Table 2. Change from baseline in measures of efficacy at six months, intention to treat analysis Efficacy measure Placebo Galantamine 24 mg Galantamine 32 mg Treatment difference (95% CI)* Galantamine 24 mg Galantamine 32 mg 11 item cognitive subscale of Alzheimer's disease assessment scale   No of patients 215 220 217 Mean (SE) change from baseline 2.4 (0.41)  0.5 (0.38)  0.8 (0.43) 2.9 (1.6 to 4.1) P<0.001 3.1 (1.9 to 4.4) P<0.001 No (%) with 0 points improvement 88 (41) 138 (63) 130 (60) 21.5 (12.0 to 31.0) P<0.001 19.5 (10.0 to 29.0) P<0.001 No (%) with 4 points improvement 32 (15) 64 (29) 70 (32) 14.0 (6.0 to 22.0) P<0.001 17.0 (9.0 to 25.0) P<0.001 Disability assessment for dementia score   No of patients 210 212 214 Mean (SE) change from baseline  6.0 (1.08)  3.2 (1.02)  2.5 (1.07) 2.8 (0.6 to 6.1) P=0.1 3.4 (0.1 to 6.7) P<0.05 Clinician's interview based impression of change plus caregiver input   No (%) of patients 203 206 198 <0.05§ <0.001§ 1=Much improved 0 0 0 2=Moderately improved 1 (0.5) 7 (3) 9 (5) 3=Minimally improved 32 (16) 29 (14) 39 (20) 4=No change 68 (33) 91 (44) 82 (41) 5=Minimally worsened 68 (33) 57 (28) 54 (27) 6=Moderately worsened 32 (16) 17 (8) 14 (7) 7=Much worsened 2 (1) 5 (2) 1 (1) * Difference from placebo. Negative change from baseline indicates improvement. Negative change from baseline indicates deterioration. § Van Elteren test was used to test for differences in the distribution of scores between placebo and galantamine groups. For details of the observed case analysis see table 2 on BMJ's website.