We performed a Norwegian cross-sectional multicenter cohort study with public funding. Investigators from all medical universities in the country participated. This study aimed to describe a Norwegian population-based cohort on standard antihypertensive treatment, and to investigate drug adherence in patients with uncontrolled hypertension by measuring serum concentrations of commonly used antihypertensive drugs in the country.

Patients were included at four clinical centers: namely at the university hospitals in Oslo, Trondheim, Bergen, and Tromsø. Most patients were referred by general practitioners (GPs), but some were self-referred (responding to advertisements in media) or referred by nephrologists or cardiologists in hospital-based or private outpatient clinics. Participation was free of charge.

To meet the inclusion criteria, patients had to be ≥ 18 years of age, being prescribed ≥ 2 antihypertensive agents, and with a stable drug regimen during the last four weeks. Exclusion criteria comprised pregnancy, terminal illness, drug or alcohol abuse, dementia, poor Norwegian language skills, or chronic kidney disease with estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m², or urine albumin/creatinine ratio > 300 mg/mmol. All patients gave written informed consent, and the study protocol was approved by the Regional Ethical Committee of the UiT Arctic University in Norway, Tromsø, Norway (approval no. 2017/804/REK Nord). This Ethical Committee had the authority of making approval for all participating centers including subsequent changes in investigator staff, study documents, and announcements needed to recruit patients to the study.

A comprehensive description of the study design has been published previously in a half-way report [6]. In short, patients underwent a structured physician-patient interview after signing the informed consent. This interview provided information about demographic data, background, socioeconomic status (income, education, and occupation), medical history, family history, and previous treatment. All information on antihypertensive and concomitant treatment was self-reported by the patients. The total number of prescribed daily antihypertensive and concomitant pills was calculated by summing all prescribed pills per week divided by seven. After completed interview, the patients had blood samples and BPs taken. A noteworthy point in the design is that the patients were uninformed about the intention to analyze blood samples for serum drug concentrations and thereby investigate for nonadherence. This procedure was approved by the Ethical Committee and done to minimize so-called white-coat adherence.

All patients had office BP and ambulatory BP measurement (ABPM) taken with a validated device by a standardized procedure as described by the European Society of Hypertension guidelines [2]. Validated devices used in the study were Microlife WatchBP O3 (Microlife Health Management Ltd., Cambridge, UK), SunTech Oscar 2 (SunTech Medical, Morrisville, NC, USA), or Omron HEM-907 (Omron Corporation, Kyoto, Japan). Seated office BP was taken after 5 min of rest. Three measures were performed with 1 min rest between measurements, and we used the average of the last two measurements. After office measurements, the same device was used for subsequent 24 hours ABPM. Patients with a mean daytime systolic BP ≥ 135 mmHg were defined as having uncontrolled hypertension and had further follow-up visits in the main study.

All patients had their blood samples drawn and analyzed for a panel of biomarkers. In addition, a 5 mL Vacutainer tube without additives was collected for pharmacological analyses. The further preparation method has been described previously in detail [14]. Patients also brought morning urine samples for examination.

The pharmacological serum samples from all four centers were analyzed at the Department of Pharmacology at Oslo University Hospital. By use of ultra-high-pressure liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) [Agilent 1290 Infinity UHPLC combined with Agilent 6490 MS/MS (Matrix, Oslo, Norway)] the 23 most frequently prescribed antihypertensive agents in Norway were analyzed (Table 1). Experienced pharmacologists considered the patients’ drug concentrations and indicated adherence or non-adherence when related to their self-reported prescribed antihypertensive medications, dose of medications, and time of intake. Patients were defined as adherent if all prescribed antihypertensive agents were above a predefined cut-off value [6]. If one or more agents were below the cut-off value, the patient was designated as non-adherent.

All analyses were performed in SPSS 29.0 (Chicago, IL, USA). Continuous variables are shown as mean [standard deviation (SD)] and categorical variables as absolute numbers with percentages. Q-Q plots and Shapiro-Wilk test for normality were performed to verify normal distribution of samples. Fisher’s exact test or chi-square was used to compare differences between categorical variables, and students’ T-test for comparison of continuous variables. Two-sided p-value < 0.05 denoted statistical significance.

We used logistic regression analyses to identify variables associated with non-adherence. Variables with a p < 0.10 when testing for differences between groups, were included in univariate logistic regression analyses with non-adherence being the dependent variable. Significant univariate factors were included in the multivariate analyses by stepwise regression deciding the final model. We performed backward regression analysis and tested for interactions between the different covariates.

Totally, n = 1,156 patients with hypertension were included in the study. Five patients were excluded due to incomplete serum drug measurements and irregularities. Of the n = 1,151 remaining patients, n = 79 patients (6.9%) were found to be non-adherent by serum drug concentration measurements, and n = 1,072 patients (93.1%) were adherent (Figure 1).

