Analysis of failure time data with ordinal categories of response.

When failure times are observed, additional information concerning the type of failure is often recorded. A method which simultaneously models the failure times and additional information in the form of ordinal categories is discussed. An application to clinical trial data, in which the failure times are times of onset of headache, and the headaches are classified into the ordinal categories mild, moderate and severe, illustrates how this method may be used and how the final model can be interpreted. The continuation ratio model, which is used in this method, is described in detail.     

Linear combinations of biomarkers to improve diagnostic accuracy with three ordinal diagnostic categories

Many researchers have addressed the problem of finding the optimal linear combination of biomarkers to maximize the area under receiver operating characteristic (ROC) curves for scenarios with binary disease status. In practice, many disease processes such as Alzheimer can be naturally classified into three diagnostic categories such as normal, mild cognitive impairment and Alzheimer's disease (AD), and for such diseases the volume under the ROC surface (VUS) is the most commonly used index of diagnostic accuracy. In this article, we propose a few parametric and nonparametric approaches to address the problem of finding the optimal linear combination to maximize the VUS. We carried out simulation studies to investigate the performance of the proposed methods. We apply all of the investigated approaches to a real data set from a cohort study in early stage AD. Copyright © 2012 John Wiley & Sons, Ltd.     

The analysis of contingency tables with ordinal data: an application to monitoring antibiotic resistance

Rationalization of antibiotic therapy in the management of infectious diseases is helped by a knowledge of the patterns of sensitivity and resistance of bacteria to antibiotics and their possible changes both in time and from one hospital unit to another. In this paper we present the results regarding the sensitivities of several groups of bacteria and different Units of the S.Orsola-Malpighi Hospital of Bologna in the period 1995-1997. We apply recent methods of analysis of ordinal contingency tables that rely on stochastic ordering of the rows to test the assumption that a decrease (or increase) in sensitivity of bacteria to specific antibiotics has taken place against the alternative that no such thing has happened. In most cases the results seem to indicate an increase in sensitivity rather than what was expected, namely the opposite.     

A robust goodness-of-fit test statistic with application to ordinal regression models

We propose a goodness-of-fit test statistic for linear regression with heterogeneous variance, which is asymptotically chi-square if the given model is correct. The test statistic is computed as a quadratic form of observed minus predicted responses. We apply the method to a linear regression for an ordinal categorical response, the wheezing status of a child (no wheeze, wheeze with cold, wheeze apart from cold) as a function of maternal smoking and city of residence.     

Model-based screening by risk with application to Down's syndrome.

Screening for a disorder may be carried out by assessing the risk that an individual is affected given the values of variables whose distributions alter when the disorder is present. An optimal screening policy is obtained by identifying those individuals whose risk is greater than some cut-off value. This paper summarizes the way in which risk is derived from the likelihood ratio of being affected by the disorder, and compares three different methods of estimating the likelihood ratio, namely direct estimation, logistic regression and distribution modelling. For continuous variables that have a multivariate normal distribution, screening by risk is equivalent to the use of quadratic discrimination. The paper shows how estimates of the risk and associated detection and false positive rates can be derived for a screening policy which uses specified risk cut-offs. Screening by risk has the counter-intuitive property that as the separation in the distribution of screening variables between affected and unaffected individuals increases, the detection and false positive rates may both increase. The approach is explored using data on antenatal screening for Down's syndrome. The method of choice is model-based; the model is described and tested for goodness of fit. Complications arising from outliers and non-normality must be overcome before an appropriate assessment of risk can be made. The concept of shrinkage is used to estimate the detection of false positive rates that may be expected in a new data set.     

Marginalized models for longitudinal ordinal data with application to quality of life studies

Random effects are often used in generalized linear models to explain the serial dependence for longitudinal categorical data. Marginalized random effects models (MREMs) for the analysis of longitudinal binary data have been proposed to permit likelihood-based estimation of marginal regression parameters. In this paper, we propose a model to extend the MREM to accommodate longitudinal ordinal data. Maximum marginal likelihood estimation is proposed utilizing quasi-Newton algorithms with Monte Carlo integration of the random effects. Our approach is applied to analyze the quality of life data from a recent colorectal cancer clinical trial. Dropout occurs at a high rate and is often due to tumor progression or death. To deal with events due to progression/death, we used a mixture model for the joint distribution of longitudinal measures and progression/death times and use principal stratification to draw causal inferences about survivors.     

