David A. Lane

A Nonstochastic Interpretation of Reported Significance Levels

Tests of significance are often made in situations where the standard assumptions underlying the probability calculations do not hold. As a result, the reported significance levels become difficult to interpret. This article sketches an alternative interpretation of a reported significance level, valid in considerable generality. This level locates the given data set within the spectrum of other data sets derived from the given one by an appropriate class of transformations. If the null hypothesis being tested holds, the derived data sets should be equivalent to the original one. Thus, a small reported significance level indicates an unusual data set. This development parallels that of randomization tests, but there is a crucial technical difference: our approach involves permuting observed residuals; the classical randomization approach involves permuting unobservable, or perhaps nonexistent, stochastic disturbance terms.

Artificial worlds and economics, part II

In this and a companion paper (Lane 1993), I describe a class of models, called artificial worlds (AWs), that are designed to give insight into a process called emergent hierarchical organization (EHO). This paper introduces the ideas of EHO and AWs and discusses some of the interferential problems involved in trying to learn about EHO by constructing and studying the properties of AWs. It concludes by introducing two abstract AWs that address important general problems in EHO: the relation between structure and function, and the dynamics of evolutionary processes. The companion paper will discuss several AWs expressly designed to model particular economic phenomena.

Assessing methods for causality assessment of suspected adverse drug reactions.

Reproducibility and validity as currently defined are inappropriate criteria for the evaluation of methods for causality assessment. Reproducibility leads to suppression rather than resolution of real disagreements and the method used to establish validity relies on the tarnished gold standard of expert opinion. We describe six alternative criteria that attempt to address a potential users main concerns--the need to know whether to believe the results in general and in a particular case. These criteria focus on the internal structure of a method rather than its output. When we assessed the published methods by these criteria most of the methods failed most of the criteria. We believe that the problem and its solution lie at a fundamental level--real understanding of the true nature of causality assessment, which we suggest is an inherently subjective evaluation based on the multiple uncertainties that an assessor has about a case and not an objective attribute of the drug-event connection that can be determined from unambiguous evidence elicited in response to operational questions.

Utility, decision, and quality of life

This paper introduces the notion of utility and its application to the measurement of quality of life. Utility is a general concept for measuring the value individuals attach to the consequences of various courses of action. Its purpose is to guide decision making when the consequences of the alternative decisions cannot be known with certainty. How utility can be applied to clinical decision making is illustrated by example. In addition, utility is compared to other quality of life measures with respect to its scope, its purpose and its interpretation.

Aminorex, dexfenfluramine, and primary pulmonary hypertension.

BACKGROUNDnIn the late 1960s, an epidemic of primary pulmonary hypertension (PPH) occurred in Europe shortly after the introduction of aminorex fumarate, a potent anorexigen. A recently published case-control study from Europe reported that use of other anorexigens (the most prevalent of which was dexfenfluramine) was also associated with an increased risk of PPH. This led to warnings of a repeat epidemic, especially after the introduction of dexfenfluramine on the North American market.nnnOBJECTIVEnTo compare the epidemiologic associations of PPH with aminorex and dexfenfluramine, both with respect to strength of association (estimate of relative risk) and public health impact (etiologic fraction) and thus to assess the potential for a new epidemic of PPH.nnnMETHODSnWe constructed a synthetic case-control study for aminorex based on reported case series from Berne and Basel, Switzerland, and a random population sample from Hanover, Germany, and compared the results with those recently reported for dexfenfluramine. Control rates of exposure were used to estimate population exposure prevalences and, hence, etiologic fractions.nnnRESULTSnThe estimated odds ratio (and 95% confidence interval) for the association between PPH and any exposure to aminorex was 97.8 (78.9-121.3), with a corresponding etiologic fraction of 77%. The corresponding figures for dexfenfluramine were 3.7 (1.9-7.2) and 17%, respectively.nnnCONCLUSIONnThe strong association between aminorex and PPH probably led to a 5-fold increase in PPH incidence, and thus a very noticeable epidemic. The association with dexfenfluramine would result in an increase in incidence of only 20%. Based on the available evidence, a repeat PPH epidemic seems unlikely.

Analgesic use, blood dyscrasias, and case-control pharmacoepidemiology: A critique of the International Agranulocytosis and Aplastic Anemia Study

The International Agranulocytosis and Aplastic Anemia Study (IAAAS) of analgesic-induced risks of blood dyscrasias represents the current state of the art in case-control pharmacoepidemiology. We present a conceptual framework for examining the goal, methods, and analysis of an epidemiologic study of drug risks and review the IAAAS within this framework. In our view, the new risk estimates reported by the IAAAS are not inherently more accurate than existing ones, nor have they been measured in clinically and sociodemographically relevant groups of patients over the anticipated course of therapy. Thus, the reported risks cannot be used to guide clinical or regulatory decisions concerning available treatment options for such patients. Furthermore, we believe that the IAAAS methods for selection of cases and controls, ascertainment of exposure, and data analysis may well have led to invalid estimates even for those risks that are reported. We hope that closer attention to the conceptual framework we suggest and the methodologic issues we raise will enable future case-control pharmacoepidemiologic studies to provide more useful and accurate answers to questions concerning the adverse effects of drugs.

A Bayesian assessment of idiosyncratic adverse reactions to new drugs: Guillain-Barré syndrome and zimeldine.

Etude de pharmacovigilance sur la toxicite du zimeldine, un inhibiteur de la capture de serotonine, en Suede. Une methode pour le diagnostic differentiel, le BARDI, est appliquee a neuf cas de syndrome de Guillain-Barre associes au traitement

The Bayesian differential diagnosis of neutropenia associated with antiarrhythmic agents.

Methode detude basee sur une analyse mathematique bayesienne de la survenue dune neutropenie due a 1 traitement antiarythmique. Les auteurs relient la prise dun medicament particulier avec les effets secondaires observes pendant le suivi des patients

Causal propositions in clinical research and practice

The concept of causation is central to clinical research and practice. The health science literature on causality, largely contributed by epidemiologists, has examined the population-based question of whether an exposure can cause a given health outcome. Most of this literature has focused on criteria for assessing causality, rather than attempting to define it. Moreover, the population-based approach is rather distant from the individual persons in whom causes must act, which has led to different perspectives on causality among epidemiologists and health policy markers, on the one hand, and clinical practitioners and the lay public, on the other. We attempt to bridge the gap between these perspectives by defining three probabilistic causal propositions based on the locus (individual vs population) and time frame (past vs future outcome) to which they refer, beginning with the individual in whom a health outcome has already occurred (retrodictive causal propositions, i.e. It Did) and proceeding to potential causal propositions (It Can) for populations and predictive causal propositions (It Will) for individuals or populations. We conclude by showing how attention to these distinctions may help avoid common pitfalls that can impair clinical or public health decision-making.

The Bayesian Approach to Causality Assessment: An Introduction

This paper outlines an approach to causality assessment that is based on the logic of uncertainty, Bayesian probability theory. The goal of causality assessment is taken to be the calculation of the posterior odds in favor of drug causation, given all available background and case information. There are two stages to the Bayesian approach: collecting the facts and evaluating the evidence. The evaluation proceeds by a series of probability assessments that decompose the overall causality assessment into a series of component evaluations, each of which focuses on one factor or source of information. The solutions to these component problems are then combined according to the rules of probability theory to give a solution to the overall causality assessment.