Table 8.

Reported responsiveness of GAS in included studies

First author	Year	Drug study	N	Methods and results	Quality
Cusick	2006	Yes	41	Ability to detect change overtime, and ability to detect difference in change between groups was measured with regression coefficients and effect sizes. Effect size for the weighted GAS scale: 0.55 (p = 0.036), and for the Likert scale 0,91 (p = 0.003).	+/− Doubtful design or method
Gordon	1999	No	53	GAS was the most responsive measure, with the highest effect size (1.29) and the highest relative efficiency (53.7).	+/− Doubtful design or method
Hartman	1997	No	10	Effect size statistic of 2.34; paired t-test before-after of 2.9 (df = 9, p = 0.017).	+/− Doubtful design or method
Khan	2008	No	24	Effect size 9.0, t = 10.0, Standardized response mean = 2.4	+/− Doubtful design or method
Palisano	1993	No	21	Of the 84 goals that were formulated for the study, similar information was obtained with the behavioral objective and GAS formats for 33 (39 %) of the goals, and change that could not be measured with the behavioral objective format was measured with the GAS format for 51 (61 %) of the goals. Of the 17 behavioral objectives that were not achieved, the corresponding GAS score documented progress toward the expected outcome (score of - 1) for 2 (12 %) of the goals. Of the 67 behavioral objectives that were achieved, the corresponding GAS score documented progress that exceeded he criteria for achievement of the behavioral objective (score of +1 or +2) for 49 (73 %) of the goals.	+/− Doubtful design or method
Rockwood	1993	No	45	RE = 4.5; ES = 5.0	+/− Doubtful design or method
Rockwood	1997	No	44	Relative efficiency: 7.8; Effect size: 5.11	+/− Doubtful design or method
Rockwood	2003	No	265	GAS was more responsive than other measures for functional improvement in the elderly; Effect size Cohen’s D: 7.8; SRM: 1.2; NRS: 0.58; Relative efficiency: 57.	+/− Doubtful design or method
Steenbeek	2011	No	23	Individual change score was found in 9/23 (physical), 18/23 (occupational) and 12/18 (speech), and for only one patient a change score was found in the GMFM-66	+/− Doubtful design or method
Stolee	1999	No	173	GAS ES = 3.52; Standardized response mean = 1.73; Relative efficiency = 3.14	+/− Doubtful design or method
Stolee	2012	No	90	All three measures of responsiveness indicated that GAS was able to detect meaningful change in this setting: Paired t-test: T(89) = −17.48; p <0.001, SRM = 1.85 (95 % CI 1.50–2.19), ES = 3.27	+/− Doubtful design or method
Turner-Stokes	2009	No	164	SRM: non-weighed GAS = 2.23, weighed GAS = 2.29. Effect sizes: non-weighed GAS = 3.16, weighed GAS = 3.54	+/− Doubtful design or method
Turner-Stokes	2010	Yes	90	The group was divided in responders and non-responders, based on the basis of their mean global benefit at the end of the study; across the whole sample, a change in GAS score from baseline of 6 predicted a positive response, with 52 % sensitivity, 85 % specificity, 81 % positive predictive value and 60 % negative predictive value.	+/− Doubtful design or method
Yip	1998	No	143	Standardized Response Mean was calculated for each instrument, by dividing the mean difference between post-treatment and pre-treatment status by the standard deviation of the mean change score. The SRM was 1.56 for GAS, compared with 0.89, 0.82, 0.72 and 0.54 for the Barthel, Katz, OARS-IADL, and SMMSE, respectively.	+/− Doubtful design or method