Flexible Bayesian semiparametric mixed-effects model for skewed longitudinal data.

BACKGROUND: In clinical trials and epidemiological research, mixed-effects models are commonly used to examine population-level and subject-specific trajectories of biomarkers over time. Despite their increasing popularity and application, the specification of these models necessitates a great deal of care when analysing longitudinal data with non-linear patterns and asymmetry. Parametric (linear) mixed-effect models may not capture these complexities flexibly and adequately. Additionally, assuming a Gaussian distribution for random effects and/or model errors may be overly restrictive, as it lacks robustness against deviations from symmetry. METHODS: This paper presents a semiparametric mixed-effects model with flexible distributions for complex longitudinal data in the Bayesian paradigm. The non-linear time effect on the longitudinal response was modelled using a spline approach. The multivariate skew-t distribution, which is a more flexible distribution, is utilized to relax the normality assumptions associated with both random-effects and model errors. RESULTS: To assess the effectiveness of the proposed methods in various model settings, simulation studies were conducted. We then applied these models on chronic kidney disease (CKD) data and assessed the relationship between covariates and estimated glomerular filtration rate (eGFR). First, we compared the proposed semiparametric partially linear mixed-effect (SPPLM) model with the fully parametric one (FPLM), and the results indicated that the SPPLM model outperformed the FPLM model. We then further compared four different SPPLM models, each assuming different distributions for the random effects and model errors. The model with a skew-t distribution exhibited a superior fit to the CKD data compared to the Gaussian model. The findings from the application revealed that hypertension, diabetes, and follow-up time had a substantial association with kidney function, specifically leading to a decrease in GFR estimates. CONCLUSIONS: The application and simulation studies have demonstrated that our work has made a significant contribution towards a more robust and adaptable methodology for modeling intricate longitudinal data. We achieved this by proposing a semiparametric Bayesian modeling approach with a spline smoothing function and a skew-t distribution.

Spatial mapping of colorectal cancer screening uptake and associated factors.

OBJECTIVE: Over the past decades, it has been understood that the availability of screening tests has contributed to a steady decline in incidence of colorectal cancer (CRC). However, it is also seen that there is a geographic disparity in the use of such tests across small areas. The aim of this study is to examine small-area level barrier factors that may impact CRC screening uptake and to delineate coldspot (low uptake of screening) counties in Florida. METHODS: Data on the percentages of county-level CRC screening uptakes in 2016 and county-level barrier factors for screening were obtained from the Florida Department of Health, Division of Public Health Statistics & Performance Management. Bayesian spatial beta models were used to produce posterior probability of deceedance to identify coldspots for CRC screening rates. RESULTS: Unadjusted screening rates using sigmoidoscopy or colonoscopy test ranged from 56.8 to 85%. Bayesian spatial beta models were fitted to the proportion data. At an ecological level, we found that an increasing rate of CRC screening uptake for either of the test types (colon/rectum exam, stool-based test) was strongly associated with a higher health insurance coverage, and lower percentage of population that speak English less than very well (immigration) at county level. Eleven coldspot counties out of 67 total were also identified. CONCLUSION: This study suggests that health insurance disparities in the use of CRC screening tests are an important factor that may need more attention for resource allocation and health policy targeting small areas with low uptake of screening.

Colorectal Cancer Risk Perceptions Among Black Men in Florida.

PURPOSE: We examined colorectal cancer (CRC) risk perceptions among Black men in relation to socio-demographic characteristics, disease prevention factors, and personal/family history of CRC. METHODS: A self-administered cross-sectional survey was conducted in five major cities in Florida between April 2008 and October 2009. Descriptive statistics and multivariable logistic regression were performed. RESULTS: Among 331 eligible men, we found a higher proportion of CRC risk perceptions were exhibited among those aged ≥ 60 years (70.5%) and American nativity (59.1%). Multivariable analyses found men aged ≥ 60 had three times greater odds of having higher CRC risk perceptions compared to those ≤ 49 years (95% CI = 1.51-9.19). The odds of higher CRC risk perception for obese participants were more than four times (95% CI = 1.66-10.00) and overweight were more than twice the odds (95% CI = 1.03-6.31) as compared to healthy weight/underweight participants. Men using the Internet to search for health information also had greater odds of having higher CRC risk perceptions (95% CI = 1.02-4.00). Finally, men with a personal/family history of CRC were ninefold more likely to have higher CRC risk perceptions (95% CI = 2.02-41.79). CONCLUSION: Higher CRC risk perceptions were associated with older age, being obese/overweight, using the Internet as a health information source, and having a personal/family history of CRC. Culturally resonate health promotion interventions are sorely needed to elevate CRC risk perceptions for increasing intention to screen among Black men.

