MultiBUGS: A Parallel Implementation of the BUGS Modelling Framework for Faster Bayesian Inference

Mediation Analysis of Racial Disparity for Infant Mortality Using Bayesian Estimation of Potential Outcomes

Background/Objectives: While the overall rate of infant mortality in the United States has been decreasing over decades, the racial disparity, defined as the difference between races, has increased. Even though a person's race cannot change, it may be possible to identify factors that mediate or cause this racial disparity. Evaluating the factors that mediate or cause racial disparity is imperative because current clinical recommendations could be based on preventative modalities that are more effective for white women and their children. Methods: A Bayesian approach modeled the data from the full United States National Natality Database for the years 2016 to 2018. The binomial rate parameters for each combination of race and mediators provided the potential outcomes. Estimating the mediation outcomes, including total effect, controlled direct effect, mediated effect, and proportion mediated used common counterfactual definitions for these probabilities. Results: Maternal smoking, low birthweight, and teenage maternity interacted in causing racial disparity for infant mortality. The proportion of racial disparity attributable to low birthweight was approximately 0.73, with only small variations attributable to maternal smoking and teenage maternity. Conclusions: The novel approach facilitated modeling of multiple mediators. Low birthweight caused racial disparity for infant mortality. The model can be extended to evaluate additional mediational factors with the objective of identifying the preventable causes.

JAGS model specification for spatiotemporal epidemiological modelling

Bayesian inference in modelling infectious diseases using Bayesian inference using Gibbs Sampling (BUGS) is notable in the last two decades in parallel with the advancements in computing and model development. The ability of BUGS to easily implement the Markov chain Monte Carlo (MCMC) method brought Bayesian analysis to the mainstream of infectious disease modelling. However, with the existing software that runs MCMC to make Bayesian inferences, it is challenging, especially in terms of computational complexity, when infectious disease models become more complex with spatial and temporal components, in addition to the increasing number of parameters and large datasets. This study investigates two alternative subscripting strategies for creating models in Just Another Gibbs Sampler (JAGS) environment and their performance in terms of run times. Our results are useful for practitioners to ensure the efficiency and timely implementation of Bayesian spatiotemporal infectious disease modelling.

Bayesian Spatio-Temporal Multilevel Modelling of Patient-Reported Quality of Life following Prostate Cancer Surgery

BACKGROUND

Globally, prostate cancer is the second leading cause of cancer deaths among males. It is the most commonly diagnosed cancer in Australia. The quality of life of prostate cancer patients is poorer when compared to the general population due to the disease itself and its related complications. However, there is limited research on the geographic pattern of quality of life and its risk factors in Victoria. Therefore, an examination of the spatio-temporal pattern and risk factors of poor quality of life, along with the impact of spatial weight matrices on estimates and model performance, was conducted.

METHOD

A retrospective study was undertaken based on the Prostate Cancer Outcome Registry-Victoria data. Patient data (n = 5238) were extracted from the Prostate Cancer Outcome Registry, a population-based clinical quality outcome assessment from 2015 to 2021. A Bayesian spatio-temporal multilevel model was fitted to identify risk factors for poor quality of life. This study also evaluated the impact of distance- and adjacency-based spatial weight matrices. Model convergence was assessed using Gelman-Rubin statistical plots, and model comparison was based on the Watanabe-Akaike Information Criterion.

RESULTS

A total of 1906 (36.38%) prostate cancer patients who had undergone surgery experienced poor quality of life in our study. Belonging to the age group between 76 and 85 years (adjusted odds ratio (AOR) = 2.90, 95% credible interval (CrI): 1.39, 2.08), having a prostate-specific antigen level between 10.1 and 20.0 (AOR = 1.33, 95% CrI: 1.12, 1.58), and being treated in a public hospital (AOR = 1.35, 95% CrI: 1.17, 1.53) were significantly associated with higher odds of poor quality of life. Conversely, residing in highly accessible areas (AOR = 0.60, 95% CrI: 0.38, 0.94) was significantly associated with lower odds of poor prostate-specific antigen levels. Variations in estimates and model performance were observed depending on the choice of spatial weight matrices.

CONCLUSION

Belonging to an older age group, having a high prostate-specific antigen level, receiving treatment in public hospitals, and remoteness were statistically significant factors linked to poor quality of life. Substantial spatio-temporal variations in poor quality of life were observed in Victoria across local government areas. The distance-based weight matrix performed better than the adjacency-based matrix. This research finding highlights the need to reduce geographical disparities in quality of life. The statistical methods developed in this study may also be useful to apply to other population-based clinical registry settings.

