BACKGROUND: Estimation of the effect of different durations of treatments on safety outcomes is challenging in real-world studies because only people who survive for a long time can receive treatment for a long time. Emulating a target trial that aligns eligibility, treatment assignment, and the beginning of follow-up avoids immortal time bias. The effect of drugs with anticholinergic properties (AC) on the risk of dementia is not well established, and effect estimates on an absolute scale have not been reported.
OBJECTIVE: To estimate the effect of different durations of antihypertensive drugs with AC properties (AC AHT) on the incidence of dementia using real-world data.
METHODS: We identified initiators of AC AHT drugs aged 50 years or older without a history of cognitive impairment from the CPRD database from January 2001 through 31 December 2017. We emulated a target trial comparing three AC AHT treatment duration strategies, assigning patients to the strategy their data were compatible with at baseline: (i) ≤ 3-year (yr) strategy (after yr 3, patients can stop AC AHT or continue with a non-AC AHT), (ii) 3-6-yr strategy (after yr 6, patients can stop AC AHT or continue with a non-AC AHT), and (iii) > 6-yr strategy. Under any strategy, patients can switch between AC AHT drugs when indicated or discontinue in the presence of hypotension. We estimated the effect on the 10-yr risk of dementia under complete adherence by creating clones and censoring them when they deviated from the assigned strategy. Risk differences (RDs) and risk ratios (RRs) were estimated by taking the ≤ 3-yr strategy as reference and adjusting by baseline and time-varying covariates via inverse probability weighting with 95% confidence intervals (95% CI) computed via bootstrapping (360 runs).
RESULTS: At baseline, 77,365 eligible individuals were assigned to each treatment strategy, contributing 8,648 yrs of follow-up to the ≤ 3-yr strategy, 7,565 yrs of follow-up to the 3-6-yr strategy, and 9,859 yrs of follow-up to the > 6-yr strategy. The 10-yr risk of dementia was 6.4% (95% CI, 5.8%-7.0%) for the ≤ 3-yr strategy, 6.2% (95% CI, 5.2%-7.1%) for the 3-6-yr strategy, and 5.1% (95% CI, 4.7%-5.5%) for the > 6-yr strategy. Compared with the ≤ 3-yr strategy, the 10-yr RD for the 3-6-yr strategy was −0.3% (95% CI, −1.3%-0.8%) and for the > 6-yr strategy was −1.3% (95% CI, −2.0 to −0.5). The corresponding RRs were 1.0 (95% CI, 0.8-1.2) and 0.8 (95% CI, 0.7-0.9), respectively.
CONCLUSIONS: Our results are compatible with a small absolute protective effect on the risk of dementia with longer use of AC AHT.
