Psychometric properties and accuracy of the European Portuguese version of the Pittsburgh Sleep Quality Index in clinical and non-clinical samples

Abstract

To assess the reliability and validity concerning the formal European Portuguese version of the Pittsburgh Sleep Quality Index [PSQI (EP)], its accuracy, and optimal cut-off point. N = 564 volunteers (18–80 years old) recruited in several settings (e.g., university campuses; work place; home; sleep consultations), agreed to complete the PSQI (EP). Subgroups completed additional measures: Insomnia Severity Index (ISI), STOP-Bang, Glasgow Sleep Effort Scale, or responded to a supplementary question about perceived sleep problems. As to internal consistency, Cronbach’s α = 0.75. Principal component analysis revealed a unidimensional structure. Six PSQI (EP) components and total scores were able to discriminate individuals who did versus did not describe having any sleep problem; all PSQI (EP) scores were significantly higher (denoting poorer quality) in participants suffering from a sleep disorder. Most Cohen’s d values showed large magnitude associations. PSQI (EP) and ISI scores were highly correlated, but no significant correlations were found considering STOP-Bang. ROC analysis confirmed an optimal cut-off point > 5 of the PSQI (EP) to detect self-reported poor/good sleepers in non-clinical settings. To discriminate non-clinical from clinical sleep patients, the optimal cutoff was > 7, and AUC = 0.94. The European Portuguese version of the PSQI performs as a reliable, valid, and accurate measure of overall sleep quality in Portuguese participants. Furthermore, results suggest that PSQI (EP) can discriminate poor sleepers in non-clinical settings, in addition to demonstrating high clinical accuracy in signaling potential sleep-disorder cases. In conclusion, the PSQI (EP) is a suitable tool to assess general sleep quality in Portuguese participants, both for clinical or non-clinical applications.