A Bayesian approach to estimating the regression coefficients of a multinomial logit model

A Bayesian approach to estimation of the regression coefficients of a multinominal logit model with ordinal scale response categories is presented. A Monte Carlo method is used to construct the posterior distribution of the link function. The link function is treated as an arbitrary scalar function. Then the Gauss-Markov theorem is used to determine a function of the link which produces a random vector of coefficients. The posterior distribution of the random vector of coefficients is used to estimate the regression coefficients. The method described is referred to as a Bayesian generalized least square (BGLS) analysis. Two cases involving multinominal logit models are described. Case I involves a cumulative logit model and Case II involves a proportional-odds model. All inferences about the coefficients for both cases are described in terms of the posterior distribution of the regression coefficients. The results from the BGLS method are compared to maximum likelihood estimates of the regression coefficients. The BGLS method avoids the nonlinear problems encountered when estimating the regression coefficients of a generalized linear model. The method is not complex or computationally intensive. The BGLS method offers several advantages over Bayesian approaches. ^

Logistic regression models for ordinal response: A study of self-efficacy in colorectal cancer screening

The ordinal logistic regression models are used to analyze the dependant variable with multiple outcomes that can be ranked, but have been underutilized. In this study, we describe four logistic regression models for analyzing the ordinal response variable. ^ In this methodological study, the four regression models are proposed. The first model uses the multinomial logistic model. The second is adjacent-category logit model. The third is the proportional odds model and the fourth model is the continuation-ratio model. We illustrate and compare the fit of these models using data from the survey designed by the University of Texas, School of Public Health research project PCCaSO (Promoting Colon Cancer Screening in people 50 and Over), to study the patient’s confidence in the completion colorectal cancer screening (CRCS). ^ The purpose of this study is two fold: first, to provide a synthesized review of models for analyzing data with ordinal response, and second, to evaluate their usefulness in epidemiological research, with particular emphasis on model formulation, interpretation of model coefficients, and their implications. Four ordinal logistic models that are used in this study include (1) Multinomial logistic model, (2) Adjacent-category logistic model [9], (3) Continuation-ratio logistic model [10], (4) Proportional logistic model [11]. We recommend that the analyst performs (1) goodness-of-fit tests, (2) sensitivity analysis by fitting and comparing different models.^

HPV vaccine coverage in Texas counties: An application of multilevel, small area estimation

Health departments, research institutions, policy-makers, and healthcare providers are often interested in knowing the health status of their clients/constituents. Without the resources, financially or administratively, to go out into the community and conduct health assessments directly, these entities frequently rely on data from population-based surveys to supply the information they need. Unfortunately, these surveys are ill-equipped for the job due to sample size and privacy concerns. Small area estimation (SAE) techniques have excellent potential in such circumstances, but have been underutilized in public health due to lack of awareness and confidence in applying its methods. The goal of this research is to make model-based SAE accessible to a broad readership using clear, example-based learning. Specifically, we applied the principles of multilevel, unit-level SAE to describe the geographic distribution of HPV vaccine coverage among females aged 11-26 in Texas.^ Multilevel (3 level: individual, county, public health region) random-intercept logit models of HPV vaccination (receipt of ≥ 1 dose Gardasil® ) were fit to data from the 2008 Behavioral Risk Factor Surveillance System (outcome and level 1 covariates) and a number of secondary sources (group-level covariates). Sampling weights were scaled (level 1) or constructed (levels 2 & 3), and incorporated at every level. Using the regression coefficients (and standard errors) from the final models, I simulated 10,000 datasets for each regression coefficient from the normal distribution and applied them to the logit model to estimate HPV vaccine coverage in each county and respective demographic subgroup. For simplicity, I only provide coverage estimates (and 95% confidence intervals) for counties.^ County-level coverage among females aged 11-17 varied from 6.8-29.0%. For females aged 18-26, coverage varied from 1.9%-23.8%. Aggregated to the state level, these values translate to indirect state estimates of 15.5% and 11.4%, respectively; both of which fall within the confidence intervals for the direct estimates of HPV vaccine coverage in Texas (Females 11-17: 17.7%, 95% CI: 13.6, 21.9; Females 18-26: 12.0%, 95% CI: 6.2, 17.7).^ Small area estimation has great potential for informing policy, program development and evaluation, and the provision of health services. Harnessing the flexibility of multilevel, unit-level SAE to estimate HPV vaccine coverage among females aged 11-26 in Texas counties, I have provided (1) practical guidance on how to conceptualize and conduct modelbased SAE, (2) a robust framework that can be applied to other health outcomes or geographic levels of aggregation, and (3) HPV vaccine coverage data that may inform the development of health education programs, the provision of health services, the planning of additional research studies, and the creation of local health policies.^

