Clinical Psychometrics

Clinical Psychometrics

Per Bech

A John Wiley & Sons, Ltd., Publication

This edition first published 2012 © 2012 by John Wiley & Sons, Ltd Danish original title: Klinisk psykometri, by Per Bech, ISBN 97887628-1011-2, copyright Munksgaard Danmark, Copenhagen 2011. This edition of Klinisk psykometri is published with the title “Clinical Psychometrics”, by arrangement with Munksgaard Danmark. Wiley-Blackwell is an imprint of John Wiley & Sons, formed by the merger of Wiley’s global Scientific, Technical and Medical business with Blackwell Publishing. Registered Office John Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK Editorial Offices 9600 Garsington Road, Oxford, OX4 2DQ, UK 111 River Street, Hoboken, NJ 07030-5774, USA For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell The right of the author to be identified as the author of this work has been asserted in accordance with the UK Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. The contents of this work are intended to further general scientific research, understanding, and discussion only and are not intended and should not be relied upon as recommending or promoting a specific method, diagnosis, or treatment by physicians for any particular patient. The publisher and the author make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of fitness for a particular purpose. In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of medicines, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each medicine, equipment, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. Readers should consult with a specialist where appropriate. The fact that an organization or Website is referred to in this work as a citation and/or a potential source of further information does not mean that the author or the publisher endorses the information the organization or Website may provide or recommendations it may make. Further, readers should be aware that Internet Websites listed in this work may have changed or disappeared between when this work was written and when it is read. No warranty may be created or extended by any promotional statements for this work. Neither the publisher nor the author shall be liable for any damages arising herefrom. Library of Congress Cataloging-in-Publication Data Bech, Per. [Klinisk psykometri. English] Clinical psychometrics / Per Bech. – 1st ed. p. ; cm. Includes bibliographical references and index. ISBN 978-1-118-32978-8 (pbk. : alk. paper) 1. Psychometrics. 2. Psychiatry. I. Title. [DNLM: 1. Psychometrics–history. 2. Factor Analysis, Statistical. 3. Psychology, Clinical– instrumentation. 4. Psychopharmacology. BF 39] BF39.B417 2012 150.1′5195–dc23

2012009839 A catalogue record for this book is available from the British Library. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Cover image: © Todd Harrison – iStockphoto.com Cover design by Sarah Dickinson

Set in 9.5/12pt Minion by SPi Publisher Services, Pondicherry, India

1 2012

http://www.wiley.com/wiley-blackwell

I attempted to effect the scientific in my psychopathology by methodological investigations, not by a dogmatic exposition of a complete psychiatric epistemology.

Karl Jaspers (1950)

The debt of psychiatry to the psychologist is now great and growing. From [Eysenck’s] rigorous inquiries, sustained and resourcefully developed over years, psychiatry stands to gain an impetus and accuracy in some essential matters which will advance it and reinforce the free play of clinical skill and insight.

Aubrey Lewis (1952)

Emil Kraepelin is probably the most outstanding psychiatrist who ever lived. Max Hamilton (1978)

To Ole Rafaelsen, a man larger than life, and to Erling Dein who showed me how to use Occam’s razor in psychopathology

About the author, ix Preface, x

Introduction, 1

1. Classical psychometrics, 3

Emil Kraepelin: Symptom check list and pharmacopsychology, 6 Charles Spearman: Factor analysis and intelligence tests, 10 Harold Hotelling: Principal Component Analysis, 13 Hans Eysenck: Factor analysis and personality questionnaires, 15 Max Hamilton: Factor analysis and rating scales, 20 Pierre Pichot: Symptom rating scales and clinical validity, 23

2. Modern psychiatry: DSM-IV/ICD-10, 27

Focusing on reliability, 27 Focusing on validity, 28 Quantitative, dimensional diagnosis, 29

