Feldman and Lane (2007)138
Design: Parallel double-blind RCT
Countries: Australia, Canada, Ireland, Italy, South Africa and the UK
No. of centres: 37
No. randomised: 678
Maximum follow-up: 26
MMSE range included: 10–26
Funding: Commissioned by Novartis Pharma AG (Switzerland) | AD (DSM-IV criteria) and probable AD (NINCDS-ADRDA)
Responsible caregiver | Severe and unstable cardiac disease
Severe and obstructuive pulmonary disease
Other life-threatening conditions
Use of anticholoinergic drugs, health food supplements containing ACh precursors, putative memory enhancers or insulin
Use of psychotropic drugs, with the exception of chloral hydrate, short-acting benzodiazepines and haloperidol ≤ 3 days in succession and not < 72 hours before any efficacy assessment) | Sample attrition/dropout: 553 of 678 completed study. In total, 125 withdrew after allocation: AEs (n = 83); ECG abnormalities (n = 4); laboratory abnormalities (n = 1); withdrawn consent (n = 14); protocol violation (n = 8); treatment failure (n = 2); failure to attend (n = 7); other reasons (n = 6). Differences between groups was only on AEs (rivastigmine t.i.d. 11%; rivastigmine b.i.d. 17%; placebo 9%)
Randomisation and allocation: Randomisation procedure not described. Rivastigmine and placebo tablets were identical and the number taken was the same at each dose in all groups
Power calculation: The study sample size was determined on the basis of an estimated three-point difference between rivastigmine administered b.i.d. and placebo on the ADAS-cog, an estimated 0.4-point difference between b.i.d. and placebo on the CIBIC-plus and an increased proportion of responders with CIBIC-plus ratings of > 4 of 20% within the b.i.d. rivastigmine group (35% rivastigmine vs 15% placebo). Sample sizes of 192 per group were required. For practical reasons the sample size was chosen as 200 [(ITT) population]. An individual power of 90% guaranteed protection of the global power in view of the requirement that both ADAS-cog and CIBIC-plus analyses should be significant at the 0.0499 level | Therapy common to all participants: None
Study funding: Commissioned by Novartis Pharma AG (Switzerland)
Other conflicts: HF has received honoraria for consulting, advisory boards and for participation in CME programmes sponsored by Novartis. HF has also received grant-in-aid funding for research from Novartis. RL is an employee of Novartis. The study was commissioned by Novartis Pharma AG in Switzerland |
Mowla et al. (2007)139
Design: Parallel double-blind RCT
Country: Not reported. Lead author based in the Islamic Republic of Iran
No. of centres: Not reported
No. randomised: 122
Maximum follow-up: 12
MMSE range included: 10–24
Funding: Shiraz University of Medical Sciences | AD (DSM-IV criteria)
Brief Cognitive Rating Score mean 3–5
Hachinski Ischaemic Score < 4
Adequate level of premorbid intelligence (IG > 80, global assessment) | Dementia of other aetiology
Severe organic disease (tumours, severe infectious disease, brain trauma, epilepsy, cerebrovascular malformations, alcohol or drug abuse)
Other psychiatric disorders (Hamilton Depression Scale, 17-item version, total score < 10) | Sample attrition/dropout: 98 of 122 completed study. Dropouts: rivastigmine arm n = 7; fluoxetine plus rivastigmine n = 9; placebo n = 8. Major cause of withdrawal in fluoxetine plus rivastigmine arm was AEs; in placebo arm it was loss of efficacy
Randomisation and allocation: Computer-generated (on-site) randomisation – whether researchers were able to view randomisation sequence prior to allocation is not reported. Same number of pills for all trial arms, but appearance of these pills not reported (simply described as ‘similar’)
Power calculation: Not reported | Therapy common to all participants: Single-blind placebo 6-week run-in period to exclude placebo responders
Study funding: Shiraz University of Medical Sciences
Other conflicts: Not reported
Notes: 12-week mean MMSE/WMS/ ADL/HAM scores in the fluoxetine plus rivastigmine arm were much lower than in the other arms – potential error? |
Winblad et al. (2007)140
Design: Parallel double-blind RCT
Countries: Chile, Czech Republic, Denmark, Finland, Germany, Guatemala, Israel, Italy, Republic of Korea, Mexico, Norway, Peru, Poland, Portugal, Russian Federation, Slovak Republic, Sweden, Taiwan (Province of China), the USA, Uruguay and Venezuela
No. of centres: 100
No. randomised: 1195
Maximum follow-up: 24
MMSE range included: 10–20
Funding: Novartis Pharma AG, Basel, Switzerland | AD (DSM-IV criteria) and probable AD (NINCDS-ADRDA criteria) [brain scan (MRI or CT) used for establishing these criteria must have been done within 1-year prior to randomisation]
Age 50–85 years
Living with someone in the community or, if living alone, in daily contact with a responsible caregiver | Advanced, severe, progressive, or unstable disease of any type that could interfere with study assessments or put the patient at special risk
Any condition other than AD that could explain the dementia
Use of any investigational drugs, new psychotropic or dopaminergic agents, cholinesterase inhibitors or anticholinergic agents during the 4 weeks prior to randomisation | Sample attrition/dropout: 970 of 1195 patients completed study. Reasons for dropout: AEs, withdrawn consent, lost to follow-up, death and unsatisfactory therapeutic effect. No difference between groups
Randomisation and allocation: Automated random assignment of treatment using an interactive voice-response system. Blocking was done on a study centre basis. All personnel directly involved in the conduct of the study remained unaware of the active treatment groups until all data had been retrieved and finalised for analysis
Appearance of tablets, patches and placebo not reported
Power calculation: In previous placebo-controlled trials of the rivastigmine capsule in patients with AD, a treatment difference to placebo in the ADAS-cog change from baseline of approximately 2.5 points was observed in the ITT analysis. In the current trial, a non-inferiority margin was predefined as 1.25 points on the ADAS-cog to preserve 50% of this effect, which was considered the smallest value that could represent a clinically meaningful difference. To determine the power of this study, the assumptions on delta (difference in means) and SD for the change in ADAS-cog and ADCS-CGIC from baseline were based on 24-week data from the rivastigmine capsule studies that used the ADAS-cog and CIBIC-plus. The ADCS-CGIC scale is comparable to the CIBIC-plus, which was used in previous rivastigmine capsule studies. To ensure that the study had adequate power, 1040 evaluable patients were needed. In order to reach an overall power of 80% for all of the first three hypotheses (which is defined as the product of the individual powers), the sample size was 260 patients per treatment group | Therapy common to all participants: None reported
Study funding: Novartis Pharma AG, Basel, Switzerland
Other conflicts: Three co-authors (SZ, JN, RL) are employees of Novartis. Remaining authors were investigators (BW, NA, GG, MO, CS) and/or Study Publication Committee members (BW, JC, NA, GG, MO, SZ, JN, RL). BW, JC, NA, GG, MO and CS have provided consultation services to many pharmaceutical companies that develop dementia drugs, including Novartis. A writing committee prepared an initial draft of the manuscript, based on a report provided by Novartis, and all authors contributed to its finalisation through interactive review
Data were collected by investigators and co-investigators, entered into a central database using electronic data-capture software and analysed by Novartis Pharma AG, which vouches for the data and the analysis |
