Multiple imputation for missing continuous and dichotomous data: The effects of patterns of missingness and variable correlations on type I error with small samples

The purpose of this study is to investigate the effects of predictor variable correlations and patterns of missingness with dichotomous and/or continuous data in small samples when missing data is multiply imputed. Missing data of predictor variables is multiply imputed under three different multivariate models: the multivariate normal model for continuous data, the multinomial model for dichotomous data and the general location model for mixed dichotomous and continuous data. Subsequent to the multiple imputation process, Type I error rates of the regression coefficients obtained with logistic regression analysis are estimated under various conditions of correlation structure, sample size, type of data and patterns of missing data. The distributional properties of average mean, variance and correlations among the predictor variables are assessed after the multiple imputation process. ^ For continuous predictor data under the multivariate normal model, Type I error rates are generally within the nominal values with samples of size n = 100. Smaller samples of size n = 50 resulted in more conservative estimates (i.e., lower than the nominal value). Correlation and variance estimates of the original data are retained after multiple imputation with less than 50% missing continuous predictor data. For dichotomous predictor data under the multinomial model, Type I error rates are generally conservative, which in part is due to the sparseness of the data. The correlation structure for the predictor variables is not well retained on multiply-imputed data from small samples with more than 50% missing data with this model. For mixed continuous and dichotomous predictor data, the results are similar to those found under the multivariate normal model for continuous data and under the multinomial model for dichotomous data. With all data types, a fully-observed variable included with variables subject to missingness in the multiple imputation process and subsequent statistical analysis provided liberal (larger than nominal values) Type I error rates under a specific pattern of missing data. It is suggested that future studies focus on the effects of multiple imputation in multivariate settings with more realistic data characteristics and a variety of multivariate analyses, assessing both Type I error and power. ^

Border intervention by Promotores for type 2 diabetes

Purpose of the study. This study had two components. The first component of the study was the development and implementation of an infrastructure that integrated Promotores who teach diabetes self-management into a community clinic. The second component was a six-month randomized clinical trial (RCT) designed to test the effectiveness of the Promotores in changing knowledge, beliefs, and HbA1c levels among Mexican American patients with type 2 diabetes. ^ Methods. Starfield's adaptation of the Donbedian structure, process, and outcome methodology was used to develop a clinic infrastructure that allowed the integration of Promotores as diabetes educators. The RCT of the culturally sensitive Promotores-led 10-week diabetes self-management program compared the outcomes of 63 patients in the intervention group with 68 patients in a wait-list, usual care control group. Participants were Mexican Americans, at least 18 years of age, with type 2 diabetes, who were patients at a Federally Qualified Health Center on the Texas-Mexico border. At baseline, three months, and six months, data were collected using the Diabetes Knowledge Questionnaire (DKQ, the Health Beliefs Questionnaire (HBQ, and HbA1c levels were drawn by the clinic laboratory. A mixed model methodology was used to analyze the data. ^ Results. The infrastructure to support a Promotores-led diabetes self-management course designed in concert with administration, the physicians, and the CDE, resulted in (1) employment of Promotores to teach diabetes self-management courses; (2) integration of provider and nurse oversight of course design and implementation; (3) management of Promotora training, and the development of teaching competencies and skills; (4) coordination of care through communication and documentation policies and procedures; (5) utilization of quality control mechanisms to maintain patient safety; and (6) promotion of a culturally competent approach to the educational process. The RCT resulted in a significant improvement in the intervention group's DKQ scores over time (F [1, 129] = 4.77, p = 0.0308), and in treatment by time (F [2, 168] = 5.85, p = 0.0035). Neither the HBQ scores nor the HbA1c changed over time. However, the baseline HbA1c was 7.49, almost at the therapeutic level. The DKQ, HBQ, and HbA1c results were significantly affected by age; the DKQ and HbA1c by years with diabetes. ^ Conclusions. The clinic model provides a systematic approach to safely address the educational needs of large numbers of patients with type 2 diabetes who live in communities that suffer from a lack of health care professionals. The Promotores-led diabetes self-management course improved the knowledge of patients with diabetes and may be a culturally sensitive strategy for meeting patient educational needs. The low baseline HbA1c levels in this border community suggested that patients in this Federally Qualified Health Center on the Texas-Mexico border were experiencing good medical management of their diabetes. ^