Effectiveness of Fludrocortisone Plus Hydrocortisone versus Hydrocortisone Alone in Septic Shock: A Systematic Review and Network Meta-Analysis of Randomized Controlled Trials

Rationale: The use of hydrocortisone in adult patients with septic shock is controversial, and the effectiveness of adding fludrocortisone to hydrocortisone remains uncertain. Objectives: To assess the comparative effectiveness and safety of fludrocortisone plus hydrocortisone, hydrocortisone alone, and placebo or usual care in adults with septic shock. Methods: A systematic review and a Bayesian network meta-analysis of peer-reviewed randomized trials were conducted. The primary outcome was all-cause mortality at last follow-up. Treatment effects are presented as relative risks (RRs) with 95% credible intervals (CrIs). Placebo or usual care was the reference treatment. Measurements and Main Results: Among 7,553 references, we included 17 trials (7,688 patients). All-cause mortality at last follow-up was lowest with fludrocortisone plus hydrocortisone (RR, 0.85; 95% CrI, 0.72-0.99; 98.3% probability of superiority, moderate-certainty evidence), followed by hydrocortisone alone (RR, 0.97; 95% CrI, 0.87-1.07; 73.1% probability of superiority, low-certainty evidence). The comparison of fludrocortisone plus hydrocortisone versus hydrocortisone alone was based primarily on indirect evidence (only two trials with direct evidence). Fludrocortisone plus hydrocortisone was associated with a 12% lower risk of all-cause mortality compared with hydrocortisone alone (RR, 0.88; 95% CrI, 0.74-1.03; 94.2% probability of superiority, moderate-certainty evidence). Conclusions: In adult patients with septic shock, fludrocortisone plus hydrocortisone was associated with lower risk of all-cause mortality at last follow-up than placebo and hydrocortisone alone. The scarcity of head-to-head trials comparing fludrocortisone plus hydrocortisone versus hydrocortisone alone led our network meta-analysis to rely primarily on indirect evidence for this comparison. Although we undertook several sensitivity analyses and assessments, these findings should be considered while also acknowledging the heterogeneity of included trials.

Assessing the Negative Binomial-Lindley model for crash hotspot identification: Insights from Monte Carlo simulation analysis

Identifying hazardous crash sites (or hotspots) is a crucial step in highway safety management. The Negative Binomial (NB) model is the most common model used in safety analyses and evaluations - including hotspot identification. The NB model, however, is not without limitations. In fact, this model does not perform well when data are highly dispersed, include excess zero observations, or have a long tail. Recently, the Negative Binomial-Lindley (NB-L) model has been proposed as an alternative to the NB. The NB-L model overcomes several limitations related to the NB, such as addressing the issue of excess zero observations in highly dispersed data. However, it is not clear how the NB-L model performs regarding the hotspot identification. In this paper, an innovative Monte Carlo simulation protocol was designed to generate a wide range of simulated data characterized by different means, dispersions, and percentage of zeros. Next, the NB-L model was written as a Full-Bayes hierarchical model and compared with the Full-Bayes NB model for hotspot identification using extensive simulation scenarios. Most previous studies focused on statistical fit, and showed that the NB-L model fits the data better than the NB. In this research, however, we investigated the performance of the NB-L model in identifying the hazardous sites. We showed that there is a trade-off between the NB-L and NB when it comes to hotspot identification. Multiple performance metrics were used for the assessment. Among those, the results show that the NB-L model provides a better specificity in identifying hotspots, while the NB model provides a better sensitivity, especially for highly dispersed data. In other words, while the NB model performs better in identifying hazardous sites, the NB-L model performs better, when budget is limited, by not selecting non-hazardous sites as hazardous.

Human ratings take time: A hierarchical facets model for the joint analysis of ratings and rating times

Performance assessments increasingly utilize onscreen or internet-based technology to collect human ratings. One of the benefits of onscreen ratings is the automatic recording of rating times along with the ratings. Considering rating times as an additional data source can provide a more detailed picture of the rating process and improve the psychometric quality of the assessment outcomes. However, currently available models for analyzing performance assessments do not incorporate rating times. The present research aims to fill this gap and advance a joint modeling approach, the "hierarchical facets model for ratings and rating times" (HFM-RT). The model includes two examinee parameters (ability and time intensity) and three rater parameters (severity, centrality, and speed). The HFM-RT successfully recovered examinee and rater parameters in a simulation study and yielded superior reliability indices. A real-data analysis of English essay ratings collected in a high-stakes assessment context revealed that raters exhibited considerably different speed measures, spent more time on high-quality than low-quality essays, and tended to rate essays faster with increasing severity. However, due to the significant heterogeneity of examinees' writing proficiency, the improvement in the assessment's reliability using the HFM-RT was not salient in the real-data example. This discussion focuses on the advantages of accounting for rating times as a source of information in rating quality studies and highlights perspectives from the HFM-RT for future research on rater cognition.