Logistic regression with Markov chains as covariates

The tobacco-specific nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) is an obvious carcinogen for lung cancer. Since CBMN (Cytokinesis-blocked micronucleus) has been found to be extremely sensitive to NNK-induced genetic damage, it is a potential important factor to predict the lung cancer risk. However, the association between lung cancer and NNK-induced genetic damage measured by CBMN assay has not been rigorously examined. ^ This research develops a methodology to model the chromosomal changes under NNK-induced genetic damage in a logistic regression framework in order to predict the occurrence of lung cancer. Since these chromosomal changes were usually not observed very long due to laboratory cost and time, a resampling technique was applied to generate the Markov chain of the normal and the damaged cell for each individual. A joint likelihood between the resampled Markov chains and the logistic regression model including transition probabilities of this chain as covariates was established. The Maximum likelihood estimation was applied to carry on the statistical test for comparison. The ability of this approach to increase discriminating power to predict lung cancer was compared to a baseline "non-genetic" model. ^ Our method offered an option to understand the association between the dynamic cell information and lung cancer. Our study indicated the extent of DNA damage/non-damage using the CBMN assay provides critical information that impacts public health studies of lung cancer risk. This novel statistical method could simultaneously estimate the process of DNA damage/non-damage and its relationship with lung cancer for each individual.^

Renal Cryoablation: Investigation of Periprocedural Visualization Tools and Treatment Response Quantification

Cryoablation for small renal tumors has demonstrated sufficient clinical efficacy over the past decade as a non-surgical nephron-sparing approach for treating renal masses for patients who are not surgical candidates. Minimally invasive percutaneous cryoablations have been performed with image guidance from CT, ultrasound, and MRI. During the MRI-guided cryoablation procedure, the interventional radiologist visually compares the iceball size on monitoring images with respect to the original tumor on separate planning images. The comparisons made during the monitoring step are time consuming, inefficient and sometimes lack the precision needed for decision making, requiring the radiologist to make further changes later in the procedure. This study sought to mitigate uncertainty in these visual comparisons by quantifying tissue response to cryoablation and providing visualization of the response during the procedure. Based on retrospective analysis of MR-guided cryoablation patient data, registration and segmentation algorithms were investigated and implemented for periprocedural visualization to deliver iceball position/size with respect to planning images registered within 3.3mm with at least 70% overlap and a quantitative logit model was developed to relate perfusion deficit in renal parenchyma visualized in verification images as a result of iceball size visualized in monitoring images. Through retrospective study of 20 patient cases, the relationship between likelihood of perfusion loss in renal parenchyma and distance within iceball was quantified and iteratively fit to a logit curve. Using the parameters from the logit fit, the margin for 95% perfusion loss likelihood was found to be 4.28 mm within the iceball. The observed margin corresponds well with the clinically accepted margin of 3-5mm within the iceball. In order to display the iceball position and perfusion loss likelihood to the radiologist, algorithms were implemented to create a fast segmentation and registration module which executed in under 2 minutes, within the clinically-relevant 3 minute monitoring period. Using 16 patient cases, the average Hausdorff distance was reduced from 10.1mm to 3.21 mm with average DSC increased from 46.6% to 82.6% before and after registration.