3. Modern dimensional psychometrics, 32

Ronald A. Fisher: From Galton’s pioneer work to the suffi cient statistic, 32 Georg Rasch: From Guttman’s pioneer work to item response theory

analysis (IRT), 34 Sidney Siegel: Non-parametric statistics, 38 Robert J. Mokken: Non-parametric analysis for item response

theory (IRT), 39

4. Modern psychometrics: Item categories and suffi cient statistics, 43

Rensis Likert: Scale step measurements, 43 John Overall: Brief, suffi cient rating scales, 45

Contents

vii

viii Contents

Clinical versus psychometric validity, 48 Item-response theory versus factor analysis, 49 Jacob Cohen: Eff ect size, 50

5. The clinical consequence of IRT analyses: The pharmacopsychometric triangle, 53

Eff ect size and clinical signifi cance, 53 Th e pharmacopsychometric triangle, 56 Antidementia medication, 59 Antipsychotic medication, 60 Antimanic medication, 65 Antidepressive medication, 66 Antianxiety medication, 69 Mood stabilising medications, 72 Combination of antidepressants, 73

6. The clinical consequence of IRT analyses: Health-related quality of life, 74

Th e WHO-5 Questionnaire, 78

7. The clinical consequences of IRT analyses: The concept of stress, 82

Post-traumatic stress disorder, 82 Th e work-related stress condition, 84 Integration of Selye’s medical stress model, 85

8. Questionnaires as ‘blood tests’, 89

Population studies in depression and anxiety, 89 Th e predictive validity of WHO-5, 92 Screening scales, 92

9. Summary and perspectives, 95

10. Epilogue: Who’s carrying Einstein’s baton?, 103

Glossary, 109

Appendices, 114

References, 185

Index, 196

ix

Per Bech

Per Bech received a medical degree from Copenhagen University in 1969. In 1972 he received a gold medal award from Århus University for his thesis on the dose-response relationship between cannabis ( tetrahydrocannabinole) and various psychological measurements, including time experience and reaction time in simulated car driving.

He completed a doctorate thesis (Dr. Med. Sci) at Copenhagen University on the clinical and psychometrical validity of rating scales in depression and mania in 1981.

He was appointed Professor of Psychiatry at Odense University in 1992 and in 2008 he was appointed Professor of Applied or Clinical Psychometrics at Copenhagen University.

Since 1981 he has held the post of chief psychiatrist at The Mental Health Centre North Zealand in Hillerød (Capital Region of Denmark) and is Head of the Psychiatric Research Unit there. He is an honorary member of the Royal College of Psychiatrists and of the European Psychiatric Association (EPA).

About the author

x

The first edition of this book was the original Danish version published in January 2011, as an introduction to the very broad field covering clinical psychology, psychiatry and clinical psychopharmacology. It was an attempt to follow Kraepelin’s rating scale approach and his pharmacop sycho-metrics as they have developed in the twentieth century, especially with the introduction of psychopharmacology in the 1960s. The central concept here is the Pharmacopsychometric Triangle, in which (A) covers desired clinical effect, (B) unwanted effects, or side effects, and (C) patient-reported quality of life. In connection with (A), short psychometric scales are described which can be used to measure such classes of drugs as antide mentias, antipsychotics, antimanics, antidepressants, antianxiety drugs, and mood stabilisers.

The psychometric performances of scales for (A), (B) and (C) are described with reference to both factor analysis and to item response theory models. These models have been amended for readers without mathematical knowl-edge. However, throughout the book experienced psychiatrists are referred to as an index of validity in an attempt to bring the symptoms home to the dimensions within (A), (B) and (C) where they belong.

My thanks when preparing the Danish version of my book went, as so often before, to Peter Allerup, Professor of Theoretical Psychometrics at the University of Århus. He has been a ‘basic factor’ for my work with rating scales over nearly 40 years! My research coordinator Lone Lindberg has made a unique contribution, with invaluable help in typing and layout. Gabriele Bech-Andersen and Susan Søndergaard are behind the translation procedures for the scales in the Danish version, and Susan has translated this English version from the Danish. Ove Aaskoven has been my statistical research assistant for many years, often in collaboration with Peter Allerup. Finally, I owe a debt of thanks to the Munksgaard editors Marie Schack and Daniel R. Andersen who made helpful suggestions for the earlier Danish versions.