Measuring Worksite Health Promotion Programs: an application of Structural Equation Modeling with ordinal data

This paper presents a model for measuring the outcome of Worksite Health Promotion Programs through an application of Structural Equation Modeling with ordinal data. We model the function “being healthy” as a vector comprised of three latent or unobservable variables: Health Status, Lifestyle and Stress. Each variable can be measured only indirectly through a set of manifest or observable ordinal indicators. The objective is to derive and analyze the distributions, and changes in distributions over time, of the latent variables on an individual level. The model is analyzed empirically on data from three large Swedish manufacturing firms.     

Variable selection in covariate dependent random partition models: an application to urinary tract infection

Lower urinary tract symptoms can indicate the presence of urinary tract infection (UTI), a condition that if it becomes chronic requires expensive and time consuming care as well as leading to reduced quality of life. Detecting the presence and gravity of an infection from the earliest symptoms is then highly valuable. Typically, white blood cell (WBC) count measured in a sample of urine is used to assess UTI. We consider clinical data from 1341 patients in their first visit in which UTI (i.e. WBC ) is diagnosed. In addition, for each patient, a clinical profile of 34 symptoms was recorded. In this paper, we propose a Bayesian nonparametric regression model based on the Dirichlet process prior aimed at providing the clinicians with a meaningful clustering of the patients based on both the WBC (response variable) and possible patterns within the symptoms profiles (covariates). This is achieved by assuming a probability model for the symptoms as well as for the response variable. To identify the symptoms most associated to UTI, we specify a spike and slab base measure for the regression coefficients: this induces dependence of symptoms selection on cluster assignment. Posterior inference is performed through Markov Chain Monte Carlo methods. Copyright © 2015 John Wiley & Sons, Ltd.     

The value of explicitly emulating a target trial when using real world evidence: an application to colorectal cancer screening

Observational analyses for causal inference often rely on real world data collected for purposes other than research. A frequent goal of these observational analyses is to use the data to emulate a hypothetical randomized experiment, i.e., the target trial, that mimics the design features of a true experiment, including a clear definition of time zero with synchronization of treatment assignment and determination of eligibility. We review a recent observational analysis that explicitly emulated a target trial of screening colonoscopy using insurance claims from U.S. Medicare. We then compare this explicit emulation with alternative, simpler observational analyses that do not synchronize treatment assignment and eligibility determination at time zero and/or do not allow for repeated eligibility. This empirical comparison suggests that lack of an explicit emulation of the target trial leads to biased estimates, and shows that allowing for repeated eligibility increases the statistical efficiency of the estimates.     

HIV infection among women in prison: an assessment of risk factors using a nonnominal methodology.

The relative contributions of needle use practices and sexual behaviors to human immunodeficiency virus (HIV) antibody seropositivity among 394 women incarcerated in Quebec were determined by risk factor assessment and serology with a nonnominal methodology. HIV positivity was found in 6.9% (95% confidence interval [CI] = 4.6, 9.9) of all participants and in 13% (95% CI = 8.6, 18.6) of women with a history of injection drug use. HIV seropositivity among women with a history of injection drug use was predicted by sexual or needle contact with a seropositive person, self-reported genital herpes, and having had a regular sexual partner who injected drugs, but it was not predicted by prostitution. Nonnominal testing is an ethical alternative to mandatory and anonymous unlinked testing among correctional populations.     

Epidemiologic evaluation of screening for risk factors: application to genetic screening.

To assess the usefulness of screening for risk factors, we derived arithmetic relationships between screening parameters (sensitivity, specificity, and positive predictive value PPV) and risk factor frequency, disease frequency and relative risk. We evaluated these relationships in the special case of genetic markers and disease susceptibility. It can be shown that even in the face of very large relative risks, sensitivity and positive predictive value are affected by the relative magnitude of disease and genetic marker frequencies. When the genetic marker is less frequent than the disease, PPV increases with increasing relative risk but sensitivity remains low. When the genetic marker is more frequent than the disease, sensitivity increases with increasing relative risk but PPV remains low. When marker and disease frequencies are equal, both PPV and sensitivity increase with increasing relative risks, but very high relative risks (greater than 100) have to be obtained for rare diseases. Depending on the goals of the screening program, these relationships can be used to predict the relative magnitudes of false positives (low PPV) and false negatives (low sensitivity). This approach can be generalized to evaluate nongenetic risk factors in screening programs as well.     

