G Welch1, K Weinger, B Anderson, W H Polonsky. 1. Behavioural and Mental Health Research Division, Joslin Diabetes Center, Boston, MA 02115, USA. Garry.Welch@Joslin.Harvard.edu
Abstract
AIM: Responsiveness (sensitivity to change over time) is a key psychometric quality for an outcome measure. We examined the responsiveness of the Problem Areas In Diabetes (PAID) questionnaire, a measure of diabetes-specific emotional distress. METHODS: PAID data were obtained from seven diabetes intervention studies following a literature search that included both published papers and conference abstracts. To estimate responsiveness we used two indices: (i) a statistical test (the dependent t-test), and (ii) a commonly used effect size index (Cohen's d). RESULTS: Mean patient PAID scores improved from baseline to follow-up for all seven studies. Specifically, t-statistics ranged from t= 8.5 (P < 0.001) to t= 2.1 (P < 0.06). Effect size results ranged from 0.32 (i.e. small) for a disease management intervention to 0.65 (i.e. moderate) for an intensive medical/educational intervention. CONCLUSIONS: Despite the pilot nature of the studies, the pattern of findings provided strong support for the responsiveness of the PAID. Information on responsiveness helps clinical researchers select measures, accurately estimate sample size to ensure adequate statistical power, and prioritize outcomes to be assessed.
AIM: Responsiveness (sensitivity to change over time) is a key psychometric quality for an outcome measure. We examined the responsiveness of the Problem Areas In Diabetes (PAID) questionnaire, a measure of diabetes-specific emotional distress. METHODS: PAID data were obtained from seven diabetes intervention studies following a literature search that included both published papers and conference abstracts. To estimate responsiveness we used two indices: (i) a statistical test (the dependent t-test), and (ii) a commonly used effect size index (Cohen's d). RESULTS: Mean patient PAID scores improved from baseline to follow-up for all seven studies. Specifically, t-statistics ranged from t= 8.5 (P < 0.001) to t= 2.1 (P < 0.06). Effect size results ranged from 0.32 (i.e. small) for a disease management intervention to 0.65 (i.e. moderate) for an intensive medical/educational intervention. CONCLUSIONS: Despite the pilot nature of the studies, the pattern of findings provided strong support for the responsiveness of the PAID. Information on responsiveness helps clinical researchers select measures, accurately estimate sample size to ensure adequate statistical power, and prioritize outcomes to be assessed.
Authors: Young Sil Eom; Hwa Sun Park; Sei-Hyun Kim; Sun Mee Yang; Moon Suk Nam; Hyoung Woo Lee; Ki Young Lee; Sihoon Lee; Yeun Sun Kim; Ie Byung Park Journal: Diabetes Metab J Date: 2011-04-30 Impact factor: 5.376
Authors: Athanasia Papathanasiou; Sue Shea; Anastasios Koutsovasilis; Andreas Melidonis; Eustathios Papavasiliou; Christos Lionis Journal: Ment Health Fam Med Date: 2008-06
Authors: Lloyd Paul Aiello; Allison R Ayala; Andrew N Antoszyk; Bambi Arnold-Bush; Carl Baker; Neil M Bressler; Michael J Elman; Adam R Glassman; Lee M Jampol; Michele Melia; Jared Nielsen; Howard A Wolpert Journal: JAMA Ophthalmol Date: 2015-08 Impact factor: 7.389