Incidence of typhoid fever in Burkina Faso, Democratic Republic of the Congo, Ethiopia, Ghana, Madagascar, and Nigeria (the Severe Typhoid in Africa programme): a population-based study

BACKGROUND

Typhoid Fever remains a major cause of morbidity and mortality in low-income settings. The Severe Typhoid in Africa programme was designed to address regional gaps in typhoid burden data and identify populations eligible for interventions using novel typhoid conjugate vaccines.

METHODS

A hybrid design, hospital-based prospective surveillance with population-based health-care utilisation surveys, was implemented in six countries in sub-Saharan Africa. Patients presenting with fever (≥37·5°C axillary or ≥38·0°C tympanic) or reporting fever for three consecutive days within the previous 7 days were invited to participate. Typhoid fever was ascertained by culture of blood collected upon enrolment. Disease incidence at the population level was estimated using a Bayesian mixture model.

FINDINGS

27 866 (33·8%) of 82 491 participants who met inclusion criteria were recruited. Blood cultures were performed for 27 544 (98·8%) of enrolled participants. Clinically significant organisms were detected in 2136 (7·7%) of these cultures, and 346 (16·2%) Salmonella enterica serovar Typhi were isolated. The overall adjusted incidence per 100 000 person-years of observation was highest in Kavuaya and Nkandu 1, Democratic Republic of the Congo (315, 95% credible interval 254-390). Overall, 46 (16·4%) of 280 tested isolates showed ciprofloxacin non-susceptibility.

INTERPRETATION

High disease incidence (ie, >100 per 100 000 person-years of observation) recorded in four countries, the prevalence of typhoid hospitalisations and complicated disease, and the threat of resistant typhoid strains strengthen the need for rapid dispatch and implementation of effective typhoid conjugate vaccines along with measures designed to improve clean water, sanitation, and hygiene practices.

FUNDING

The Bill & Melinda Gates Foundation.

Global variation in diabetes diagnosis and prevalence based on fasting glucose and hemoglobin A1c

Fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) are both used to diagnose diabetes, but these measurements can identify different people as having diabetes. We used data from 117 population-based studies and quantified, in different world regions, the prevalence of diagnosed diabetes, and whether those who were previously undiagnosed and detected as having diabetes in survey screening, had elevated FPG, HbA1c or both. We developed prediction equations for estimating the probability that a person without previously diagnosed diabetes, and at a specific level of FPG, had elevated HbA1c, and vice versa. The age-standardized proportion of diabetes that was previously undiagnosed and detected in survey screening ranged from 30% in the high-income western region to 66% in south Asia. Among those with screen-detected diabetes with either test, the age-standardized proportion who had elevated levels of both FPG and HbA1c was 29-39% across regions; the remainder had discordant elevation of FPG or HbA1c. In most low- and middle-income regions, isolated elevated HbA1c was more common than isolated elevated FPG. In these regions, the use of FPG alone may delay diabetes diagnosis and underestimate diabetes prevalence. Our prediction equations help allocate finite resources for measuring HbA1c to reduce the global shortfall in diabetes diagnosis and surveillance.

Use of second-generation antipsychotics in autism spectrum disorder: a systematic review and meta-analysis protocol

INTRODUCTION

Atypical antipsychotics have been studied to treat autism spectrum disorder (ASD). However, like little is known about whether these drugs are effective and safe when compared in controlled and non-controlled settings. This study aims to assess the efficacy and safety of second-generation antipsychotics in ASD in randomised controlled trials (RCT) and observational studies.

METHODS AND ANALYSIS

This systematic review will include RCT and prospective cohorts evaluating second-generation antipsychotics in people 5 years and older diagnosed with ASD. Searches will be conducted in Medline, Embase, Cochrane Library, Epistemonikos, Lilacs, CINAHL, PsycINFO, trial registries and grey literature databases without restriction on publication status, year of publication and language. The primary outcomes will be symptoms of aggressive behaviour, quality of life for the individual or their careers, and discontinuation or dropouts/withdrawals of antipsychotics due to adverse events. The secondary outcomes are other not serious adverse events and adherence to pharmacotherapy. Selection, data extraction, and quality assessment will be performed by pairs of reviewers, independently. The Risk of Bias 2 (RoB 2) and Risk of Bias in Non-Randomised Studies of Interventions (ROBINS-I) tools will be used to assess the risk of bias in the included studies. If appropriate, a meta-analysis and network meta-analysis will be conducted to synthesise the results. The overall quality of the evidence for each outcome will be determined by the Recommendation, Assessment, Development and Evaluation approach.

