A MULTILEVEL INVESTIGATION FROM THE AESOP STUDY
Preface
This thesis is a result of research pursued at the University of Cambridge, Department of Psychiatry, during the summer of 2012. The analyses in this paper have been undertaken together with my main supervisor Dr. James Kirkbride, Sir Henry Wellcome Research Fellow at the University of Cambridge. I am grateful for all his invaluable feedback and continued support during the creation of this thesis. It has been a genuine learning experience.
I am further grateful for the encouragement and valuable comments from my supervisors Dr. Mats Bogren, senior psychiatric consultant at Skåne University Hospital Lund, and Dr. Anna Lindgren, Associate Professor at the Department of Mathematical Statistics, Lund University.
I also want to thank Professor Peter B. Jones at the University of Cambridge, whose support has been vital for my research in England, and Associate Professor Elisabet Holst at Lund University, whose support for my somewhat atypical course has never wavered.
This thesis has been written for the Medical Programme (Läkarprogrammet) as well as for a bachelor’s degree in Industrial Engineering and Management, both at Lund University. This procedure has been communicated clearly to the Faculty of Medicine as well as the Department of Mathematical Statistics. I believe that the reader will find that competences from and an interest in both these fields have been fundamental for the creation of this thesis.
Sincerely,
Fredrik Oher, Dec 2012
Content
Abstract 4
Populärvetenskaplig sammanfattning 5
1. Introduction 6
1.1. Background 6
1.2 Conceptualization of psychotic disorders 7
1.3 Etiology 8
1.4 Patophysiology and Possible Mechanisms 10
1.5 Context and Significance of Thesis 11
1.6 Objectives 12
2. Methods 14
2.1. Sample 14
2.2. Outcome data (psychotic symptom dimensions and cluster 14
2.3. Individual level sociodemographic & clinical data 15
2.4. Neighbourhood level exposures and confounders 16
2.5. Statistical analysis 16
2.5.1. Symptom dimension overview and transformation 16
2.5.2. Centre & Symptom dimension comparison 17
2.5.3. Multilevel linear regression analysis approach 17
2.5.4. Model Building 17
2.5.5 Analytical approach for symptom dimensions 18
2.5.6. Analytical approach for paranoia 19
2.6. Sensitivity analysis 19
3. Results 20
3.1 Sample characteristics 20
3.2 Distribution of symptom dimensions between centres 20
3.3. Multilevel modeling of symptom dimensions 20
3.4. Sensitivity analysis 21
3.5. Multilevel modeling of paranoia 21
3.6. Symptom dimension and paranoia differences by diagnostic category 21
4. Discussion 24
4.2 Methodological considerations 24
4.3 Meaning of findings 25
5. Appendix 28
6. References Error! Bookmark not defined.
Abstract
Background
The impact of our social environment on our mental health is important. Many have studied how the environment influences the incidence and prevalence of different psychiatric disorders. Few have investigated how our environment influences the symptom dimensions within psychiatric conditions such as for example psychotic disorders.
Methods
With data from 469 subjects presenting with first episode psychosis in southeast London and Nottingham, five main symptom dimensions were identified: mania, reality distortion, negative symptoms, depressive symptoms and disorganization. Different statistical methods were pursued to investigate the impact of the environment on these symptom dimensions.
Results
The levels of reality distortion were found to be higher in London than in Nottingham (β=0.17; 95%CI: 0.07, 0.27). Further analysis indicated that this effect might have been stronger for subjects diagnosed with non-affective psychosis. Levels of depressive symptoms were also found to be higher in London (β=0.22; 95%CI: 0.06, 0.37) while levels of disorganization where found to be higher in Nottingham (β=-0.05; 95%CI: -0.09, -0.02). No differences were observed for mania or negative symptoms.
Conclusion
The notion of higher rates of reality distortion in a more urban environment is consistent with the idea that environmental stimuli could influence our levels of positive psychotic symptoms. While more work is needed, the results in this thesis indicate that this effect might be valid primarily for non-affective psychotic disorders, suggesting possibly different etiological backgrounds for non-affective and affective disorders.
Populärvetenskaplig sammanfattning
Den sociala omgivningens påverkan på vår mentala hälsa är viktig att försöka förstå. Mycket forskning har ägnats åt att undersöka hur omgivningen påverkar antalet personer som insjuknar med olika typer av psykiatriska diagnoser. Få har dock undersökt hur omgivningen påverkar symptombilden inom psykiatriska tillstånd som till exempel psykotiska sjukdomar. Psykos är ett tillstånd som kan innefatta till exempel hallucinationer och vanföreställningar, ofta i kombination med andra psykologiska symptom.
