Real-world treatment gaps in hypertension management: drug switching, prescribing patterns and apparent resistant hypertension – a retrospective observational study

Although antihypertensive pharmacotherapy is well established, the treatment gaps encountered in outpatient practice — particularly drug switching, class-specific tolerability signals and real-world resistant hypertension — remain poorly characterized in Türkiye. To quantify antihypertensive prescribing patterns, drug switch patterns and the prevalence of, and factors associated with, apparent resistant hypertension (aRH) in a large outpatient population. Single-centre retrospective registry-based study of all 4,500 adult hypertensive patients (ICD-10 I10–I15) followed in Internal Medicine and Cardiology outpatient clinics during 1 January – 31 December 2024. Prescribing status and blood pressure (BP) were ascertained cross-sectionally at the last outpatient visit within the window; drug switches were captured over the full 12 months. All primary analyses were repeated in a pre-specified core hypertensive cohort excluding chronic kidney disease (CKD) and heart failure (HF) ( n = 3,814). Pre-specified sensitivity analyses restricted the aRH definition to patients receiving a renin-angiotensin system (RAS) blocker, calcium channel blocker (CCB) and diuretic, and restricted spironolactone uptake to biochemically eligible patients (potassium < 5.0 mmol/L, eGFR ≥ 45 mL/min/1.73 m²). Multivariable logistic regression was performed with variables screened at p < 0.20 plus clinically essential covariates; multicollinearity was assessed with variance inflation factors (VIF); a second model was restricted to the at-risk population receiving ≥ 3 agents including a diuretic. The cohort was 60.8 ± 12.0 years old and 53.4% female. CCB (53.3%) was the most used class, followed by ARB (47.5%), diuretic (41.0%), β-blocker (40.5%) and ACEi (29.0%); 68.4% were on combination therapy and office BP was below the study threshold of 140/90 mmHg in 59.0%. In the core cohort, CCB use rose to 57.2% and β-blocker use fell to 36.2%, with 60.1% below threshold. A first documented switch was recorded in 16.8% of patients (16.1% in the core cohort), driven chiefly by presumed insufficient efficacy assessed on office BP alone (38.3%) and side effects (32.8%); ACEi had the highest class-specific first-switch rate (10.8%), more than double that of ARB (5.1%), a gradient that persisted but narrowed when restricted to side-effect-attributed switches (3.5% vs. 1.9%, p = 0.004). Apparent resistant hypertension was present in 15.3% of the cohort (12.7% under a core-triple-restricted definition; 14.6% in the core cohort). In multivariable analysis (pseudo-R² 0.030; AUC 0.622; Hosmer-Lemeshow p = 0.358; all VIF < 1.1), factors independently associated with aRH were baseline SBP (per 10 mmHg: aOR 1.25, 95% CI 1.18–1.31), diabetes (aOR 1.47, 1.19–1.81), CKD (aOR 1.48, 1.13–1.92) and Cardiology follow-up (aOR 1.24, 1.05–1.47); discrimination was too low for individual risk prediction, and clinic was not associated with aRH in the restricted model (aOR 1.20, p = 0.204) or in the core cohort (aOR 1.19, p = 0.064). Spironolactone was used in 1.8% of the whole cohort and in 11.5% (79/689) of aRH patients — 10.7% among the 81.6% who were biochemically eligible — leaving approximately nine in ten aRH patients without a fourth-line mineralocorticoid receptor antagonist. β. In this real-world Turkish outpatient population, prescribing broadly followed guideline-recommended patterns, but two gaps stand out. First, ACEi carried a first-switch rate twice that of ARB, and side-effect-attributed switching was significantly more frequent with ACEi than ARB, consistent with — though not independent evidence for — the existing guideline preference for an ARB in patients intolerant of an ACE inhibitor. Second, only approximately one in nine patients meeting aRH criteria, and one in nine of those biochemically eligible, received a mineralocorticoid receptor antagonist. Because adherence was unverified and out-of-office BP unavailable, the aRH estimate should be read as an upper bound and the aRH analyses as exploratory and hypothesis-generating.