In this English version editor Jesper Konradsen has raised challenging queries, especially on the philosophical lines running through it, with

Preface

Preface xi

focus on the development of psychometrics from a philosophical start to mathematical aspects of measuring mental stages, to clinical validity and dose–response relationships and then back to the philosophy of Wittgenstein, which brings symptoms home to form relevant syndromes or dimensions.

1

Clinical Psychometrics, First Edition. Per Bech.© 2012 John Wiley & Sons, Ltd. Published 2012 by John Wiley & Sons, Ltd.

Clinical psychiatry has incorporated psychology as an important auxiliary subject in the same way as neuropharmacology and neuroanatomy. As a branch of medicine, clinical psychiatry has especially attempted to deter-mine the organic cause of mental disorders; and before the establishment of psychometrics, the psychological approach to patients was seen as a non-organic explanatory model for mental disorders. Freud’s psychoanalysis, in particular, was seen as a psychological explanatory model; partly because psychiatry was regarded for many years as an atypical branch of medicine due to the non-testability of the Freudian theories, which were thus without clinical validity ( 1 ).

The scientific approach to psychology launched by psychometrics has resulted in psychiatry being regarded as a clinical branch of medicine. This only took place with the 1987 publication of Feinstein’s monograph on clini-metrics ( 2 ). Finding a comprehensive overview of the role of psychometrics in clinical psychiatry has proved difficult. The following is an attempt to put this to rights.

It falls naturally to divide clinical psychometrics into two eras. The first of these, the classical era, covers the period from 1879 to 1945. It is the era of the greatest names: Wilhelm Wundt who founded psychometrics in 1879 and his two most important pupils; Kraepelin and Spearman. The modern period developed after 1945 has Eysenck, Hamilton and Pichot as the major psychom-etricians. They developed the questionnaires and rating scales archetypal of modern clinical psychometrics in the period from 1945 to the 1970s ( 3 ). From a statistical point of view, however, Francis Galton and his London psychomet-ric laboratory (founded in 1884) are essential elements, together with Galton’s two most important ‘students’ (Pearson and Fisher) and the three people (Rasch, Siegel and Mokken) who developed the psychometric analyses that are

Introduction

2 Clinical Psychometrics

archetypal of modern clinical psychometrics in the period from 1945 to the 1970s ( 4 ) (see Figure  I.1 ).

The most obvious impact of modern psychometric research, which has resulted in short valid rating scales and the descriptive statistics of effect sizes, is the pharmacopsychometric triangle. It was the revolution in phar-macology 50 years ago that led to the rebirth of Kraepelin’s pharmacopsy-chology, now crystallised in the pharmacopsychometric triangle, the major focus of this book.

Psychometrics

WundtLeipzig (1879–1904)

GaltonLondon (1884)

Kraepelin Spearman (1904)Factor analysis

(1883)DSM III/IV

ICD-10(1994)

(1892)Pharmaco-psychology

Pichot(1974)

Eysenck(1953)

Hamilton(1967)

Pearson (1911)The grammar of science

Fisher (1922)Sufficient statistic

Siegel (1956)Nonparametric

statistic

Item scoreLikert (1932)

Anchoring points

Rasch (1960)IRT

Total scoreCattell (1973)Transferability

Hotelling (1933)PCA

Figure I.1 Psychometrics

3

Clinical Psychometrics, First Edition. Per Bech.© 2012 John Wiley & Sons, Ltd. Published 2012 by John Wiley & Sons, Ltd.

