The factorial analysis of a quality of life scale for people addicted to drugs in methadone programs

Abstract

Background The evaluation scale of drug users Quality of life (QoL) in a substitution program with methadone, was developed with 21 items, two subscales, “family and economic situation” (11 items) and “personal satisfaction” (10 items) [1]. Concerning their reliability, in the validation study for the Portuguese population in 2005 (n=236) was obtained an Alfa of Cronbach of 0.88 and in a recent study in 2011(n=308) of 0.93 [1]. Objective To determine a scale factorial structure and its psychometric properties. Methods Methodological study. Participants – 180 drug users participated, aging an average of 41 years (SD=7.58 [24-69]), mostly men (73.3%), single (55.6%) and with children (52.8%), from 3 outpatients drug units. Data analyses - The correlation matrix of the items was evaluated through exploratory and confirmatory factorial analysis. The factorial load as well as the internal consistency estimated the dimensionality. Results In the reliability analysis with 21 items was found an Alfa of Cronbach of 0.89, all communalities >0.40, KMO=0.88 (p<0.001) and 58.42% of explained variance by 5 factors. When extracted item 18 (item-total correlation 0.16), all items assumed an item-total correlation > 0.20; alpha increased (α=0.90) and KMO increased to 0.88 maintaining the stability of the Bartlett test; communalities maintained above 0.40; total variance explained by the 5 factors increased to 60.4%, but 5 factors diverged from theoretical matrix and 10 items weighed in more than one factor. Through confirmatory analysis (excluding item 18) forcing for the 2 original scale factors, we verified that 4 items had commonalities <0.30. The total variance explained after the spin fell to 43.11% and 4 items weighed in more than one factor. We explored with 3 factors, KMO maintained in 0.88, Bartlett test remained within criteria, total variance explained after de rotation stayed 49.6%. Although this structure presents 3 items weigh in more than one factor, justifies its maintenance by underlying the theoretical model. Alpha is higher to initial (α=0.90). The extraction of any item will weaken scale consistency. The most stable structure was with 3 renamed factors: 1- Personal satisfaction and self-care (8 items) (38.5% explained variance); Social Family situation (8 items) (7.4%); 3- Socio professional and economic situation (4 items) (6.5%). The fidelity of the scale is reinforced by the internal consistency of its subscales, factor 1 α=0.85; factor 2 α=0.79; factor 3 α=0.72 and by the correlation between them (0.51-0.67; p<0.01). Conclusions Good internal consistency. Factorial analysis supported by the theoretical matrix. Good discriminant capacity by differences pointed out in some variables.