on the surprise question: "Would you be surprised if this patient died in the next 12 months?"

Optimal end-of-life care requires identifying patients that are near the end of life. The extent to which attending physicians and trainee physicians agree on the prognoses of their patients is unknown. We investigated agreement between attending and trainee physician on the surprise question: "Would you be surprised if this patient died in the next 12 months?", a question intended to assess mortality risk and unmet palliative care needs. This was a multicentre prospective cohort study of general internal medicine patients at 7 tertiary academic hospitals in Ontario, Canada. General internal medicine attending and senior trainee physician dyads were asked the surprise question for each of the patients for whom they were responsible. Surprise question response agreement was quantified by Cohen's kappa using Bayesian multilevel modeling to account for clustering by physician dyad. Mortality was recorded at 12 months. Surprise question responses encompassed 546 patients from 30 attending-trainee physician dyads on academic general internal medicine teams at 7 tertiary academic hospitals in Ontario, Canada. Patients had median age 75 years (IQR 60-85), 260 (48%) were female, and 138 (25%) were dependent for some or all activities of daily living. Trainee and attending physician responses agreed in 406 (75%) patients with adjusted Cohen's kappa of 0.54 (95% credible interval 0.41 to 0.66). Vital status was confirmed for 417 (76%) patients of whom 160 (38% of 417) had died. Using a response of "No" to predict 12-month mortality had positive likelihood ratios of 1.84 (95% CrI 1.55 to 2.22, trainee physicians) and 1.51 (95% CrI 1.30 to 1.72, attending physicians), and negative likelihood ratios of 0.31 (95% CrI 0.17 to 0.48, trainee physicians) and 0.25 (95% CrI 0.10 to 0.46, attending physicians). Trainee and attending physician responses to the surprise question agreed in 54% of cases after correcting for chance agreement. Physicians had similar discriminative accuracy; both groups had better accuracy predicting which patients would survive as opposed to which patients would die. Different opinions of a patient's prognosis may contribute to confusion for patients and missed opportunities for engagement with palliative care services.

Phenological mismatch, carryover effects, and marine survival in a wild steelhead trout Oncorhynchus mykiss population

Climate-driven changes in the oceans, such as shifts in prey timing and abundance, could influence variability in population productivity of marine fishes. For example, according to the match/mismatch hypothesis, the temporal matching of the young salmon outmigration from freshwater to the ocean relative to the timing of availability of their prey could influence their marine survival. Indeed, understanding patterns and processes of marine survival is particularly pressing in many salmon and steelhead trout populations due to recent declines. To determine whether phenological mismatches between juvenile salmonids and their prey could contribute to low ocean survival, we analyzed the migration timing and ocean survival of 22,116 tagged juvenile steelhead trout Oncorhynchus mykiss over 12 years from the Wind River, Washington State, USA. We used a Bayesian multilevel modelling approach with variable selection to assess how survival was associated with body size, river exit date, the biological spring transition date (the day when northern zooplankton first appeared in the coastal region near the Columbia River estuary), and the degree of mismatch (the effect of the interaction between individual outmigration timing and biological spring transition date). The variables with the highest probability of contributing to individual survival were fish size (100%), river exit date (99%), the interaction between year and river exit date (91%), and the biological spring transition date (64%). Fish that were larger than average at outmigration had higher ocean survival, providing further evidence that freshwater growing conditions have carryover effects on marine survival. Years with greater annual phenological mismatches such as those years with late biological spring transition dates (i.e., occurring after June 1st), or warm sea surface temperatures, had sufficiently low marine survival to compromise recovery goals. Substantial intra-annual variation in outmigration timing buffered the population from inter-annual variation in optimal outmigration timing. Collectively these findings indicate that freshwater growing conditions, migration timing, and the timing of high-quality food availability in the nearshore coastal environment work in concert to influence individual survival and annual smolt-to-adult returns.