ETHICS AND DISSEMINATION

This study will systematically summarise the existing evidence evaluating the use of second-generation antipsychotics for treating ASD, in controlled and uncontrolled studies. The results of this review will be disseminated through peer-reviewed publications and conference presentations.

PROSPERO REGISTRATION NUMBER

CRD42022353795.

More future synergies and less trade-offs between forest ecosystem services with natural climate solutions instead of bioeconomy solutions

To reach the Paris Agreement, societies need to increase the global terrestrial carbon sink. There are many climate change mitigation solutions (CCMS) for forests, including increasing bioenergy, bioeconomy, and protection. Bioenergy and bioeconomy solutions use climate-smart, intensive management to generate high quantities of bioenergy and bioproducts. Protection of (semi-)natural forests is a major component of "natural climate solution" (NCS) since forests store carbon in standing biomass and soil. Furthermore, protected forests provide more habitat for biodiversity and non-wood ecosystem services (ES). We investigated the impacts of different CCMS and climate scenarios, jointly or in isolation, on future wood ES, non-wood ES, and regulating ES for a major wood provider for the international market. Specifically, we projected future ES given by three CCMS scenarios for Sweden 2020-2100. In the long term, fulfilling the increasing wood demand through bioenergy and bioeconomy solutions will decrease ES multifunctionality, but the increased stand age and wood stocks induced by rising greenhouse gas (GHG) concentrations will partially offset these negative effects. Adopting bioenergy and bioeconomy solutions will have a greater negative impact on ES supply than adopting NCS. Bioenergy or bioeconomy solutions, as well as increasing GHG emissions, will reduce synergies and increase trade-offs in ES. NCS, by contrast, increases the supply of multiple ES in synergy, even transforming current ES trade-offs into future synergies. Moreover, NCS can be considered an adaptation measure to offset negative climate change effects on the future supplies of non-wood ES. In boreal countries around the world, forestry strategies that integrate NCS more deeply are crucial to ensure a synergistic supply of multiple ES.

Scalable proximal methods for cause-specific hazard modeling with time-varying coefficients

Survival modeling with time-varying coefficients has proven useful in analyzing time-to-event data with one or more distinct failure types. When studying the cause-specific etiology of breast and prostate cancers using the large-scale data from the Surveillance, Epidemiology, and End Results (SEER) Program, we encountered two major challenges that existing methods for estimating time-varying coefficients cannot tackle. First, these methods, dependent on expanding the original data in a repeated measurement format, result in formidable time and memory consumption as the sample size escalates to over one million. In this case, even a well-configured workstation cannot accommodate their implementations. Second, when the large-scale data under analysis include binary predictors with near-zero variance (e.g., only 0.6% of patients in our SEER prostate cancer data had tumors regional to the lymph nodes), existing methods suffer from numerical instability due to ill-conditioned second-order information. The estimation accuracy deteriorates further with multiple competing risks. To address these issues, we propose a proximal Newton algorithm with a shared-memory parallelization scheme and tests of significance and nonproportionality for the time-varying effects. A simulation study shows that our scalable approach reduces the time and memory costs by orders of magnitude and enjoys improved estimation accuracy compared with alternative approaches. Applications to the SEER cancer data demonstrate the real-world performance of the proximal Newton algorithm.

BIMAM—a tool for imputing variables missing across datasets using a Bayesian imputation and analysis model

Motivation

Combination of multiple datasets is routine in modern epidemiology. However, studies may have measured different sets of variables; this is often inefficiently dealt with by excluding studies or dropping variables. Multilevel multiple imputation methods to impute these ‘systematically’ missing data (as opposed to ‘sporadically’ missing data within a study) are available, but problems may arise when many random effects are needed to allow for heterogeneity across studies. We show that the Bayesian IMputation and Analysis Model (BIMAM) implemented in our tool works well in this situation.

General features

BIMAM performs imputation and analysis simultaneously. It imputes both binary and continuous systematically and sporadically missing data, and analyses binary and continuous outcomes. BIMAM is a user-friendly, freely available tool that does not require knowledge of Bayesian methods. BIMAM is an R Shiny application. It is downloadable to a local machine and it automatically installs the required freely available packages (R packages, including R2MultiBUGS and MultiBUGS).