Med hjälp av data från 469 personer med förstaepisods-psykos från sydöstra London och Nottingham kunde Demjaha et al. identifiera fem huvudsakliga symptomdimensioner inom det psykotiska sjukdomsspektrumet: mani, förvrängd verklighetsuppfattning, negativa symptom, depressiva symptom samt desorganisation. Med hjälp av olika statistiska metoder undersökte vi sedan, för samma urvalsgrupp, omgivningens påverkan på dessa fem symptomdimensioner.
Vi fann att nivåerna av förvrängd verklighetsuppfattning var högre i London än i Nottingham. En utökad analys antydde att denna effekt var starkare för de personer som diagnosticerats med en så kallad icke-affektiv psykotisk sjukdom (såsom schizofreni) jämfört med så kallad affektiv psykotisk sjukdom (såsom depression med psykos). Det fanns även tecken på ökad nivå av depressiva symptom i London och ökad nivå av desorganisation i Nottingham. Inga skillnader observerades för mani eller negativa symptom.
Det har spekulerats att stimuli från omgivningen kan påverka risken att utveckla så kallade positiva psykotiska symptom. Vårt fynd att högre nivåer av förvrängd verklighetsuppfattning kunde identifieras i en mer urban miljö (London jämfört med Nottingham) passar väl med denna idé. Resultaten i detta arbete indikerar att denna effekt främst verkar gälla för icke-affektiva psykotiskta sjukdomstillstånd, vilket skulle kunna tyda på olika etiologiska bakgrunder (uppkomstorsaker) för icke-affektiva och affektiva sjukdomstillstånd.
1. Introduction
1.1. Background
Psychosis is a descriptive psychiatric term referring to an altered condition of the mind, where delusions and divergent perceptions of reality are central features. The condition is characterized by distorted interpretations of stimuli and experiences, which can influence attention, emotional state and social life. Symptoms that might occur include hallucinations, confusion and behavioral abnormalities, such as disorganized speech and grossly disorganized or catatonic behaviour1.
Psychosis can arise in several instances, including as a core feature of a number of mental illnesses, as a result of neurological damage, through intoxication, or as a side effect to pharmaceuticals. Psychotic disorders, i.e. syndromes where psychosis is a core feature, are generally divided into affective and non-affective disorders, such as bipolar disorder and schizophrenia, respectively. In schizophrenia and similar syndromes, psychosis generally occurs episodically2. In certain personality disorders, psychosis can also occur as an accompanying, but non-defining, feature.
For individuals with schizophrenia and their families, the cost in terms of suffering and lower quality of life can be substantial. Life expectancy is 12-15 years shorter in people with schizophrenia, due to increased rate of suicides and decreased somatic health3. Costs are also considerable for society; In Sweden, it has been estimated that the cost per affected individual and year is equivalent to roughly 50,000 British Pounds (500,000 Swedish kronor). The largest part is attributed welfare support to the individual4, but medicine and hospitals costs are also considerable. In Europe, the total cost to health services and society of psychotic disorders is only eclipsed by dementia and common mental disorders,5 and few syndromes entail as much human suffering 6, 7. Understanding the causes, course and treatment of such disorders is therefore of great importance.
As a phenomenon, schizophrenia might have a long history, but the first evident accounts of the syndrome seem to have appeared as late as the beginning of the early nineteenth century. This stands in contrast to depression and mania, which reportedly already appeared in ancient texts8. During the end of the nineteenth century, the German psychiatrist Emil Kraepelin suggested a distinction in the classification of mental disorders between what he termed dementia praecox and mood disorders9. Dementia praecox, meaning “early dementia”, was an attempt to describe a syndrome that appeared early in life, and resulted in reduced psychic functioning6. Kraeplin believed that dementia praecox was a result of brain damage, an incurable process that with time would totally alter the personality of the individual9. The distinction between “early dementia” and mood disorders became the foundation underpinning modern psychiatry, broadly classifying subjects with mood disorders (such as depression) or with psychotic disorders (such as schizophrenia). Today, we note that such distinctions are not always clear and that features of the two sets of disorders can occur together (see below).
At the turn of the twentieth century, the Swiss psychiatrist Eugen Bleuler concluded that the syndrome described by Kraepelin was not a type of dementia, based upon the observation that some of his patients seemed improved over time rather than deteriorated. He also believed that manic or depressive symptoms could appear within this syndrome9. He decided to call this condition schizophrenia, meaning roughly “splitting of the mind”, intending to describe the separation of function of thinking, memory, perception, emotion, and personality.
