Article type
Year
Abstract
Background: Alzheimer’s dementia (AD) is the most common type of dementia. However, it is unclear which cognitive enhancer is optimal for severe AD. Patient-level data from people with AD can be helpful to explore patient-level variation per treatment response. Pooling individual patient data (IPD) from multiple randomised clinical trials (RCTs) of clinical interventions is considered the ‘gold standard’ analysis.
Objectives: To examine the comparative efficacy and safety of cognitive enhancers by patient characteristics, such as AD severity and sex, and to assess treatment-by-covariate interactions through IPD network meta-analysis (NMA).
Methods: We searched for RCTs with adults comparing cognitive enhancers. The primary outcome was cognition using the Mini-Mental State Examination (MMSE), and the secondary outcome was serious adverse events (SAEs). For eligible RCTs, we requested IPD from authors, sponsors and data sharing platforms. We assessed for consistency between results from published RCTs and provided IPD. We applied an available case analysis for each study, but we plan to explore the impact of missing data through the use of informative missingness parameters in NMA. We captured reasons for missing participants and time to SAE if this was available. We conducted a 2-stage analysis: at 1st stage IPD from included studies were aggregated to study-level summary; at 2nd stage the trial parameter estimates were synthesized in a random-effects NMA. We summarized evidence using the odds ratio (OR) and mean difference (MD), respectively. In the main analysis, we used crude ORs and MDs and did not adjust for any patient characteristics. In a further analysis we included patient-level covariates with interaction terms in the model including the patient characteristics that were provided. We combined aggregated data from RCTs for which we were unable to obtain IPD. Subgroup and meta-regression NMA was undertaken for all potential effect modifiers requested from data providers, whenever data were provided.
Results: We included 108 RCTs and received IPD for 17 (16%) RCTs. Of the 17 RCTs, we were able to include 12 RCTs in our NMA with complete data. Access to IPD via proprietary sponsor-specific platforms restricted us from combining IPD in a one-stage NMA model. In most IPD, we encountered a high dropout rate (up to 72%), for which most publications used the last observation carried forward imputation method. NMA results including IPD and/or aggregate data will be presented at the Cochrane Colloquium.
Conclusions: An advantage of our IPD-NMA is that we were able to include outcome data, which were not reported in the original publications. Our study will provide insight on personalised medicine for patients with AD.
Patient or healthcare consumer involvement: People with Alzheimer’s Dementia require personalised medicine to optimise their healthcare. Evidence from high quality systematic reviews and patient-level network meta-analyses influence patient care since they are used to tailor decision making.
Objectives: To examine the comparative efficacy and safety of cognitive enhancers by patient characteristics, such as AD severity and sex, and to assess treatment-by-covariate interactions through IPD network meta-analysis (NMA).
Methods: We searched for RCTs with adults comparing cognitive enhancers. The primary outcome was cognition using the Mini-Mental State Examination (MMSE), and the secondary outcome was serious adverse events (SAEs). For eligible RCTs, we requested IPD from authors, sponsors and data sharing platforms. We assessed for consistency between results from published RCTs and provided IPD. We applied an available case analysis for each study, but we plan to explore the impact of missing data through the use of informative missingness parameters in NMA. We captured reasons for missing participants and time to SAE if this was available. We conducted a 2-stage analysis: at 1st stage IPD from included studies were aggregated to study-level summary; at 2nd stage the trial parameter estimates were synthesized in a random-effects NMA. We summarized evidence using the odds ratio (OR) and mean difference (MD), respectively. In the main analysis, we used crude ORs and MDs and did not adjust for any patient characteristics. In a further analysis we included patient-level covariates with interaction terms in the model including the patient characteristics that were provided. We combined aggregated data from RCTs for which we were unable to obtain IPD. Subgroup and meta-regression NMA was undertaken for all potential effect modifiers requested from data providers, whenever data were provided.
Results: We included 108 RCTs and received IPD for 17 (16%) RCTs. Of the 17 RCTs, we were able to include 12 RCTs in our NMA with complete data. Access to IPD via proprietary sponsor-specific platforms restricted us from combining IPD in a one-stage NMA model. In most IPD, we encountered a high dropout rate (up to 72%), for which most publications used the last observation carried forward imputation method. NMA results including IPD and/or aggregate data will be presented at the Cochrane Colloquium.
Conclusions: An advantage of our IPD-NMA is that we were able to include outcome data, which were not reported in the original publications. Our study will provide insight on personalised medicine for patients with AD.
Patient or healthcare consumer involvement: People with Alzheimer’s Dementia require personalised medicine to optimise their healthcare. Evidence from high quality systematic reviews and patient-level network meta-analyses influence patient care since they are used to tailor decision making.