Comparing the factorial structure, invariance, and predictive validity of transtheoretical model constructs for alcohol use across restricted and unrestricted settings

With substance abuse treatment expanding in prisons and jails, understanding how behavior change interacts with a restricted setting becomes more essential. The Transtheoretical Model (TTM) has been used to understand intentional behavior change in unrestricted settings, however, evidence indicates restrictive settings can affect the measurement and structure of the TTM constructs. The present study examined data from problem drinkers at baseline and end-of-treatment from three studies: (1) Project CARE (n = 187) recruited inmates from a large county jail; (2) Project Check-In (n = 116) recruited inmates from a state prison; (3) Project MATCH, a large multi-site alcohol study had two recruitment arms, aftercare (n = 724 pre-treatment and 650 post-treatment) and outpatient (n = 912 pre-treatment and 844 post-treatment). The analyses were conducted using cross-sectional data to test for non-invariance of measures of the TTM constructs: readiness, confidence, temptation, and processes of change (Structural Equation Modeling, SEM) across restricted and unrestricted settings. Two restricted (jail and aftercare) and one unrestricted group (outpatient) entering treatment and one restricted (prison) and two unrestricted groups (aftercare and outpatient) at end-of-treatment were contrasted. In addition TTM end-of-treatment profiles were tested as predictors of 12 month drinking outcomes (Profile Analysis). Although SEM did not indicate structural differences in the overall TTM construct model across setting types, there were factor structure differences on the confidence and temptation constructs at pre-treatment and in the factor structure of the behavioral processes at the end-of-treatment. For pre-treatment temptation and confidence, differences were found in the social situations factor loadings and in the variance for the confidence and temptation latent factors. For the end-of-treatment behavioral processes, differences across the restricted and unrestricted settings were identified in the counter-conditioning and stimulus control factor loadings. The TTM end-of-treatment profiles were not predictive of drinking outcomes in the prison sample. Both pre and post-treatment differences in structure across setting types involved constructs operationalized with behaviors that are limited for those in restricted settings. These studies suggest the TTM is a viable model for explicating addictive behavior change in restricted settings but calls for modification of subscale items that refer to specific behaviors and caution in interpreting the mean differences across setting types for problem drinkers. ^

Clinical factors associated with growth and mortality of pediatric patients on HAART at Mulago PIDC, Uganda, 2003--2006

The global social and economic burden of HIV/AIDS is great, with over forty million people reported to be living with HIV/AIDS at the end of 2005; two million of these are children from birth to 15 years of age. Antiretroviral therapy has been shown to improve growth and survival of HIV-infected individuals. The purpose of this study is to describe a cohort of HIV-infected pediatric patients and assess the association between clinical factors, with growth and mortality outcomes. ^ This was a historical cohort study. Medical records of infants and children receiving HIV care at Mulago Pediatric Infectious Disease Clinic (PIDC) in Uganda between July 2003 and March 2006 were analyzed. Height and weight measurements were age and sex standardized to Centers for Disease Control and prevention (CDC) 2000 reference. Descriptive and logistic regression analyses were performed to identify covariates associated with risk of stunting or being underweight, and mortality. Longitudinal regression analysis with a mixed model using autoregressive covariance structure was used to compare change in height and weight before and after initiation of highly active antiretroviral therapy (HAART). ^ The study population was comprised of 1059 patients 0-20 years of age, the majority of whom were aged thirteen years and below (74.6%). Mean height-for-age before initiation of HAART was in the 10th percentile, mean weight-for-age was in the 8th percentile, and the mean weight-for-height was in the 23rd percentile. Initiation of HAART resulted in improvement in both the mean standardized weight-for-age Z score and weight-for-age percentiles (p <0.001). Baseline age, and weight-for-age Z score were associated with stunting (p <0.001). A negative weight-for-age Z score was associated with stunting (OR 4.60, CI 3.04-5.49). Risk of death decreased from 84% in the >2-8 years age category to 21% in the >13 years age category respectively, compared to the 0-2 years of age (p <0.05). ^ This pediatric population gained weight significantly more rapidly than height after starting HAART. A low weight-for-age Z score was associated with poor survival in children. These findings suggest that age, weight, and height measurements be monitored closely at Mulago PIDC. ^

