In Denmark, a network of administrative and clinical registers and databases with nationwide coverage enables large-scale epidemiological research. Using a unique identifier (ie, the CPR number) all residents can be followed on an individual level linking different databases pertaining to demographic variables (ie, age, sex, death, education level), hospital admissions and diagnoses (International Classification of Diseases 10th edition [ICD-10]), hospital procedures and operations, as well as medicine dispensed at any pharmacy in Denmark. ^,, The registers have been described elsewhere in more detail and have been used extensively and validated, especially for cardiovascular research. ^,, Enrollment in the databases is mandatory for all citizens in Denmark and does not require individual consent.

The study was conducted as a nationwide cohort study including all patients ≥40 years of age initiating treatment for hypertension with either a BB (metoprolol, bisoprolol, carvedilol, atenolol, propranolol, and other [<1%]) or a CCB (amlodipine, lercanidipine, verapamil, and other [<1%]) between 2012 and 2023. Patients were required to be naïve to the mentioned drugs. Hypertension was defined as treatment with at least 2 types of antihypertensive medication as BBs are no longer a first-line therapy for hypertension. As such, all patients were in antihypertensive treatment with at least 1 other drug prior to BB or CCB treatment initiation. Using claimed prescriptions for defining hypertension in the Danish registers is a method that has been previously used and validated. Patients with other potential indications for BB- and CCB treatment (eg, heart failure, tachyarrhythmias, ischemic heart disease, hyperthyroidism, and essential tremor) were excluded to obtain a homogenous study population and appropriate comparator groups. Likewise, patients with medical history of any of the outcomes were excluded as well. ( Figure 1 ) Codes (eg, ICD-10 and ATC codes) used for defining variables for inclusion and exclusion criteria, please see Supplementary Table S1.

Flowchart. Flowchart showing the application of exclusion and inclusion criteria into the final cohort.

Five different outcomes identified as markers of potential quality-of-life reducing side-effects to BB treatment were predetermined as: (1) depression, (2) erectile dysfunction, (3) anxiety or insomnia, (4) gastrointestinal side-effects, (5) dizziness, fainting or orthostatic hypotension. (1) Depression was defined as a claimed prescription of an antidepressant; (2) erectile dysfunction by a claim of a prescription for phosphodiesterase type 5 inhibitor; (3) anxiety or insomnia by a claimed prescription of anxiolytics or sedating medication used for treating insomnia; (4) gastrointestinal side-effects by diagnostic codes for constipation, or a claimed prescription of a laxative; (5) dizziness or fainting by diagnostic codes for dizziness, fainting, collapse, or orthostatic hypotension. “Any outcome” was defined as well as a composite of all described outcomes except erectile dysfunction as only male patients are included in the analysis of this specific outcome. Specific ICD-10 and ATC codes are listed in Supplementary Table S1.

Patients were followed from the date of their first-time claim of either a BB or a CCB until event of interest, emigration, death, end of study period (31 December 2023), or 1 year of follow-up. Specifically for patients initiated on CCB, follow-up was also halted if a prescription of a BB was claimed during follow-up since this would place these patients “at risk” in this active comparator design. On the contrary as CCB treatment was assumed not to be associated with the outcomes, a claimed prescription of CCB in the BB group during follow-up was ignored.

Baseline characteristics were defined at index-date (initiation of treatment with either a BB or CCB). Comorbidity included in the study was a medical history of chronic obstructive pulmonary disease, stroke, chronic kidney disease, peripheral artery disease, and cancer. (Supplementary Table S1). Comorbidities were captured using diagnostic codes (ICD-10) in a period of 5 years prior to study inclusion as done previously. Medical treatment considered at baseline was treatment with renin-angiotensin-system inhibitors, thiazides, loop diuretics, statins, antiplatelets, anticoagulants, nonsteroid anti-inflammatory drugs, opioids, and glucose lowering drugs (used as a proxy for diabetes mellitus). Educational level was categorized as elementary- or high school, vocational education, higher education, or unknown.

Characteristics in categories at baseline were shown as both total counts and percentages within each group of interest. Age was presented as median with interquartile range (IQR). To describe the absolute risk of possible BB side effects, the Aalen-Johansen estimator was used to calculate crude 1-year absolute risks with 95% confidence intervals (CI) treating death as competing risk. Using an active comparator design, where BB treatment was compared with CCB treatment, standardized absolute 1-year risks following initiation of both drugs and risk ratios were computed. First, a multivariable, cause-specific cox model was fitted for the specific outcomes including the following predetermined variables: age, sex, educational level, and medical history of stroke, diabetes mellitus, chronic obstructive pulmonary disease, cancer, and chronic kidney disease. Second, the G-formula was executed to compute average (adjusted) treatment effects and provide adjusted absolute risks and corresponding risk ratios (RR) comparing BB treatment with CCB treatment. ^,

Several supplementary analyses were performed: 1) analyses were performed divided into two age strata (>65 years and ≤ 65 years of age) examining different presentations of BB side-effects according to age, 2) analyses were performed stratified by the specific BB/CCB subtype, 3) analyses were performed stratified by sex (erectile dysfunction not included), 4) and an analysis investigating a potential dose-response relationship was performed. The category “high-dose” was defined as patients claiming prescriptions for the highest tablet strength available for the different BBs (≥100mg metoprolol, ≥10mg bisoprolol, ≥25mg carvedilol, ≥100mg atenolol, ≥80mg propranolol). Patients claiming the “high dose” were then compared with patients receiving other and lower strengths of the drugs categorized as “low dose”. As a marker of treatment adherence during follow-up, we descriptively assessed the proportion of patients claiming a second prescription after treatment initiation within a year using the Aalen-Johansen estimator with death as a competing risk. A similar approach was applied to show the proportion of patients claiming a third prescription following a second. Further, to investigate the different outcomes’ potential impact on adherence, the proportion of claiming another prescription was calculated in a similar manner among patients experiencing an outcome. Ultimately, an analysis examining time trends (ie, stratifying analyses before and after 2017) as well as analyses by frailty score were performed. All data management, figures, and statistical analyses were conducted using R version 4.4.1.

Retrospective, observational studies based on administrative data do no not require ethical approval in Denmark. Permission was granted to the institution by the responsible government data institute and the project has been registered.

No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper and its final contents.