The worth of a screening program: an application of a statistical decision model for the benefit evaluation of screening projects.

A statistical decision model is applied to the benefit evaluation of screening projects to derive an expression which provides upper and lower limits for average benefits in terms of prevalance rates of screen positives and negatives, and the average cost of screening and referral. Possible applications of such a technique are discussed and a numerical example is given.     

Risk assessment using exposure intensity: an application to vermiculite mining.

Estimation of exposure-response relations from epidemiological data is complicated by the fact that exposures usually vary in intensity over time. Cumulative exposure indices, which do not separate the effects of intensity and duration, are commonly used to circumvent this problem. In this paper the estimation of relative risk for specific ranges of exposure intensity from such data is considered using existing statistical methods for fitting multivariate relative risk models. This has the advantage that it does not assume that exposure intensity and duration have equivalent effects on risk. It also throws light on the possible existence of a threshold. The procedure was applied to data from a cohort of 406 vermiculite miners to examine the lung cancer risk associated with exposure to fibrous tremolite, which contaminated the vermiculite. The pattern of exposure-response differed substantially from that obtained using a cumulative exposure index to assess risk.     

Development and application of a generic methodology to assess the quality of clinical guidelines.

BACKGROUND: Despite clinical guidelines penetrating every aspect of clinical practice and health policy, doubts persist over their ability to improve patient care. We have designed and tested a generic critical appraisal instrument, that assesses whether developers have minimized the biases inherent in creating guidelines, and addressed the requirements for effective implementation. DESIGN: Thirty-seven items describing suggested predictors of guideline quality were grouped into three dimensions covering the rigour of development, clarity of presentation (including the context and content) and implementation issues. The ease of use, reliability and validity of the instrument was tested on a national sample of guidelines for the management of asthma, breast cancer, depression and coronary heart disease, with 120 appraisers. A numerical score was derived to allow comparison of guidelines within and between diseases. RESULTS: The instrument has acceptable reliability (Cronbach's alpha coefficient, 0.68-0.84; intra-class correlation coefficient, (0.82-0.90)). The results provided some evidence of validity (Pearson's correlation coefficient between appraisers' dimension scores and their global assessment was 0.49 for dimension one, 0.63 for dimension two and 0.40 for dimension three). The instrument could differentiate between national and local guidelines and was easy to apply. There was variation in the performance of guidelines with most not achieving a majority of criteria in each dimension. CONCLUSIONS: Use of this instrument should encourage developers to create guidelines that reflect relevant research evidence more accurately. Potential users or groups adapting guidelines for local use could apply the instrument to help decide which one to follow. The National Health Service Executive is using the instrument to assist in deciding which guidelines to recommend to the UK National Health Service. This methodology forms the basis of a common approach to assessing guideline quality in Europe.     

Stochastic league tables: an application to diabetes interventions in the Netherlands

The aim of this paper is to discuss the use of stochastic league tables approach in cost-effectiveness analysis of diabetes interventions. It addresses the common grounds and differences with other methods of presenting uncertainty to decision-makers. This comparison uses the cost-effectiveness results of medical guidelines for Dutch diabetes type 2 patients in primary and secondary care. Stochastic league tables define the optimum expansion pathway as compared to baseline, starting with the least costly and most cost-effective intervention mix. Multi-intervention cost-effectiveness acceptability curves are used as a way to represent uncertainty information on the cost-effectiveness of single interventions as compared to a single alternative. The stochastic league table for diabetes interventions shows that in case of low budgets treatment of secondary care patients is the most likely optimum choice. Current care options of diabetes complications are shown to be inefficient compared to guidelines treatment. With more resources available one may implement all guidelines and improve efficiency. The stochastic league table approach and multi-intervention cost-effectiveness acceptability curves in uncertainty analysis lead to similar results. In addition, the stochastic league table approach provides policy makers with information on affordability by budget level. It fulfils more adequately the information requirements to choose between interventions, using the efficiency criterion.