Availability

BIMAM is available at [www.alecstudy.org/bimam].

Forecasting of Lung Cancer Incident Cases at the Small-Area Level in Victoria, Australia

Predicting lung cancer cases at the small-area level is helpful to quantify the lung cancer burden for health planning purposes at the local geographic level. Using Victorian Cancer Registry (2001-2018) data, this study aims to forecast lung cancer counts at the local government area (LGA) level over the next ten years (2019-2028) in Victoria, Australia. We used the Age-Period-Cohort approach to estimate the annual age-specific incidence and utilised Bayesian spatio-temporal models that account for non-linear temporal trends and area-level risk factors. Compared to 2001, lung cancer incidence increased by 28.82% from 1353 to 1743 cases for men and 78.79% from 759 to 1357 cases for women in 2018. Lung cancer counts are expected to reach 2515 cases for men and 1909 cases for women in 2028, with a corresponding 44% and 41% increase. The majority of LGAs are projected to have an increasing trend for both men and women by 2028. Unexplained area-level spatial variation substantially reduced after adjusting for the elderly population in the model. Male and female lung cancer cases are projected to rise at the state level and in each LGA in the next ten years. Population growth and an ageing population largely contributed to this rise.

The Effects of Parent Ages on Birth Defects

BACKGROUND

Men and women, in the United States, are having children at considerably older ages. This changing demographic has been associated with multiple neonatal adverse birth outcomes that are currently considered to constitute a major public health crisis. The objective of this study was to evaluate the risk of parent age on birth defects and to provide results that can serve as a starting point for more specific mediation modeling. The modeling estimated the effects of parent age on birth defects controlling for confounding between maternal and paternal age and separated the mediating effect of chromosomal disorders, including Down syndrome.

METHODS

Data containing approximately 15,000,000 birth records were obtained from the United States Natality database for the years 2014 to 2018. A Bayesian modeling approach was used to estimate adjusted risks of parent ages both unadjusted and adjusted for the other parent's age and for the mediational effect of chromosomal disorders, including Down syndrome.

RESULTS

Increasing maternal age was associated with increased risks for hypospadias and cyanotic congenital heart disease. Increasing maternal and paternal ages were associated with decreasing risks for gastroschisis. For limb reduction defect, cleft lip and all defects combined, the risk of maternal age was U-shaped with the lowest risks observed at approximately age 35y. Paternal age was not associated with an increase in the birth prevalence of birth defects.

CONCLUSION

Advancing maternal age was associated with increased birth prevalence of hypospadias and cyanotic congenital heart disease and associated with a lower birth prevalence for gastroschisis. Both older and younger maternal ages were related to limb reduction defect and cleft lip. Advancing paternal age was not associated with an increased birth prevalence of birth defects but was associated with a decreased birth prevalence of gastroschisis.

Association between Receipt of Guideline-Concordant Lung Cancer Treatment and Individual- and Area-Level Factors: A Spatio-Temporal Analysis

BACKGROUND

Guideline-concordant treatment (GCT) of lung cancer has been observed to vary across geographic regions over the years. However, there is little evidence as to what extent this variation is explained by differences in patients' clinical characteristics versus contextual factors, including socioeconomic inequalities.

METHODS

This study evaluated the independent effects of individual- and area-level risk factors on geographic and temporal variation in receipt of GCT among patients with lung cancer. Receipt of GCT was defined on the basis of the National Comprehensive Cancer Network guidelines. We used Bayesian spatial-temporal multilevel models to combine individual and areal predictors and outcomes while accounting for geographically structured and unstructured correlation and linear and nonlinear trends.

RESULTS

Our study included 4,854 non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC) cases, reported to the Victorian Lung Cancer Registry between 2011 and 2018. Area-level data comprised socioeconomic disadvantage and remoteness data at the local government area level in Victoria, Australia. Around 60.36% of patients received GCT, and the rates varied across geographic areas over time. This variation was mainly associated with poor performance status, advanced clinical stages, NSCLC types, public hospital insurance, area-level deprivation, and comorbidities.

CONCLUSIONS

This study highlights the need to address disparities in receipt of GCT among patients with lung cancer with poor performance status, NSCLC, advanced clinical stage, stage I-III SCLC, stage III NSCLC, public hospital insurance, and comorbidities, and living in socioeconomically disadvantaged areas.

IMPACT

Two-year mortality outcomes significantly improved with GCT. Interventions aimed at reducing these inequalities could help to improve lung cancer outcomes.