Systematic interpretation of molecular beacon polymerase chain reaction for identifying rpoB mutations in Mycobacterium tuberculosis isolates with mixed resistant and susceptible bacteria

Detection of multidrug-resistant tuberculosis (MDR-TB), a frequent cause of treatment failure, takes 2 or more weeks to identify by culture. RIF-resistance is a hallmark of MDR-TB, and detection of mutations in the rpoB gene of Mycobacterium tuberculosis using molecular beacon probes with real-time quantitative polymerase chain reaction (qPCR) is a novel approach that takes ≤2 days. However, qPCR identification of resistant isolates, particularly for isolates with mixed RIF-susceptible and RIF-resistant bacteria, is reader dependent and limits its clinical use. The aim of this study was to develop an objective, reader-independent method to define rpoB mutants using beacon qPCR. This would facilitate the transition from a research protocol to the clinical setting, where high-throughput methods with objective interpretation are required. For this, DNAs from 107 M. tuberculosis clinical isolates with known susceptibility to RIF by culture-based methods were obtained from 2 regions where isolates have not previously been subjected to evaluation using molecular beacon qPCR: the Texas–Mexico border and Colombia. Using coded DNA specimens, mutations within an 81-bp hot spot region of rpoB were established by qPCR with 5 beacons spanning this region. Visual and mathematical approaches were used to establish whether the qPCR cycle threshold of the experimental isolate was significantly higher (mutant) compared to a reference wild-type isolate. Visual classification of the beacon qPCR required reader training for strains with a mixture of RIF-susceptible and RIF-resistant bacteria. Only then had the visual interpretation by an experienced reader had 100% sensitivity and 94.6% specificity versus RIF-resistance by culture phenotype and 98.1% sensitivity and 100% specificity versus mutations based on DNA sequence. The mathematical approach was 98% sensitive and 94.5% specific versus culture and 96.2% sensitive and 100% specific versus DNA sequence. Our findings indicate the mathematical approach has advantages over the visual reading, in that it uses a Microsoft Excel template to eliminate reader bias or inexperience, and allows objective interpretation from high-throughput analyses even in the presence of a mixture of RIF-resistant and RIF-susceptible isolates without the need for reader training.^

Design of mixed-longitudinal studies

Cross-sectional designs, longitudinal designs in which a single cohort is followed over time, and mixed-longitudinal designs in which several cohorts are followed for a shorter period are compared by their precision, potential for bias due to age, time and cohort effects, and feasibility. Mixed longitudinal studies have two advantages over longitudinal studies: isolation of time and age effects and shorter completion time. Though the advantages of mixed-longitudinal studies are clear, choosing an optimal design is difficult, especially given the number of possible combinations of the number of cohorts and number of overlapping intervals between cohorts. The purpose of this paper is to determine the optimal design for detecting differences in group growth rates.^ The type of mixed-longitudinal study appropriate for modeling both individual and group growth rates is called a "multiple-longitudinal" design. A multiple-longitudinal study typically requires uniform or simultaneous entry of subjects, who are each observed till the end of the study.^ While recommendations for designing pure-longitudinal studies have been made by Schlesselman (1973b), Lefant (1990) and Helms (1991), design recommendations for multiple-longitudinal studies have never been published. It is shown that by using power analyses to determine the minimum number of occasions per cohort and minimum number of overlapping occasions between cohorts, in conjunction with a cost model, an optimal multiple-longitudinal design can be determined. An example of systolic blood pressure values for cohorts of males and cohorts of females, ages 8 to 18 years, is given. ^