More than a century ago, psychology was defined as the science of human mental manifestations and phenomena. However, it was psychometrics (the science of measuring these mental manifestations and phenomena) that made psychology scientific. Thus, psychometrics is a purely psychological area of research.

From a historical point of view, psychology branched out from philosophy as an independent university discipline at the close of the nineteenth century. It all started in Leipzig in 1879. Here the philosopher Wilhelm Wundt (1832–1920) established his psychological laboratory at the university. Formally, however, his laboratory remained under the faculty of philosophy. Wundt succeeded in detaching psychology from philosophy, especially freeing it from the influence of Emanuel Kant, an extremely influential philosopher who stated that it is impossible to measure manifestations of the mind in the same way as physical objects ( 5 ). With his criticism of pure reason, Kant (1724–1804) established the very important distinction between ‘the essential nature of things’ (things in themselves) and ‘things as they seem’ (i.e., that which we sense or perceive as a phenomenon when faced with the object we are examining).

Figure  1.1 illustrates Kant ’ s philosophical approaches with reference to present day psychiatry, according to which depression is understood to be a clinical phenomenological perception (shared phenomenology of depressive symptoms) as measured by the six depression symptoms contained in the Hamilton Depression Scale (HAM-D 6 , see Figure  3.1). Modern neuropsy-chiatry attempts to describe the depression behind the phenomenological perception, i.e., depression ‘in itself ’, as we believe it to be present in the brain, for example, as a serotonin 1A receptor problem (impairment).

The area of research now known as brain research is just such an attempt to measure the processes presumed to be taking place in the brain, that is ‘das

1 Classical psychometrics

4 Clinical Psychometrics

Ding an sich’. As pointed out by Sontag, reality has increasingly grown to resemble what the camera shows us ( 6 ). It is reality itself when the neuropsy-chiatric camera demonstrates receptor binding in the brain, while clinical reality is increasingly becoming what the camera visualises for us by means of assessment scales or patient-related questionnaires.

The ability to describe reality as it is in itself, i.e., looking at the world unclouded by any preconception of it, has been debated by such neo-Kantentians as Wittgenstein and Quine ( 7 ). The quantification of endophenotypes or deep phenotypes is probably the most scientific image of the world. However, we do not have endophenotypes to tell us whether we indeed can describe reality, e.g., the brain, as it is itself. Wittgenstein tells us that he does not want to say whether we can or cannot describe reality as it is in itself. He wanted, as stated by Putman to bring our phenomenological items back to their home in clinical psychiatry. This is what clinical psychometrics is about ( 7 ).

Figure  1.2 shows a correlation between the so-called psychotic symp-tom items in an American rating scale (see Appendix) and serotonin 2A receptor binding, which it is now possible to measure by means of positron emission tomography (PET) scanning ( 8 ). The figure shows a correlation coefficient of −0.57; this is statistically significant but not clinically sig-nificant, as the variance on the ordinate axis (the ‘psychosis’ scale) can explain only about 32% of the variance on the axis of abscissas (serotonin 2A receptor binding). If the two patients at the far left are excluded as outliers, then the negative correlation value is halved, so that less than 10% of the variance is explained.

Kant’s philosophical approach

Psychometric frame of reference(The clinical scientist)

das Ding für uns

the phenomenon for us

Things as we perceive them in timeand space when measuring them

e.g. HAM-D6

Biological frame of reference(The brain scientist)

das Ding an sich

the noumenon

Things in themselves – onlybiological comprehension is valid

e.g. serotonin 1A receptorin the brain

Figure 1.1 The philosophical background for the emergence of psychometrics

Classical psychometrics 5

The scale in Figure  1.2 shows the positive symptoms in a schizophrenia scale. In the early 1970s, the American psychiatrist Nancy Andreasen found it important to label those schizophrenic symptoms on which medication had an effect as positive. In clinical psychiatry, these were termed productive symptoms as they were often the reason for hospitalisation in a mental insti-tution. Later on, Nancy Andreasen became interested in neuropsychiatric brain imaging methods [Computer Assisted Tomography (CAT scan), Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET)], which became available in the 1980s and 90s. However, in an inter-view from 2003, she had to admit that schizophrenia is probably not located in one specific section of the brain ( 9 ). Schizophrenia affects many different brain areas that cannot be visualised as ‘das Ding an sich’.

