Analyzing a randomized trial on breast self-examination with noncompliance and missing outcomes

Recently, instrumental variables methods have been used to address non-compliance in randomized experiments. Complicating such analyses is often the presence of missing data. The standard model for missing data, missing at random (MAR), has some unattractive features in this context. In this paper we compare MAR-based estimates of the complier average causal effect (CACE) with an estimator based on an alternative, nonignorable model for the missing data process, developed by Frangakis and Rubin (1999, Biometrika, 86, 365-379). We also introduce a new missing data model that, like the Frangakis-Rubin model, is specially suited for models with instrumental variables, but makes different substantive assumptions. We analyze these issues in the context of a randomized trial of breast self-examination (BSE). In the study two methods of teaching BSE, consisting of either mailed information about BSE (the standard treatment) or the attendance of a course involving theoretical and practical sessions (the new treatment), were compared with the aim of assessing whether teaching programs could increase BSE practice and improve examination skills. The study was affected by the two sources of bias mentioned above: only 55% of women assigned to receive the new treatment complied with their assignment and 35% of the women did not respond to the post-test questionnaire. Comparing the causal estimand of the new treatment using the MAR, Frangakis-Rubin, and our new approach, the results suggest that for these data the MAR assumption appears least plausible, and that the new model appears most plausible among the three choices.     

Postcanine dental form in the mustelidae and viverridae (Carnivora: Mammalia)

This study investigates whether the gross morphology of mustelid and viverrid postcanine dentitions corresponds with differences in diet. For each species, the predominant foods ingested are used to form predictions of dental form and measurements of the carnassial and molar teeth determine the extent of shearing and crushing surfaces on the postcanine teeth. Principal components analysis distinguishes species according to morphological differences in the dentition and these differences are compared with predictions of dental form based on diet. Dietarily specialized species are more likely to be correspondingly specialized in the dentition and species with varied food sources are more likely to possess dental characteristics that are generalized in function. Consumers of foods with high fracture resistance, such as vertebrate tissue and hard-surfaced invertebrates, possess specialized shearing or crushing postcanine teeth. On the other hand, species that consume foods of lesser fracture resistance, such as fruit and soft invertebrates, differ greatly in dental form and are more generalized in dental function. A few species possess postcanine dentitions that do not correspond with diet; the absence of dental-dietary correlation in these species suggests that other factors, such as phylogeny, are important in determining dental form.     

Dynamic conditionally linear mixed models for longitudinal data

We develop a new class of models, dynamic conditionally linear mixed models, for longitudinal data by decomposing the within-subject covariance matrix using a special Cholesky decomposition. Here 'dynamic' means using past responses as covariates and 'conditional linearity' means that parameters entering the model linearly may be random, but nonlinear parameters are nonrandom. This setup offers several advantages and is surprisingly similar to models obtained from the first-order linearization method applied to nonlinear mixed models. First, it allows for flexible and computationally tractable models that include a wide array of covariance structures; these structures may depend on covariates and hence may differ across subjects. This class of models includes, e.g., all standard linear mixed models, antedependence models, and Vonesh-Carter models. Second, it guarantees the fitted marginal covariance matrix of the data is positive definite. We develop methods for Bayesian inference and motivate the usefulness of these models using a series of longitudinal depression studies for which the features of these new models are well suited.     

Analysis of failure time data with dependent interval censoring

This article develops a method for the analysis of screening data for which the chance of being screened is dependent on the event of interest (informative censoring). Because not all subjects make all screening visits, the data on the failure of interest is interval censored. We propose a model that will properly adjust for the dependence to obtain an unbiased estimate of the nonparametric failure time function, and we provide an extension for applying the method for estimation of the regression parameters from a (discrete time) proportional hazards regression model. The method is applied on a data set from an observational study of cytomegalovirus shedding in a population of HIV-infected subjects who participated in a trial conducted by the AIDS Clinical Trials Group.     

Survival analysis using auxiliary variables via multiple imputation, with application to AIDS clinical trial data

We develop an approach, based on multiple imputation, to using auxiliary variables to recover information from censored observations in survival analysis. We apply the approach to data from an AIDS clinical trial comparing ZDV and placebo, in which CD4 count is the time-dependent auxiliary variable. To facilitate imputation, a joint model is developed for the data, which includes a hierarchical change-point model for CD4 counts and a time-dependent proportional hazards model for the time to AIDS. Markov chain Monte Carlo methods are used to multiply impute event times for censored cases. The augmented data are then analyzed and the results combined using standard multiple-imputation techniques. A comparison of our multiple-imputation approach to simply analyzing the observed data indicates that multiple imputation leads to a small change in the estimated effect of ZDV and smaller estimated standard errors. A sensitivity analysis suggests that the qualitative findings are reproducible under a variety of imputation models. A simulation study indicates that improved efficiency over standard analyses and partial corrections for dependent censoring can result. An issue that arises with our approach, however, is whether the analysis of primary interest and the imputation model are compatible.     