Nightmare disorder and low back pain in veterans: cross-sectional association and effect over time.

Low back pain (LBP) disproportionately impacts US military veterans compared with nonveterans. Although the effect of psychological conditions on LBP is regularly studied, there is little published to date investigating nightmare disorder (NMD) and LBP. The purpose of this study was to (1) investigate whether an association exists between NMD and LBP and (2) estimate the effect of NMD diagnosis on time to LBP. We used a retrospective cohort design with oversampling of those with NMD from the Veterans Health Administration (n = 15 983). We used logistic regression to assess for a cross-sectional association between NMD and LBP and survival analysis to estimate the effect of NMD on time to LBP, up to 60-month follow-up, conditioning on age, sex, race, index year, Charlson Comorbidity Index, depression, anxiety, insomnia, combat exposure, and prisoner of war history to address confounding. Odds ratios (with 95% confidence intervals [CIs]) indicated a cross-sectional association of 1.35 (1.13 to 1.60) and 1.21 (1.02 to 1.42) for NMD and LBP within 6 months and 12 months pre- or post-NMD diagnosis, respectively. Hazard ratios (HRs) indicated the effect of NMD on time to LBP that was time-dependent-HR (with 95% CIs) 1.27 (1.02 to 1.59), 1.23 (1.03 to 1.48), 1.19 (1.01 to 1.40), and 1.10 (0.94 to 1.29) in the first 3, 6, 9, and 12 months post-diagnosis, respectively-approximating the null (1.00) at >12 months. The estimated effect of NMD on LBP suggests that improved screening for NMD among veterans may help clinicians and researchers predict (or intervene to reduce) risk of future back pain.

Geographic variation and association of risk factors with incidence of colorectal cancer at small-area level.

PURPOSE: Examining spatial distribution of colorectal cancer (CRC) incidence or mortality is helpful for developing cancer control and prevention programs or for generating hypotheses. Such an investigation involves describing the spatial variation of risk factors for CRC and identifying hotspots. The aim of this study is to identify county-level risk factors that may be associated with the incidence of CRC and to map hotspots for CRC in Florida. METHODS: County-level CRC cases, recorded in 2018, were obtained from the Florida Department of Health, Division of Public Health Statistics & Performance Management (DPHSM). Data on county-level risk factors were also obtained from the same source. We used Bayesian spatial models for relative incidence rates and produced posterior predictive that indicates excess risk (hotspots) for CRC. RESULTS: The county-level unadjusted incidence rates range from .462 to 3.142. After fitting a Bayesian spatial model to the data, the results show that a decreasing risk of CRC is strongly associated with an increasing median income, higher percentage of Black population, and higher percentage of sedentary life at county level. Using exceedance probability, it is also observed that there are clustering and hotspots of high CRC incidence rates in Charlotte County in South Florida, Hernando, Sumter and Seminole counties in central Florida and Union and Washington counties in north Florida. CONCLUSION: Among few county-level variables that significantly explained the spatial variation of CRC, income disparity may need more attention for resource allocation and developing preventive intervention in high-risk areas for CRC.

Bayesian censored piecewise regression mixture models with skewness.

This paper presents censored mixture regression models with piecewise growth curves for assessing longitudinal data that exhibit multiphasic features. Such features may include censoring, skewness, measurement errors in covariates, and mixtures of unobserved subpopulations. In the process of describing those features, identification of differential effects of predictors on a response variable for a heterogeneous population (subpopulations) has recently been highly sought. Regression mixture models are key methods for assessing differential effects of predictors. In this article, we extend regression mixture models with normal distribution to incorporate (i) skew-normal distribution, (ii) left-censoring, (iii) measurement errors, and (iv) piecewise growth mixture modeling for describing multiphasic trajectories over time where the observed observations come from a mixture of unobserved subgroups. The proposed methods are illustrated using real data from an AIDS clinical study and a Bayesian approach.

