Exploring the factor structure of the mini-ICF-APP in an inpatient clinical sample, according to the psychiatric diagnosis

Abstract

Introduction

Psychosocial functioning is a key factor determining prognosis, severity, impairment and quality of life in people who have a mental disorder. The mini-ICF-APP was developed to provide a standardised classification of functioning and disability. However, despite its gaining popularity little is known about its structure and performance. This paper examines the structure of the mini-ICF-APP using factor analysis techniques.

Materials and methods

In a clinical sample of 3178 patients, with psychiatric diagnoses from several ICD-10 categories, we analysed internal consistency, item inter-correlations and the factorial structure of the data, with reference to ICD-10 diagnostic categories; Neurocognitive Disorders; Alcohol Use Disorders; Substance Use Disorders; Schizophrenia and Psychotic Disorders; Bipolar Disorder; Major Depressive Disorder; Anxiety Disorders; Personality Disorders; and Neurodevelopmental Disorders.

Results

We found good internal consistency and item inter-correlations (Cronbach alpha = 0.92) for the mini-ICF-APP. We were able to identify pivotal domains (flexibility, assertiveness and intimate relationships), which demonstrate sub-threshold influences on other domains. The factor analysis yielded a one-factor model as ideal for the whole sample and for all diagnostic categories. For some diagnostic categories the data suggested a two or three-factor model, however, with poorer fit indices.

Conclusions

The factor structure of the mini-ICF-APP appears to modify according to the main diagnosis. However, a one-factor model demonstrates better fit regardless of diagnostic category. Consequently, we consider the mini-ICF-APP to be a trans-diagnostic measurement instrument for the assessment and grading of psychosocial functioning. The use of the mini-ICF-APP sum score seems to best reflect the degree of impairment in an individual, even taking into account that affected domains may lead to sub-threshold effects on other domains.