Combining complete multivariate outcomes with incomplete covariate information: a latent class approach

This work was motivated by the need to combine outcome information from a reference population with risk factor information from a screened subpopulation in a setting where the analytic goal was to study the association between risk factors and multiple binary outcomes. To achieve such an analytic goal, this article proposes a two-stage latent class procedure that first summarizes the commonalities among outcomes using a reference population sample, then analyzes the association between outcomes and risk factors. It develops a pseudo-maximum likelihood approach to estimating model parameters. The performance of the proposed method is evaluated in a simulation study and in an illustrative analysis of data from the Women's Health and Aging Study, a recent investigation of the causes and course of disability in older women. Combining information in the proposed way is found to improve both accuracy and precision in summarizing multiple categorical outcomes, which effectively diminishes ambiguity and bias in making risk factor inferences.     

Ultrastructure of developing tooth plates in the Australian lungfish, Neoceratodus forsteri (Osteichthyes: Dipnoi)

While the lungfish dentition is partially understood as far as morphology and light microscopic structure is concerned, the ultrastructure is not. Each tooth plate is associated with a dental lamina that develops from the inner layer of endodermal cells that form the oral epithelium. Dentines, bone and cartilage of the jaws differentiate from mesenchyme cells aggregating beneath the oral endothelium. Enamel, in the developing and in the mature form, has similarities to that of other early vertebrates, but unusual characters appear as development proceeds. Ameloblasts are capable of secreting enamel, and, with mononuclear osteoclasts, of remodelling the bone below the tooth plate. The forms of dentine, all based largely on an extracellular matrix of collagen and mineralised with biological apatite, differ from each other and from the underlying bone in the ultrastructure of associated cells and in the mineralised extracellular matrices produced. Cell processes emerging from the odontoblasts and from the osteoblasts vary in length, degree of branching and of anastomoses between the processes, although all of the cell types have large amounts of rough endoplasmic reticulum. Mineralisation of the extracellular matrices varies among the enamel, dentines and bone in the tooth plate. In addition, the development of the hard tissues of the tooth plates indicates that many of the similarities in fine structure of the dentition in lungfish, to tissues in other fish and amphibia, apparent early in development, disappear as the dentition matures.     

Monotremes as Pretribosphenic Mammals

An abundance of evidence points to the conclusion that monotremes are phyletically allied with pretribosphenic, rather than with tribosphenic, mammals. Monotremes do not have a tribosphenic dentition. Character analyses that apply tribosphenic cusp terminology to monotreme dentitions are implicitly limited thereby. A review of the molar dentition of living and fossil monotremes suggests that upper molars are composed of a strongly developed pretribosphenic paracone and metacone and a series of stylar cusps attached to them in a bicrescentic, or dilambdodont, fashion. The lower molars are composed of a trigonid, with a pretribosphenic protoconid, paraconid, and metaconid, and distal metacristid. The paraconid of m1 is reduced or lost. The talonid is composed of the pretribosphenic hypoconid, hypoconulid, and cristid obliqua. There is no evidence for a tribosphenic entoconid, nor for a talonid basin. There was no tribosphenic protocone. Monotremes are not related to other taxa included in Australosphenida. The dentition of Cretaceous taxa, such as Teinolophos and Steropodon, apparently still functioned by orthal mechanisms, whereas by the medial Paleocene (Monotrematum) and later (Obdurodon), monotremes appear to have accommodated a diet of soft-bodied organisms that left little trace of a mastication regime that had changed to apical wear via propalinal motion. Monotremes appear to be modern representatives of a Mesozoic radiation of pretribosphenic mammals centered largely in Gondwana, where they still reside today.     

Generalized linear mixture models for handling nonignorable dropouts in longitudinal studies

This paper presents a method for analysing longitudinal data when there are dropouts. In particular, we develop a simple method based on generalized linear mixture models for handling nonignorable dropouts for a variety of discrete and continuous outcomes. Statistical inference for the model parameters is based on a generalized estimating equations (GEE) approach (Liang and Zeger, 1986). The proposed method yields estimates of the model parameters that are valid when nonresponse is nonignorable under a variety of assumptions concerning the dropout process. Furthermore, the proposed method can be implemented using widely available statistical software. Finally, an example using data from a clinical trial of contracepting women is used to illustrate the methodology.