Subtyping schizophrenia

Thesis (PhD)-- Stellenbosch University, 2014.

ENGLISH ABSTRACT: Schizophrenia is a phenotypically heterogeneous disorder believed to have a strong genetic component. Limiting its clinical heterogeneity by means of subtyping may help to shed light on some of the genetic underpinnings of the disease. This study describes the application of factor analysis (FA), latent class analysis (LCA) and factor mixture modeling in a sample of 734 Xhosa-speaking schizophrenic subjects using factor analytically derived variables previously identified in an independent sample of this population. LCA was performed on the following 8 SANS and SAPS items identified by preliminary exploration of the data: eye contact, auditory hallucinations, global hallucinations score, global delusions score, grooming, affective non-responsiveness, spontaneous movement, and commenting voices. A four class model provided the best fit. Classes 1 and 2 were characterized by predominantly positive and predominantly negative symptoms, respectively, class 3 by both positive and negative symptoms and class 4 by few or absent symptoms. A history of cannabis use or abuse increased the probability of a subject being allocated to class 1, while being male made a person more likely to be included in class 2. Factor mixture modelling was performed by first using latent class analysis, then factor analysis and then the factor mixture analysis were done. The fit among these three types were then investigated. The results show that factor mixture modelling uncovered a heterogeneous latent variable structure that fits the data well with the latent classes capturing distinct positive symptom/behaviours and factors capturing severity variations. This study, the first to report on the latent class structure of schizophrenia in a sample of patients from a sub-Saharan African population, supports the universality of specific latent classes across ethnic boundaries. The results further support reports that gender, sibpair status and cannabis use may influence the phenomenology of schizophrenia. The identification of subgroups may represent an intermediate step in the search for endophenotypes of schizophrenia.

AFRIKAANSE OPSOMMING: Skisofrenie is „n psigiatriese steuring met „n heterogene fenotipe en „n vermoedelik sterk genetiese vatbaarheid. Ten einde die lig te werp op die genetiese onderbou van skisofrenie word gepoog om die kliniese heterogenisiteit te beperk deur middel van subgroepering. Hierdie studie beskryf die gebruik van latente klas analise (LKA) in „n groep van 734 Xhosa-sprekendes met skisofrenie. Die LKA word baseer op die gebruik van veranderlikes wat deur middel van faktor analise op simptome in „n onafhanklike studiegroep van Xhosa-sprekendes met skisofrenie verkry is. Die LKA is gedoen op die volgende 8 “SAPS” en “SANS” veranderlikes wat deur voorlopige ondersoek van die data ge-indentifiseer is: oogkontak, gehoorshallusinasies, globale hallusinasie telling, globale waantelling, selfversorging, affektiewe nie-responsiwiteit, spontane beweging en stemme wat kommentaar lewer. „n Vierklas oplossing het die beste passing getoon. Klas 1 en 2 is gekenmerk deur oorwegend positiewe en negatiewe simptome onderskeidelik, klas 3 het beide positiewe en negatiewe simptome gehad en klas 4 het baie min of geen simptome getoon nie. „n Geskiedenis van kannabis gebruik of misbruik het die kans verhoog dat die individue in klas 1 gevind sou word, terwyl manlike geslag as veranderlike die kanse verhoog het vir allokasie in klas 2. Faktor mengsel modelering is gedoen deur eers „n latent klas analise te voltooi, gevolg deur „n faktor analise, en laastens „n factor mengsel analise. Die passing tussen die drie analises is daarna evalueer. Faktor mengsel modelering toon „n heterogene latente klas struktuur wat voldoen aan die passingsvereistes. Die latente klasse blyk spesifieke positiewe simptome/gedrag te verteenwoordig, terwyl die factor grad van erns variasie aandui. Hierdie studie is die eerste om die latente klas struktuur van skisofrenie in „n subsahara-Afrika populasie, die Xhosa, te beskryf. Die resultate onderstreep die universialiteit van die latente struktuur van skisofrenie se simptome oor etniese grense heen. Verder ondersteun die resultate die moontlike rol van geslag, aangetaste sibstatus en kannabis gebruik in skisofrenie se fenomenologie. Die identifisering van die subgroepe mag „n intermediêre stap in die soektog vir endofenotipes van skisofrenie verteenwoordig.