Association of antiherpetic medication with incident dementia in Denmark, Wales, Germany and Scotland

Abstract

AbstractBackgroundSeveral epidemiological studies from Taiwan, all using the same data resource, found significant associations between herpes virus infection, antiherpetic medication and dementia. We conducted a multicenter observational cohort study using health registry data from Wales, Germany, Scotland, and Denmark to investigate potential associations between antiherpetic medication and incident dementia, and also to comprehensively investigate such associations broken down according to medication type and dose, type of herpes virus, and dementia subtype.Method2.5 million people aged 65 and older were followed up using routinely collected linked electronic health records in four separate national observational cohort studies. Exposure and outcome were classified using coded data from prescriptions and from primary and secondary care. Data were analyzed using survival analysis with time‐dependent covariates. Confounders were age, year, sex, socioeconomic status, and comorbidities.ResultResults were heterogenous across cohorts (Figure 1), with a tendency for decreased dementia risk in people exposed to antiherpetic medication (Wales: HR 0.91, 95% CI 0.86 to 0.97; Denmark: HR 0.91 (0.89 to 0.93); Germany: HR 1.08 (0.98 to 1.20); Scotland: HR 0.98 (0.64 to 1.49). Associations were not affected by number of treatments, herpes diagnosis, type of dementia, or specific type of medication. People diagnosed with herpes but not exposed to antiherpetic medication were at higher dementia risk in the German cohort (HR: 1.18; 1.09 to 1.28) but not in the Welsh cohort (HR: 0.95; 0.88 to 1.02).ConclusionResults from the four large cohorts allow us to exclude any major association of short‐term exposure to antiherpetic medication with dementia. Because neither type of dementia nor type of herpes infection modified the association, the small but significant decrease in dementia incidence with antiherpetic administration mayeflect unmeasured confounding and misclassification.