Effect of a lifestyle physical activity intervention on transtheoretical model variables

Purpose. To determine the effect a stage-based, lifestyle physical activity intervention has on Transtheoretical Model variables in a population of breast cancer survivors. ^ Methods. Sedentary breast cancer survivors (N=60) were randomized to either a standard care study condition or to a 6-month, 21-session intervention. The Transtheoretical Model variables stage of change, self-efficacy, decisional balance (pros and cons to exercise), and processes of change were measured at baseline, 3 months, and 6 months. ^ Results. Women in the lifestyle group had significantly higher self-efficacy than women in the standard care group (F=9.55, p=0.003). Although there was not a significant difference between the two groups for perceived pros of exercise, there was a significant difference between the groups for perceived cons of exercise. Women in the lifestyle group perceived significantly fewer cons of exercise at both 3 and 6 months compared with women in the standard care condition (F=5.416, p=0.025). Between baseline and the 6 month assessment, the intervention also had an effect on three of the processes of change, while seven of the processes were not significantly affected by the intervention. ^ Conclusions. Data from the pilot study suggest that a stage-based, lifestyle physical activity intervention has an effect on Transtheoretical Model variables, which have been shown to facilitate exercise adoption, and should be tested in a larger trial. ^

Diabetic nephropathy in African-Americans with type 1 diabetes: The new Jersey 725 study

Diabetic nephropathy is the most common cause of end-stage renal disease (ESRD) in the United States. African-Americans and patients with type 1 diabetes (T1D) are at increased risk. We studied the rate and factors that influenced progression of glomerular filtration rate (GFR) in 401 African-American T1D patients who were followed for 6 years through the observational cohort New Jersey 725 study. Patients with ESRD and/or GFR<20 ml/min were excluded. The mean (SD) baseline GFR was 106.8 (27.04) ml/min and it decreased by 13.8 (mean, SD 32.2) ml/min during the 6-year period (2.3 ml/min/year). In patients with baseline macroproteinuria, GFR decreased by 31.8 (39.0) ml/min (5.3 ml/min/year) compared to 8.2 (mean, SD 27.6) ml/min (1.3 ml/min/year) in patients without it (p<0.00001). Six-year GFR fell to <20 ml/min in 5.25% of all patients, but in 16.8% of macroproteinuric patients.^ A model including baseline GFR, proteinuria category and hypertension category, explained 35% of the 6-year GFR variability (p<0.0001). After adjustment for other variables in the model, 6-year GFR was 24.9 ml/min lower in macroproteinuric patients than in those without proteinuria (p=0.0001), and 12.6 ml/min lower in patients with treated but uncontrolled hypertension compared to normotensive patients (p=0.003). In this sample of patients, with an elevated mean glycosylated hemoglobin of 12.4%, glycemic control did not independently influence GFR deterioration, nor did BMI, cholesterol, gender, age at diabetes onset or socioeconomic level.^ Taken together, our findings suggest that proteinuria and hypertension are the most important factors associated with GFR deterioration in African-American T1D patients.^

A novel method for evaluating an interaction effect of correlated continuous covariates in a linear model

