OBJECTIVE: To determine changes in medication adherence in two cohorts of heart failure patients differing by year of data collection and using a direct method of adherence detection - serum drug level testing. METHODS: We added a second cohort of patients to a prospective monocentric registry of chronic heart failure patients (LEVEL-CHF registry). The two cohorts share the same inclusion criteria but differ by the year of enrolment (2018 and 2020). Stable patients with heart failure with reduced ejection fraction were enrolled in a specialized university hospital center. RESULTS: We included 402 records of 366 individual patients, 274 in 2018 and 128 in 2020. 36 patients were enrolled in both cohorts. Of the total 81% of patients were fully adherent, and 19% were non-adherent to a varying degree. Between 2018 and 2020 there was a statistically significant increase in BMI (P=0.047) and fasting glycemia (P=0.009). Patients in the 2020 cohort were less adherent than those in the 2018 cohort (P<
0.01). Patients in the two cohorts had similarly severe heart failure and did not substantially differ in NYHA class. There were no statistically significant differences between adherent and non-adherent patients after adjusting for multiple comparisons. CONCLUSIONS: In this comparison, most patients were fully adherent to all their medication and very few were non-adherent to multiple medications. We found no clinically relevant differences between adherent and non-adherent patients. Serum drug level testing is an effective method of adherence testing in clinical practice.