Multilevel modeling, prevalence, and predictors of hypertension in Ghana: Evidence from Wave 2 of the World Health Organization's Study on global AGEing and adult health.

BACKGROUND AND AIMS: Hypertension is a major public health issue, an important risk factor for cardiovascular diseases and stroke, especially in developing countries where the rates remain unacceptably high. In Africa, hypertension is the leading driver of cardiovascular disease and stroke deaths. Identification of critical risk factors of hypertension can help formulate targeted public health programs and policies aimed at reducing the prevalence and its associated morbidity, disability, and mortality. This study attempts to develop multilevel regression, an in-depth statistical model to identify critical risk factors of hypertension. METHODS: This study used data on 4667 individuals aged ≥18 years from the nationally representative World Health Organization Study on global AGEing and adult health (SAGE) Ghana Wave 2 conducted in 2014/2015. Multilevel regression modeling was employed to identify critical risk factors for hypertension based on systolic blood pressure (SBP) (ie, SBP > 140 mmHg). Of the 4667, 27.3% were hypertensive. Final data on 4381 individuals residing in 3790 households were analyzed using multilevel models, and results were presented as adjusted odds ratios (aOR) and their associated 95% confidence intervals (CI). RESULTS: Risk factors for hypertension identified were age (aOR) = 5.4, 95% CI: 4.11-7.09), obesity (aOR = 1.51, 95% CI: 1.19-1.91), marital status (aOR = 0.75, 95% CI: 0.64-0.89), perceived health state (moderate; aOR = 1.38, 95% CI: 1.15-1.65 and bad/very bad; aOR = 1.35, 95% CI: 1.0-1.83), and difficulty with self-care (aOR = 1.64, 95% CI: 1.1-2.44). We found unobserved significant differences in the likelihood of hypertension prevalence between different households. CONCLUSION: Addressing the problem of obesity, targeting specific interventions to those aged over 50 years, and improvement in the general health of Ghanaians are paramount to reducing the prevalence and its associated morbidity, disability, and mortality. Lifestyle modification in the form of dietary intake, knowledge provision supported with strong public health message, and political will could be beneficial to the management and prevention of hypertension.

Geostatistical analysis, web-based mapping, and environmental determinants of under-5 stunting: evidence from the 2014 Ghana Demographic and Health Survey.

BACKGROUND: Stunting rates in children younger than 5 years are among the most important health indicators globally. At the national level, malnutrition accounts for about 40% of under-5 deaths in Ghana. Disease risk mapping provides opportunities for disease surveillance and targeted interventions. We aimed to estimate and map under-5 stunting prevalence in Ghana, with the goal of identifying communities at higher risk where interventions and further research can be targeted. METHODS: For this modelling study, we used data from the 2014 Ghana Demographic and Health Survey. Analyses were done on 2734 children residing in 415 geographical clusters. The outcome variable was the number of stunted children younger than 5 years in each sampled cluster. We employed a Bayesian geostatistical model to investigate both measured and unmeasured spatial risk factors for child stunting, comparing the performance of non-spatial (adjusting for selected covariates without spatial correlation), spatial (including spatial correlation), and null spatial (without the selected covariates) models. We then visualised the stunting prevalence across Ghana by mapping the predicted prevalence and exceedance probabilities to resolutions as refined as 5 km × 5 km. FINDINGS: In 2014, 535 (19·6%) of 2734 children surveyed in Ghana were stunted. Elevation (log odds mean -0·0017, 95% credible interval -0·0034 to -0·0001), precipitation (0·0403, 0·0192 to 0·0615), and aridity (-3·7013, -6·5478 to -0·8723) were environmental and climatic factors associated with stunting in the non-spatial model, but were not significant in the spatial model. Substantial geographical variations in prevalence of childhood stunting were found. The predicted mean stunting prevalence was 27·7% (SD 3·7%) with predicted prevalence ranging from 4·2% to 45·1% across Ghana. Children residing in parts of the Northern region were at highest risk of stunting, whereas parts of the Greater Accra, Brong-Ahafo, Ashanti, and Eastern regions showed some of the lowest prevalence. INTERPRETATION: There are substantial geographical differences in childhood stunting across Ghana. Our prevalence maps can be used as an effective tool to identify communities that require targeted interventions by programme managers and implementers, as part of an overall strategy to reduce the burden of malnutrition in a country with limited public health resources. FUNDING: None.

