Analysis code for the paper:
Associations between midlife anticholinergic medication use and subsequent cognitive decline
Mark J Rawle, Wallis Lau, Arturo Gonzalez-Izquierdo, Praveetha Patalay, Marcus Richards, Daniel Davis
Link to be provided to paper once published.
Contains three files:
Generate ACBS Score Age 69.do - Example file detailing the calculation of Anticholinergic Cognitive Burden Score from raw medication data within the National Survey for Health and Development (NSHD). This is proivded for age 69 only, but similar code conducting the process at all waves of NSHD data collection with valid medication data can be provided on request (ages 31, 36, 43, 54, 60-64 & 69).
MJR ACBS and cognition (NSHD).do - The second file provided is a collection of the code used to calculate the variables used within the paper outlined below, alongside code to create appropriate tables, figures and sensitivity analyses (the outputs of which are all provided in the submitted manuscript). Please note, some variables have been renamed from the original var names in NSHD (especially where they are used in longitudinal analysis), this has been outlined within the code.
FDR code.do - Stata code for creating sharpened q-values from a dataset of p-values. Based on code from ANDERSON, M. L. 2012. Multiple Inference and Gender Differences in the Effects of Early Intervention: A Reevaluation of the Abecedarian, Perry Preschool, and Early Training Projects. Journal of the American Statistical Association, 103, 1481-1495.
ABSTRACT
Background: Anticholinergic medication use is associated with cognitive decline and incident dementia. Our study aimed to determine if repeated exposure was associated with greater decline, and whether decline was reversed with medication cessation or reduction.
Methods: From the MRC National Survey for Health and Development, we quantified anticholinergic exposure between ages 53 and 69 years using the anticholinergic cognitive burden scale (ACBS). We used multinomial regression to estimate associations with the Addenbrooke’s Cognitive Examination (ACE-III). Longitudinal associations between ACBS and cognitive test results (world learning test WLT, timed letter search task TLST) at three time points (age 53, 60-64 and 69) were assessed using mixed and fixed effects linear regression models.
Results: Anticholinergic exposure was associated cross-sectionally with lower ACE-III scores at age 69, with the greatest effects in those with high exposure at ages 60-64 (mean difference: -2.34, 95%CI -3.51, -1.17). Longitudinally, both mild-moderate and high ACBS scores were linked to lower WLT scores, again with high exposure showing larger effects (mean difference with contemporaneous exposure: -0.90, 95%CI -1.63, -0.17; mean difference with lagged exposure: -1.53, 95%CI -2.43, -0.64). Associations remained in fixed effects models (mean difference with contemporaneous exposure: -1.78, 95%CI -2.85, -0.71; mean difference with lagged exposure: -2.23, 95%CI -3.33, -1.13). Associations with TLST were noted only in isolated contemperanous exposure (mean difference -7.85, -13.14 p<0.01).
Conclusions: Anticholinergic exposure throughout mid and later life was associated with lower cognitive function. Associations persisted in individuals even when their anticholinergic medication use decreased over the course of the study.