1.2 Conceptualization of psychotic disorders
There are two principal manuals widely used to diagnose psychiatric disorders. The first is the Diagnostic and Statistical Manual for Mental Disorders (DSM) published by the American Psychiatric Association; the other is the International Classification of Diseases [ICD], published by the World Health Organization. The manuals have been revised several times since they first came out, with new versions of each (DSM-V and ICD-11, respectively), due to be published in 201310 and 201511.
With regard to the diagnosis of schizophrenia, the approaches of DSM-IV and ICD-10 are quite similar. The basis is derived from the identification of specific signs, symptoms and disabilities, as well as the course of the illness. As a simplified summary, both DSM and ICD demand the identification of at least two of the following symptoms for the duration of at least one month: delusions, prominent hallucinations, disorganized speech, disorganized or catatonic behavior, and negative symptoms. Furthermore, affective syndrome, organic etiology and drug abuse need to be cancelled out as causing factors. The major difference between the manuals lies in the duration of dysfunction. DSM demands six months of social of professional dysfunction out of which at least one month includes psychosis features while ICD only requires one month of dysfunction including psychosis features12. For an overview of a selection of psychotic disorders, see Table 1.
The distinction between schizophrenia and other psychotic conditions have changed with time and varied between different countries and cultures. The ICD and DSM manuals are attempts to standardize the diagnoses of these conditions. Nowadays, schizophrenia is regarded as the most common syndrome in the psychotic disorder group, but is sometimes (somewhat confusingly) used interchangeably with the term “psychosis”. Schizophrenia, however, is one particular constellation of psychotic symptoms which meet certain above-mentioned criteria with regard to severity and duration13.
In the beginning of the 1980s, Tim Crow and Nancy Coover Andreason separately proposed that two distinct syndromes could be discerned within the schizophrenic symptom profile14, 15. The first syndrome, referred to as the Type I or positive syndrome, included symptoms such as delusions, hallucinations, and disorganized thinking. It was called positive, because these symptoms were thought of as added to the personality. The other syndrome, called Type 2, or negative, was characterized by deficits in affective, cognitive, and social functions16.
Crow speculated that although positive and negative symptom groups could both be a part of schizophrenia, they had different etiological backgrounds, with positive symptoms thought to result from hyperdopaminergia (excess of dopamine in the brain), and negative symptoms thought to result from structural brain deficits16. Andreasen proposed that the positive and negative symptoms were opposite sides of a continuum, after demonstrating a negative correlation between the prevalence of the two symptom groups16. Today, it is generally believed that positive and negative symptoms can co-occur, although it has been observed that the two symptom groups seem to follow independent courses over time7.
Ever since the days of Kraeplin and Bleuler, there has been an ongoing debate as to whether schizophrenia represents a single disease entity or not. The DSM and ICD have traditionally followed the categorical approach, attempting to create distinctions between different mental disorders. This approach has consequently received a lot of attention during the last century. More recently, however, there has been a growing opinion that grouping mental disorders into categories is, by itself, not a satisfactory practice for effective prevention and treatment17.
As an adjunct to the categorical approach, an alternative dimensional approach has been proposed, based upon the notion that symptoms for people with schizophrenia-like disorders vary greatly and that symptoms often fit into different, replicable dimensions (as e.g. Crow and Andreason observed). Proponents of this approach argue that identifying the presence and magnitude of these dimensions within each patient is important for effective treatment, as well as the understanding of psychotic disorders. It has further been argued that evidence of subclinical experiences of psychosis in the general population support the validity of the dimensional approach17.
In 1987 Peter Liddle was the first person to use factor analysis to create symptom dimensions. Factor analysis is a mathematical method by which variables (for example symptoms) can be grouped together with other variables showing similar variance, creating a number of factors (for example symptom dimensions). Liddle presented three dimensions: psychomotor poverty (including e.g. poverty of speech and lack of spontaneous movement), disorganization (including e.g. disturbances of the form of thought), and reality distortion (including certain types of hallucinations and delusions)18. Since then, many other attempts at different divisions have been undertaken.
In a meta-analysis from 2012, researchers reviewed 39 papers that had examined the dimensional structure in patients with a broad spectrum of psychotic disorders19. They found that the majority of the studies had concluded that either four or five dimensions effectively describe the psychosis profile. The most frequently reported dimensions were the positive, negative, disorganization, and affective symptom dimensions. It is important to emphasize that dimensions does not describe particular groups of patients, but can rather be seen as a tool to describe every individual’s unique blend of symptoms.