Geospatial analysis, web-based mapping and determinants of prostate cancer incidence in Georgia counties: evidence from the 2012-2016 SEER data.

BACKGROUND: Prostate cancer (CaP) cases are high in the United States. According to the American Cancer Society, there are an estimated number of 174,650 CaP new cases in 2019. The estimated number of deaths from CaP in 2019 is 31,620, making CaP the second leading cause of cancer deaths among American men with lung cancer been the first. Our goal is to estimate and map prostate cancer relative risk, with the ultimate goal of identifying counties at higher risk where interventions and further research can be targeted. METHODS: The 2012-2016 Surveillance, Epidemiology, and End Results (SEER) Program data was used in this study. Analyses were conducted on 159 Georgia counties. The outcome variable is incident prostate cancer. We employed a Bayesian geospatial model to investigate both measured and unmeasured spatial risk factors for prostate cancer. We visualised the risk of prostate cancer by mapping the predicted relative risk and exceedance probabilities. We finally developed interactive web-based maps to guide optimal policy formulation and intervention strategies. RESULTS: Number of persons above age 65 years and below poverty, higher median family income, number of foreign born and unemployed were risk factors independently associated with prostate cancer risk in the non-spatial model. Except for the number of foreign born, all these risk factors were also significant in the spatial model with the same direction of effects. Substantial geographical variations in prostate cancer incidence were found in the study. The predicted mean relative risk was 1.20 with a range of 0.53 to 2.92. Individuals residing in Towns, Clay, Union, Putnam, Quitman, and Greene counties were at increased risk of prostate cancer incidence while those residing in Chattahoochee were at the lowest risk of prostate cancer incidence. CONCLUSION: Our results can be used as an effective tool in the identification of counties that require targeted interventions and further research by program managers and policy makers as part of an overall strategy in reducing the prostate cancer burden in Georgia State and the United States as a whole.

Bayesian semiparametric growth models for measurement error and missing data in CD4/CD8 ratio: Application to AIDS Study.

In clinical research and practice, there is often an interest in assessing the effect of time varying predictors, such as CD4/CD8 ratio, on immune recovery following antiretroviral therapy. Such predictors are measured with errors, and ignoring those measurement errors during data analysis may lead to biased results. Though parametric methods have been used for reducing biases, they usually depend on untestable assumptions. To relax those assumptions, this paper presents semiparametric mixed-effect models which deal with predictors having measurement errors and missing values. We develop a fully Bayesian approach for fitting these models and discriminating between patients who are potentially progressors or nonprogressors to severe disease condition (AIDS). The proposed methods are demonstrated using real data from an AIDS clinical study.

Evaluating construct equivalence of youth depression measures across multiple measures and multiple studies.

Construct equivalence of measures across studies is necessary for synthesizing results when combining data in meta-analysis or integrative data analysis. We discuss several assumptions required for construct equivalence, and review methods using individual-level data and item response theory (IRT) analysis for detecting or adjusting for violations of these assumptions. We apply IRT to data from 7 measures of depressive symptoms for 4,283 youth from 16 randomized prevention trials. Findings indicate that these data violate assumptions of conditional independence. Bifactor IRT models find that depression measures contain substantial reporter variance, and indicate that a single common factor model would be substantially biased. Separate analyses of ratings by youth find stronger evidence for construct equivalence, but factor invariance across sex and age does not hold. We conclude that data synthesis studies employing measures of youth depression should analyze results separately by reporter, explore more complex approaches to integrate these different perspectives, and explore methods that adjust for sex and age differences in item functioning. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Heterogeneous growth bent-cable models for time-to-event and longitudinal data: application to AIDS studies.