Interaction effect is an important scientific interest for many areas of research. Common approach for investigating the interaction effect of two continuous covariates on a response variable is through a cross-product term in multiple linear regression. In epidemiological studies, the two-way analysis of variance (ANOVA) type of method has also been utilized to examine the interaction effect by replacing the continuous covariates with their discretized levels. However, the implications of model assumptions of either approach have not been examined and the statistical validation has only focused on the general method, not specifically for the interaction effect.^ In this dissertation, we investigated the validity of both approaches based on the mathematical assumptions for non-skewed data. We showed that linear regression may not be an appropriate model when the interaction effect exists because it implies a highly skewed distribution for the response variable. We also showed that the normality and constant variance assumptions required by ANOVA are not satisfied in the model where the continuous covariates are replaced with their discretized levels. Therefore, naïve application of ANOVA method may lead to an incorrect conclusion. ^ Given the problems identified above, we proposed a novel method modifying from the traditional ANOVA approach to rigorously evaluate the interaction effect. The analytical expression of the interaction effect was derived based on the conditional distribution of the response variable given the discretized continuous covariates. A testing procedure that combines the p-values from each level of the discretized covariates was developed to test the overall significance of the interaction effect. According to the simulation study, the proposed method is more powerful then the least squares regression and the ANOVA method in detecting the interaction effect when data comes from a trivariate normal distribution. The proposed method was applied to a dataset from the National Institute of Neurological Disorders and Stroke (NINDS) tissue plasminogen activator (t-PA) stroke trial, and baseline age-by-weight interaction effect was found significant in predicting the change from baseline in NIHSS at Month-3 among patients received t-PA therapy.^

A multiplicative model for standardizing vital rates without using external standards

Standardization is a common method for adjusting confounding factors when comparing two or more exposure category to assess excess risk. Arbitrary choice of standard population in standardization introduces selection bias due to healthy worker effect. Small sample in specific groups also poses problems in estimating relative risk and the statistical significance is problematic. As an alternative, statistical models were proposed to overcome such limitations and find adjusted rates. In this dissertation, a multiplicative model is considered to address the issues related to standardized index namely: Standardized Mortality Ratio (SMR) and Comparative Mortality Factor (CMF). The model provides an alternative to conventional standardized technique. Maximum likelihood estimates of parameters of the model are used to construct an index similar to the SMR for estimating relative risk of exposure groups under comparison. Parametric Bootstrap resampling method is used to evaluate the goodness of fit of the model, behavior of estimated parameters and variability in relative risk on generated sample. The model provides an alternative to both direct and indirect standardization method. ^

Hispanic incorporation and access to health care in Houston and Harris County, Texas: Adopting an Hispanic perspective to the Aday and Andersen model

This study examines Hispanic levels of incorporation and access to health care. Applying the Aday and Andersen framework for the study of access, the study examined the relationship between two levels of Hispanic incorporation into U.S. society, i.e., mainstream versus ethnic, and potential and realized measures of access to health care. Data for the study were drawn from a 1992 telephone survey of 600 randomly selected Hispanics in Houston and Harris County.^ The hypotheses tested were: (1) Hispanics who are incorporated into mainstream society are more likely to have better potential and realized access to health care than those who are incorporated into ethnic-group enclaves regardless of their socioeconomic status (SES), health status and health needs, and (2) there is no interaction between the levels of incorporation (mainstream or ethnic) and SES, health status, and health needs in predicting potential and realized access.^ The data analysis supported Hypothesis One for the two measures of potential access. The results of bivariate and multiple logistic regression analyses indicated that for Hispanics in Houston and Harris County, being in the "mainstream" incorporation category increased their potential access to care, having "health insurance" and a "regular place of care". For the selected measure of realized access, having a "regular check-up", the analysis did not demonstrate statistically significant differences in having a regular check-up among Hispanics incorporated in the ethnic or mainstream incorporation categories.^ Hypothesis Two, that there is no interaction between the levels of incorporation and socioeconomic characteristics, health status, and health needs in predicting potential and realized access among Hispanics was supported by the data. The results of the logistic regression analysis showed that, after adjusting for socioeconomic status, health status, and health needs, the association between "level of incorporation" and the two measures of potential access ("health insurance" and having a "usual place of care") was not modified by the control variables nor by their interaction with level of incorporation. That is, the effect of incorporation on Hispanics' health insurance coverage, and having a usual place of care, was homogenous across Hispanics with different SES and health status.^ The main research implication of this dissertation is the employment of a theoretical framework for the assessment of cultural factors essential to research on migrating heterogeneous subpopulations. It also provided strategies to solve practical and methodological difficulties in the secondary analyses of data on these populations. ^