Wilhelm Wundt ’ s major achievement was to realise that mathematical models of ‘das Ding für uns’ can be used to measure the ‘shared pheno-menology’ of the state one wishes to assess quantitatively. During his stud-ies at the Heidelberg faculty of medicine, he obtained a degree in medicine. Wundt then participated in studies in the physiology of perception under Helmholtz (1821–94) and Fechner (1801–87). He observed that it was possible to get subjects to reliably assess sensory impressions when the conditions of the study were standardised, e.g., with increasing light or noise exposure.

Clinical assessment

Psychotic subscale (PANSS) (See Appendix)40

30

20

102.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00

Frontal 5-HT2A receptor binding in the brain (biological validity)

Figure 1.2 The problematic relationship between the clinical, the psychometrical and the biological frames of reference with a correlation coeffi cient of −0.57

6 Clinical Psychometrics

Wundt ’ s philosophical basis was that each manifestation of the mind corresponds to a neurobiological substrate in the brain, but in his opinion the psychometric measurement of this manifestation of the mind should only focus on the psychological phenomena (das Ding für sich) and not include any biological elements in any way. He belonged to the branch of philosophy called non-reductive monism (corresponding to Harald Høffding ’ s critical monism, which maintains that manifestations of the mind cannot be reduced to purely biological variables) ( 10 ). On the other hand, it is of course possible to reduce certain manifestations of the mind to less complicated ones in an attempt to obtain the most reliable or objective measure. He felt that it would be possible in this way to make psychology scientific within the frame of its own descriptive realm, since psychological and biological methods of description are two different ways of viewing reality.

Wundt ’ s approach was that of descriptive psychology where the various dimensions consisting of individual items (symptoms) can be added to give a total score. He was excluding the immediate, peak-experiences detached from relations, e.g., the spontaneous, stimulus-unrelated, perception-like images in the religious experience of the child, actually referred to as ‘Sensus numinis’ ( 11,12 ). The clearest description of Wundt ’ s scientific approach based on his ‘Grundzüge der psychologischen Psychologie’ is found in Vannerus’ monograph ( 13 ).

The psychometric method developed by Wundt is probably the only specific psychological method identified in mental science, i.e., in scientific psychology ( 14 ). The two most famous scientists to emerge from Wundt ’ s psychological laboratory in Leipzig were Emil Kraepelin and Charles Spearman; both of them understood that psychological measurement ( psychometrics) and biological measurement are two different ways of viewing nature.

Emil Kraepelin: Symptom check list and pharmacopsychology

Kraepelin (1856–1926) had just obtained his medical degree when he applied for a post at Wundt ’ s laboratory in 1882. As Wundt was unable to finance his salary, Kraepelin also had to take up a post as a locum at the local mental hospital in Leipzig. Thus, Kraepelin held an unsalaried position at the Wundt laboratory. Kraepelin ’ s purpose was to introduce scientific psychology into psychiatry so that his career as a psychiatrist would be furthered by his stud-ies at Wundt ’ s psychological laboratory. In his job application to Wundt, he wrote that he would give a kingdom for a [research] topic; Wundt then gave him the opportunity to examine the influence of psychoactive substances

Classical psychometrics 7

such as alcohol and the hypnotic drug chloral hydrate on volunteer research subjects. Kraepelin set out to demonstrate a dose response curve using reaction time measurements as the psychological response and psychoactive substances as the stimuli, so that increasing amounts of alcohol (number of drinks) led to lengthening reaction times. Since Wundt could see that Kraepelin had his heart set on psychiatry, he encouraged Kraepelin to employ this objective scientific method when subsequently assessing the various symptoms presented by patients suffering from mental disorders.

