Cohort A | Cohort B | Cohort C | |
---|---|---|---|
Purpose | To investigate incidence by comparing treated and untreated individuals. | To investigate incidence by comparing the different drug subclasses of each treatment. | To investigate progression by comparing treated and untreated individuals. |
Number of cohorts required | There will be three cohorts of this type, one for each treatment of interest. | There will be three cohorts of this type, one for each treatment of interest. | There will be three cohorts of this type, one for each of dementia (AD or NADD), PD and ALS. |
Exposures | Treatments for hypertension, hypercholesterolaemia and type 2 diabetes. | Treatments for hypertension, hypercholesterolaemia and type 2 diabetes. | Treatments for hypertension, hypercholesterolaemia, and type 2 diabetes. |
Start of follow-up (index date) | Date at first risk of the condition the treatment is used for or date of first diagnosis of the condition itself if there was no preceding period ‘at risk’. | Date of first prescription of a treatment of interest. | Date of first diagnosis of neurodegenerative disease of interest. |
Outcome | Diagnosis of neurodegenerative disease of interest. | Diagnosis of neurodegenerative disease of interest. | Death. |
Exclusion criteria | Individuals with <12 consecutive months of records prior to cohort entry. | Individuals prescribed treatment and control medications at the same time or with <12 consecutive months of records prior to cohort entry. | Individuals with <12 consecutive months of records prior to cohort entry. |
Statistical analysis | Conventional regression, propensity score regression, instrumental variable analysis and marginal structural models. | Conventional regression, propensity score regression, instrumental variable analysis and marginal structural models. | Conventional regression, propensity score regression and marginal structural models. |
AD, Alzheimer's disease; ALS, amyotrophic lateral sclerosis; NADD, Non-Alzheimer's disease dementias; PD, Parkinson's disease.