Application of the general linear model to assess the effect of missing data on bone marrow mononuclear cell therapy with infusion timing and follow-up MRI

With most clinical trials, missing data presents a statistical problem in evaluating a treatment's efficacy. There are many methods commonly used to assess missing data; however, these methods leave room for bias to enter the study. This thesis was a secondary analysis on data taken from TIME, a phase 2 randomized clinical trial conducted to evaluate the safety and effect of the administration timing of bone marrow mononuclear cells (BMMNC) for subjects with acute myocardial infarction (AMI).^ We evaluated the effect of missing data by comparing the variance inflation factor (VIF) of the effect of therapy between all subjects and only subjects with complete data. Through the general linear model, an unbiased solution was made for the VIF of the treatment's efficacy using the weighted least squares method to incorporate missing data. Two groups were identified from the TIME data: 1) all subjects and 2) subjects with complete data (baseline and follow-up measurements). After the general solution was found for the VIF, it was migrated Excel 2010 to evaluate data from TIME. The resulting numerical value from the two groups was compared to assess the effect of missing data.^ The VIF values from the TIME study were considerably less in the group with missing data. By design, we varied the correlation factor in order to evaluate the VIFs of both groups. As the correlation factor increased, the VIF values increased at a faster rate in the group with only complete data. Furthermore, while varying the correlation factor, the number of subjects with missing data was also varied to see how missing data affects the VIF. When subjects with only baseline data was increased, we saw a significant rate increase in VIF values in the group with only complete data while the group with missing data saw a steady and consistent increase in the VIF. The same was seen when we varied the group with follow-up only data. This essentially showed that the VIFs steadily increased when missing data is not ignored. When missing data is ignored as with our comparison group, the VIF values sharply increase as correlation increases.^

The number and timing of time -dependent covariate measurements for the Cox regression model

The problem of analyzing data with updated measurements in the time-dependent proportional hazards model arises frequently in practice. One available option is to reduce the number of intervals (or updated measurements) to be included in the Cox regression model. We empirically investigated the bias of the estimator of the time-dependent covariate while varying the effect of failure rate, sample size, true values of the parameters and the number of intervals. We also evaluated how often a time-dependent covariate needs to be collected and assessed the effect of sample size and failure rate on the power of testing a time-dependent effect.^ A time-dependent proportional hazards model with two binary covariates was considered. The time axis was partitioned into k intervals. The baseline hazard was assumed to be 1 so that the failure times were exponentially distributed in the ith interval. A type II censoring model was adopted to characterize the failure rate. The factors of interest were sample size (500, 1000), type II censoring with failure rates of 0.05, 0.10, and 0.20, and three values for each of the non-time-dependent and time-dependent covariates (1/4,1/2,3/4).^ The mean of the bias of the estimator of the coefficient of the time-dependent covariate decreased as sample size and number of intervals increased whereas the mean of the bias increased as failure rate and true values of the covariates increased. The mean of the bias of the estimator of the coefficient was smallest when all of the updated measurements were used in the model compared with two models that used selected measurements of the time-dependent covariate. For the model that included all the measurements, the coverage rates of the estimator of the coefficient of the time-dependent covariate was in most cases 90% or more except when the failure rate was high (0.20). The power associated with testing a time-dependent effect was highest when all of the measurements of the time-dependent covariate were used. An example from the Systolic Hypertension in the Elderly Program Cooperative Research Group is presented. ^