The major limitations of growth curve mixture models for HIV/AIDS data are the usual assumptions of normality and monophasic curves within latent classes. This article addresses these limitations by using non-normal skewed distributions and multiphasic patterns for outcomes of prospective studies. For such outcomes, new skew-t (ST) distributions are proposed for modeling heterogeneous growth trajectories, which exhibit not abrupt but gradual multiphasic changes from a declining trend to an increasing trend over time. We assess these clinically important features of longitudinal HIV/AIDS data using the bent-cable framework within a context of a joint modeling of time-to-event process and response process. A real dataset from an AIDS clinical study is used to illustrate the proposed methods.

Bayesian two-part bent-cable Tobit models with skew distributions: Application to AIDS studies.

This paper presents a new development of a bent-cable two-part Tobit model to identify both phasic patterns and mixture of advancing (to AIDS) and non-advancing patients of HIV. In identification of such phasic patterns, estimation of a transition period for the development of drug resistance to antiretroviral (ARV) drug or therapy is carried out using longitudinal data that have a gradual change from a declining phase to an increasing phase. In addition to phasic changes, there are also problems of skewness and left-censoring in the response variable because of a lower limit of detection. A relatively large percentage of data below limit of detection are recorded more than expected under an assumed skew-distribution. To properly accommodate these features, we present an extension of the random effects bent-cable Tobit model that incorporates a mixture of true undetectable observations and those values from a skew-normal distribution for a response with left-censoring, skewness and phasic patterns. The proposed methods are illustrated using real data from an AIDS clinical study.

Joint bent-cable Tobit models for longitudinal and time-to-event data.

In this article, we show how to estimate a transition period for the evolvement of drug resistance to antiretroviral (ARV) drug or other related treatments in the framework of developing a Bayesian method for jointly analyzing time-to-event and longitudinal data. For HIV/AIDS longitudinal data, developmental trajectories of viral loads tend to show a gradual change from a declining trend after initiation of treatment to an increasing trend without an abrupt change. Such characteristics of trajectories are also associated with a time-to-event process. To assess these clinically important features, we develop a joint bent-cable Tobit model for the time-to-event and left-censored response variable with skewness and phasic developments. Random effects are used to determine the stochastic dependence between the time-to-event process and response process. The proposed method is illustrated using real data from an AIDS clinical study.

Joint two-part Tobit models for longitudinal and time-to-event data.

In this article, we show how Tobit models can address problems of identifying characteristics of subjects having left-censored outcomes in the context of developing a method for jointly analyzing time-to-event and longitudinal data. There are some methods for handling these types of data separately, but they may not be appropriate when time to event is dependent on the longitudinal outcome, and a substantial portion of values are reported to be below the limits of detection. An alternative approach is to develop a joint model for the time-to-event outcome and a two-part longitudinal outcome, linking them through random effects. This proposed approach is implemented to assess the association between the risk of decline of CD4/CD8 ratio and rates of change in viral load, along with discriminating between patients who are potentially progressors to AIDS from patients who do not. We develop a fully Bayesian approach for fitting joint two-part Tobit models and illustrate the proposed methods on simulated and real data from an AIDS clinical study.

Bayesian inference for two-part mixed-effects model using skew distributions, with application to longitudinal semicontinuous alcohol data.

Semicontinuous data featured with an excessive proportion of zeros and right-skewed continuous positive values arise frequently in practice. One example would be the substance abuse/dependence symptoms data for which a substantial proportion of subjects investigated may report zero. Two-part mixed-effects models have been developed to analyze repeated measures of semicontinuous data from longitudinal studies. In this paper, we propose a flexible two-part mixed-effects model with skew distributions for correlated semicontinuous alcohol data under the framework of a Bayesian approach. The proposed model specification consists of two mixed-effects models linked by the correlated random effects: (i) a model on the occurrence of positive values using a generalized logistic mixed-effects model (Part I); and (ii) a model on the intensity of positive values using a linear mixed-effects model where the model errors follow skew distributions including skew- t and skew-normal distributions (Part II). The proposed method is illustrated with an alcohol abuse/dependence symptoms data from a longitudinal observational study, and the analytic results are reported by comparing potential models under different random-effects structures. Simulation studies are conducted to assess the performance of the proposed models and method.

