THE USE OF MULTIVARIATE ANALYSIS OF COVARIANCE IN EVALUATING THE COMPARABILITY OF LABORATORY DETERMINATIONS IN COOPERATIVE STUDIES

The role of clinical chemistry has traditionally been to evaluate acutely ill or hospitalized patients. Traditional statistical methods have serious drawbacks in that they use univariate techniques. To demonstrate alternative methodology, a multivariate analysis of covariance model was developed and applied to the data from the Cooperative Study of Sickle Cell Disease.^ The purpose of developing the model for the laboratory data from the CSSCD was to evaluate the comparability of the results from the different clinics. Several variables were incorporated into the model in order to control for possible differences among the clinics that might confound any real laboratory differences.^ Differences for LDH, alkaline phosphatase and SGOT were identified which will necessitate adjustments by clinic whenever these data are used. In addition, aberrant clinic values for LDH, creatinine and BUN were also identified.^ The use of any statistical technique including multivariate analysis without thoughtful consideration may lead to spurious conclusions that may not be corrected for some time, if ever. However, the advantages of multivariate analysis far outweigh its potential problems. If its use increases as it should, the applicability to the analysis of laboratory data in prospective patient monitoring, quality control programs, and interpretation of data from cooperative studies could well have a major impact on the health and well being of a large number of individuals. ^

Bayesian generalized linear models for meta-analysis of diagnostic tests

With the recognition of the importance of evidence-based medicine, there is an emerging need for methods to systematically synthesize available data. Specifically, methods to provide accurate estimates of test characteristics for diagnostic tests are needed to help physicians make better clinical decisions. To provide more flexible approaches for meta-analysis of diagnostic tests, we developed three Bayesian generalized linear models. Two of these models, a bivariate normal and a binomial model, analyzed pairs of sensitivity and specificity values while incorporating the correlation between these two outcome variables. Noninformative independent uniform priors were used for the variance of sensitivity, specificity and correlation. We also applied an inverse Wishart prior to check the sensitivity of the results. The third model was a multinomial model where the test results were modeled as multinomial random variables. All three models can include specific imaging techniques as covariates in order to compare performance. Vague normal priors were assigned to the coefficients of the covariates. The computations were carried out using the 'Bayesian inference using Gibbs sampling' implementation of Markov chain Monte Carlo techniques. We investigated the properties of the three proposed models through extensive simulation studies. We also applied these models to a previously published meta-analysis dataset on cervical cancer as well as to an unpublished melanoma dataset. In general, our findings show that the point estimates of sensitivity and specificity were consistent among Bayesian and frequentist bivariate normal and binomial models. However, in the simulation studies, the estimates of the correlation coefficient from Bayesian bivariate models are not as good as those obtained from frequentist estimation regardless of which prior distribution was used for the covariance matrix. The Bayesian multinomial model consistently underestimated the sensitivity and specificity regardless of the sample size and correlation coefficient. In conclusion, the Bayesian bivariate binomial model provides the most flexible framework for future applications because of its following strengths: (1) it facilitates direct comparison between different tests; (2) it captures the variability in both sensitivity and specificity simultaneously as well as the intercorrelation between the two; and (3) it can be directly applied to sparse data without ad hoc correction. ^

A GENETIC MARKER STUDY OF NON - INSULIN DEPENDENT DIABETICS IN THE MEXICAN-AMERICAN COMMUNITY FROM STARR COUNTY, TEXAS (PLASMA GLUCOSE, HEMOGLOBIN ANALYSIS)

Diabetes mellitus occurs in two forms, insulin-dependent (IDDM, formerly called juvenile type) and non-insulin dependent (NIDDM, formerly called adult type). Prevalence figures from around the world for NIDDM, show that all societies and all races are affected; although uncommon in some populations (.4%), it is common (10%) or very common (40%) in others (Tables 1 and 2).^ In Mexican-Americans in particular, the prevalence rates (7-10%) are intermediate to those in Caucasians (1-2%) and Amerindians (35%). Information about the distribution of the disease and identification of high risk groups for developing glucose intolerance or its vascular manifestations by the study of genetic markers will help to clarify and solve some of the problems from the public health and the genetic point of view.^ This research was designed to examine two general areas in relation to NIDDM. The first aims to determine the prevalence of polymorphic genetic markers in two groups distinguished by the presence or absence of diabetes and to observe if there are any genetic marker-disease association (univariate analysis using two by two tables and logistic regression to study the individual and joint effects of the different variables). The second deals with the effect of genetic differences on the variation in fasting plasma glucose and percent glycosylated hemoglobin (HbAl) (analysis of Covariance for each marker, using age and sex as covariates).^ The results from the first analysis were not statistically significant at the corrected p value of 0.003 given the number of tests that were performed. From the analysis of covariance of all the markers studied, only Duffy and Phosphoglucomutase were statistically significant but poor predictors, given that the amount they explain in terms of variation in glycosylated hemoglobin is very small.^ Trying to determine the polygenic component of chronic disease is not an easy task. This study confirms the fact that a larger and random or representative sample is needed to be able to detect differences in the prevalence of a marker for association studies and in the genetic contribution to the variation in glucose and glycosylated hemoglobin. The importance that ethnic homogeneity in the groups studied and standardization in the methodology will have on the results has been stressed. ^

Analysis of interval -censored events in hypertension studies: Evaluation of treatment of coronary heart disease

Many statistical studies feature data with both exact-time and interval-censored events. While a number of methods currently exist to handle interval-censored events and multivariate exact-time events separately, few techniques exist to deal with their combination. This thesis develops a theoretical framework for analyzing a multivariate endpoint comprised of a single interval-censored event plus an arbitrary number of exact-time events. The approach fuses the exact-time events, modeled using the marginal method of Wei, Lin, and Weissfeld, with a piecewise-exponential interval-censored component. The resulting model incorporates more of the information in the data and also removes some of the biases associated with the exclusion of interval-censored events. A simulation study demonstrates that our approach produces reliable estimates for the model parameters and their variance-covariance matrix. As a real-world data example, we apply this technique to the Systolic Hypertension in the Elderly Program (SHEP) clinical trial, which features three correlated events: clinical non-fatal myocardial infarction, fatal myocardial infarction (two exact-time events), and silent myocardial infarction (one interval-censored event). ^