Bayesian multilevel model on maternal mortality in Ethiopia

Maternal mortality is one of the socio-economic problems and widely considered a serious indicator of the quality of a health. Ethiopia is considered to be one of the top six sub-Saharan countries with severe maternal mortality. The objective of this study was to investigate the effects of the Demographic and Socio-economic determinant factors of maternal mortality in Ethiopia. Data from the 2016 Ethiopia Demographic and Health Survey indicated that the sample of women (15–49) was (n = 10,103). The Bayesian multilevel we were used to explore the major risk factors and regional variations in maternal mortality in Ethiopia. Markov chain Monte Carlo methods with non-informative priors have been applied. The Deviance Information Criterion model selection criteria were used to select the appropriate model. The analysis result, 145 (1.43%) mothers were died due to pregnancy. Using model selection criteria Bayesian multilevel random coefficient was found to be appropriate. With this model, Age of mother, marital status, number of living children, wealth index and Education are found to be the significant determinants of maternal mortality in Ethiopia. The study indicated that there was within and between regional variations in maternal mortality. Inference is the fully Bayesian multilevel model based on recent Markov chain Monte Carlo techniques. The socioeconomic, demographic and environmental determinants included in the study were found to be statistically significant. The result of the Bayesian multilevel model in this study has shown that educational attainment, wealth index, an age of mother, status and number of living children was a significant factor of maternal mortality.

Nonplacebo Controls to Determine the Magnitude of Ergogenic Interventions: A Systematic Review and Meta-analysis.

Placebos are used as a control treatment that is meant to be indistinguishable from the active intervention. However, where substantive placebo effects may occur, studies that do not include a nonplacebo control arm may underestimate the overall effect of the intervention (active plus placebo components). This study aimed to determine the relative magnitude of the placebo effect associated with nutritional supplements (caffeine and extracellular buffers) by meta-analyzing data from studies containing both placebo and nonplacebo control sessions. Bayesian multilevel meta-analysis models were used to estimate pooled effects and express the placebo effect as a percentage of the overall intervention effect. Thirty-four studies were included, with the median pooled effect size (ES0.5) indicating a very small (ES0.5 = 0.09 (95% credible interval (CrI), 0.01-0.17)) improvement in the performance of placebo compared with control. There was no moderating effect of exercise type (capacity or performance), exercise duration, or training status. The comparison between active intervention and control indicated a small to medium effect (ES0.5 = 0.37 (95% CrI, 0.20-0.56)). Expressed in relative terms, the placebo effect was equivalent to 25% (75% CrI, 16%-35%) and 59% (75% CrI, 34%-94%) of the total intervention effect for buffers and caffeine. These results demonstrate a very small but potentially important placebo effect with nutritional supplementation studies. A substantive proportion of supplement effects may be due to placebo effects, with the relative proportion influenced by the magnitude of the overall ergogenic effect. Where feasible, intervention studies should use nonplacebo control-arm comparators to identify the proportion of the effect estimated to come from placebo effects and avoid underestimating the overall benefits that the physiological plus psychobiological aspects associated with an intervention provide in the real world.

Estimating the statistical power to detect set-size effects in contralateral delay activity.

The contralateral delay activity (CDA) is an event-related potential component commonly used to examine the online processes of visual working memory. Here, we provide a robust analysis of the statistical power that is needed to achieve reliable and reproducible results with the CDA. Using two very large EEG datasets that examined the contrast between CDA amplitude with set sizes 2 and 6 items and set sizes 2 and 4 items, we present a subsampling analysis that estimates the statistical power achieved with varying numbers of subjects and trials based on the proportion of significant tests in 10,000 iterations. We also generated simulated data using Bayesian multilevel modeling to estimate power beyond the bounds of the original datasets. The number of trials and subjects required depends critically on the effect size. Detecting the presence of the CDA-a reliable difference between contralateral and ipsilateral electrodes during the memory period-required only 30-50 clean trials with a sample of 25 subjects to achieve approximately 80% statistical power. However, for detecting a difference in CDA amplitude between two set sizes, a substantially larger number of trials and subjects were required; approximately 400 clean trials with 25 subjects to achieve 80% power. Thus, to achieve robust tests of how CDA activity differs across conditions, it is essential to be mindful of the estimated effect size. We recommend researchers designing experiments to detect set-size differences in the CDA collect substantially more trials per subject.