To assess the construct validity of
the Patient Global Impression–Improvement (PGI-I) questionnaire in
female subjects with SUI.
METHODS: 553 women 18-65 years of age were recruited into a
pharmacological intervention trial for SUI. Patients had to report SUI as
a predominant symptom, with 4 or more incontinent episodes per week, and
deny enuresis or urge incontinence as a predominant symptom. Following a
2-week placebo lead-in, patients were randomized to receive either
duloxetine hydrochloride or placebo for 12 weeks in this double-blind
trial. Baseline and outcome measures included a short-term, fixed bladder
volume (400mL) stress pad test (SPT), incontinence episode frequency (IEF;
obtained from real-time diaries), and the previously validated 22-item
condition specific Incontinence Quality of Life (I-QOL1) questionnaire.
Total scores on I-QOL range between 0 and 100 with 0 being the worst
possible and 100 being the best possible quality of life. Post-baseline
subjects also completed PGI-I, a single item questionnaire. The question
and responses were based on a psychometrically validated question used in
psychopharmacology research.2 The question read,"Check the one number
which best describes how your urinary tract condition is now, compared to
how it was before you began taking medication in this study".
Response options, ranging from 1 to 7, are noted in the table. Construct
validity of the PGI-I was assessed by correlating the PGI-I response
category to the median percent changes from baseline to post-baseline in
IEF and SPT and the mean absolute changes from baseline to post-baseline
in I-QOL at the final study visit utilizing the non-parametric Spearman’s
correlation coefficient (rho).
RESULTS: The table lists the outcome data by PGI-I response
category and shows the highly significant correlations observed between
the PGI-I responses and the
three independent measures of improvement in SUI.
CONCLUSIONS: PGI-I responses were
correlated significantly with changes in IEF, SPT and I-QOL measures,
establishing the construct validity of this single question measure of