Bayesian analysis of piecewise growth mixture models with skew-t distributions: Application to AIDS studies.

A major problem in HIV/AIDS studies is the development of drug resistance to antiretroviral (ARV) drug or therapy. Estimating the time at which such drug resistance would develop is usually sought. The goal of this article is to perform this estimation by developing growth mixture models with change-points and skew-t distributions based on longitudinal data. For such data, following ARV treatment, the profile of each subject's viral load tends to follow a 'broken stick' like growth trajectory, indicating multiple phases of decline and increase in viral loads. These multiple phases with multiple change-points are captured by subject-specific random parameters of growth curve models. To account for heterogeneity of drug resistance among subjects, the change-points are also allowed to differ by subgroups (subpopulations) of patients classified into latent classes on the basis of trajectories of observed viral loads. The proposed methods are illustrated using real data from an AIDS clinical study.

Bayesian bent-cable growth mixture tobit models for longitudinal data with skewness and detection limit: application to AIDS studies.

This paper presents a new Bayesian methodology for identifying a transition period for the development of drug resistance to antiretroviral drug or therapy in HIV/AIDS studies or other related fields. Estimation of such a transition period requires an availability of longitudinal data where growth trajectories of a response variable tend to exhibit a gradual change from a declining trend to an increasing trend rather than an abrupt change. We assess this clinically important feature of the longitudinal HIV/AIDS data using the bent-cable framework within a growth mixture Tobit model. To account for heterogeneity of drug resistance among subjects, the parameters of the bent-cable growth mixture Tobit model are also allowed to differ by subgroups (subpopulations) of patients classified into latent classes on the basis of trajectories of observed viral load data with skewness and left-censoring. The proposed methods are illustrated using real data from an AIDS clinical study. Copyright © 2016 John Wiley & Sons, Ltd.

Testing moderation in network meta-analysis with individual participant data.

Meta-analytic methods for combining data from multiple intervention trials are commonly used to estimate the effectiveness of an intervention. They can also be extended to study comparative effectiveness, testing which of several alternative interventions is expected to have the strongest effect. This often requires network meta-analysis (NMA), which combines trials involving direct comparison of two interventions within the same trial and indirect comparisons across trials. In this paper, we extend existing network methods for main effects to examining moderator effects, allowing for tests of whether intervention effects vary for different populations or when employed in different contexts. In addition, we study how the use of individual participant data may increase the sensitivity of NMA for detecting moderator effects, as compared with aggregate data NMA that employs study-level effect sizes in a meta-regression framework. A new NMA diagram is proposed. We also develop a generalized multilevel model for NMA that takes into account within-trial and between-trial heterogeneity and can include participant-level covariates. Within this framework, we present definitions of homogeneity and consistency across trials. A simulation study based on this model is used to assess effects on power to detect both main and moderator effects. Results show that power to detect moderation is substantially greater when applied to individual participant data as compared with study-level effects. We illustrate the use of this method by applying it to data from a classroom-based randomized study that involved two sub-trials, each comparing interventions that were contrasted with separate control groups. Copyright © 2016 John Wiley & Sons, Ltd.

Piecewise Growth Mixture Tobit Models: Application to AIDS Studies.

Piecewise growth models are very flexible methods for assessing distinct phases of development or progression in longitudinal data. As an extension of these models, this paper presents piecewise growth mixture Tobit models (PGMTMs) for describing phasic changes of individual trajectories over time where the longitudinal data has a mixture of subpopulations and where left censoring due to a lower limit of detection (LOD) is also observed. There has been relatively little work done simultaneously modeling heterogeneous growth trajectories, segmented phases of progression, and left-censoring with skewed responses. The proposed methods are illustrated using real data from an AIDS clinical study. Analysis results suggested two classes of viral load growth trajectories: Class 1 started with a decline in viral load after treatment but rebound after change-point; Class 2 had a decrease the same as the Class 1 and continued a slower decrease over time.