Kraepelin published his first Psychiatric Compendium as early as 1883. In  this he attempted to focus on the symptoms presented in the different disorders ( Compendium der Psychiatrie . Verlag von Amber Abel, Leipzig, 1883). After leaving the Leipzig laboratory and starting on his career as a  psychiatrist in Munich, Kraepelin published several compendiums or textbooks on psychiatry. He revised his textbook almost bi-annually and in the 6 th edition in 1899, he was able to describe two disorders with different symptom profiles: manic-depressive disorder and schizophrenia.

Figure  1.3 shows the checklist Kraepelin used when systematically moni-toring his patients over several years in order to ascertain which symptoms possessed ‘shared phenomenology’ over this period of time. These are called

Kraepelin’s symptom checklist from his Zählkarten (counting cards)

• Nervousness

• Restlessness

• Irritability

• Depression

• Psychomotor retardation

• Aggression

• Grandiosity

• Negativistic behaviour

• Hallucinations

• Paranoid ideas

Matthias M. Weber and Eric J. Engstrom Kraepelin’s ‘diagnostic cards’:the confluence of clinical research and preconceived categories.

History of Psychiatry 1997; 8: 375 – 385.

Figure 1.3 The assessment scale or checklist used by Kraepelin (10)

8 Clinical Psychometrics

checklist symptoms, as Kraepelin only determined whether the symptom was present or absent. This type of scale is called a nominal scale. Using this method, Kraepelin was able to demonstrate that during a period of about six months, some patients presented with the first five or six symptoms in Figure  1.3 , while in other episodes of shorter duration (up to three months) they had the next two symptoms (aggression and delusions of grandeur), along with restlessness, sleep disturbance and irritability. Between these episodes of depression or mania, these patients were discharged from hospital and were socially well-functioning. Other patients, who were often lifetime residents in asylums, had the last three symptoms in Figure  1.3 . Kraepelin described them as suffering from dementia praecox (now schizophrenia), as  the disorder typically started when they were about 20 years of age and was  chronic in nature, often with an influence on intellectual functions as well. But these were consequences, not elements, of the schizophrenic symptomatology. Manic-depressive disorder, on the other hand, did not typically emerge at a specific age. Based on the original registrations by Kraepelin on his ‘Zahlkarten’ (count-ing cards) including the checklist symptoms in Figure  1.3 , Jablensky et al made a comparison using the Present State Examination (PSE). From the PSE scores the ICD-9 diagnoses of schizophrenia and manic-depressive disorder can be made. In total Jablensky et al identified 721 patients assessed by Kraepelin and found a concordance for the diagnoses of schizophrenia and manic-depressive disorder of approximately 80% with the ICD-9 diagnoses ( 15 ).

In his thesis: ‘Über die Beeinflussung einfacher psychischer Vorgänge durch einige Arzneimittel‘ (Jena, Fischer Verlag 1892), Kraepelin established the area of research he designated pharmacopsychology .

In the 8 th edition of his textbook, written between, 1909–13, Kraepelin added reflections on the psychotherapeutic effects of certain drugs such as morphine, phenemal and chloral hydrate. However, he found that the effects of these drugs on schizophrenia and manic-depressive disorder were extremely poor. He was thus able to observe the spontaneous course of illness in these two disorders.

In the schizophrenic patient, as stated previously, the condition was unremitting, while manic-depressive disorder was characterised by episodes with specific symptoms and then periods between episodes of a year or more in which the patients were completely without symptoms and thus able to function normally. In these descriptions, Kraepelin determinedly avoided including the various theories on disease circulating at that time, such as hereditary elements, stress burden and so on.

Kraepelin ’ s textbooks were not widely known outside Germany, as the two world wars made German psychiatry less acceptable. His system only began to make an international impact after World War II, not least in the USA.