The non-adherent patients were younger (56.9 vs. 63.7 years, p < 0.001), and accordingly, the time since diagnosis of hypertension was shorter. There was no difference in adherence status between genders. Fewer non-adherent patients were of European ethnicity (82.9 vs. 95.8%, p < 0.001) (Table 2). They had significantly higher office BP and ABPM, both systolic and diastolic, compared to the adherent patients. Mean ambulatory heart rate was also significantly increased. Furthermore, non-adherent patients more often had uncontrolled hypertension compared to adherent patients (68.4 vs. 52.5%, p = 0.007).

Figure 2 shows the age distributions of the non-adherent and adherent patients. The bar charts reveal that non-adherence is present in all age groups. However, the age groups with the highest percentage of non-adherence were 18–39 years and 40–59 years. Within the age group 18–39 years, n = 7 of 33 included patients were non-adherent (21.2%). The corresponding prevalence in the age group 40–59 years was 11.8%.

Non-adherent patients were prescribed more antihypertensive pills (p = 0.006) than adherent patients. However, the number of prescribed concomitant pills was not significantly different (Table 3). Further, the non-adherent patients were more prone to use only single-agent pills and fewer combination pills compared to adherent patients. The most used drug was angiotensin receptor blocker (ARB), with calcium channel blockers (CCBs) and diuretics as the second and third most used antihypertensive drugs in the study (Figure 3). There were significant differences in the prescription of the major drug classes between the non-adherent and adherent patients. Non-adherent patients were more often prescribed CCBs and angiotensin converting enzyme inhibitors (ACEis), whilst adherent patients had a higher frequency of ARBs and diuretics.

It can be seen from Table 3 that ACEis and CCBs have higher % in the non-adherent patient group with lower “n”. Further, the numbers of patients taking other antihypertensive drugs are overall lower than for ARBs and thus there was better statistical power for ARB in the subsequent multivariate analyses. Regarding the use of combination drugs, the non-adherent group used more frequently ACEis and CCBs in monotherapy compared to the adherent group, whilst the adherent group had significantly more use of ARB in combination with hydrochlorothiazide.

As shown in Table 2, there were no large differences in selected biomarkers between non-adherent and adherent patients, except that non-adherent patients had moderately increased eGFR (87.5 vs. 80.4 mL/min/1.73 m², p < 0.001) in line with their younger age and hypertension of shorter duration.

We did not find that patients with diabetes (n = 195, 16.9%) tended to be more, or less, adherent than other patients (p = 0.309). Patients with diabetes did not have significantly higher systolic BP than patients without diabetes (144.4 vs. 143.6 mmHg, p = 0.585). However, they had higher serum creatinine (83.7 vs. 79.9 µmol/L, p = 0.044) and a lower eGFR (79.6 vs. 81.2 mL/min/1.73 m²).

The final statistical multivariate model included four variables (Table 4). The factor strongest associated with non-adherence was age. We categorized age into quartiles and computed a dichotomous variable with the youngest quartile as the reference category vs. quartiles 2 to 4. The youngest age quartile had nearly four times higher odds of being non-adherent compared to the older three quartiles. The number of antihypertensive pills increased the risk of non-adherence by 26% for each pill. Non-Europeans had a 3.3 times higher risk of being non-adherent than Europeans. Patients who used ARBs were more likely to be adherent than patients not using ARBs. Interaction analyses did not demonstrate any significant interactions.

Higher BP was also significantly associated with non-adherence but might as likely be a result of non-adherence as a predictor for non-adherence and was therefore not included in the final statistical multivariate model. Most study participants were non-smokers (never smokers or previous smokers) (Table 2). Though there were some moderate differences in fractions of active smokers, the numbers were rather small, and smoking status did not make it to the final statistical multivariate model of characteristics that were associated with drug adherence shown in Table 4.

The low prevalence of non-adherent patients is a limitation by itself in our study. It also makes it difficult to do subgroup analysis with any meaningful statistical power. In this and in other studies, ethnicity has been an important factor for adherence. Our included patients do not reflect the diversity of the Norwegian society, with an overall non-European ethnicity of approximately 20%; maybe a higher number of non-European patients would have yielded different adherence levels.

The recruitment procedures for patients to be investigated, with or without potential biases, are likely a key to explaining various fractions of patients with reduced adherence to antihypertensive drugs. In the present study, we made no focus whatsoever on drug adherence or persistence, without any mention of drug analyses in blood samples neither orally nor in written informed consent (signed on the same day as blood sampling). However, we had to advertise for hypertensive patients to volunteer for the study; patients who were on stable treatment with ≥ 2 antihypertensive drugs. This activity might of course by itself have brought attention to the drug treatment and improved the drug adherence of the study participants.

Our study also had strengths. Investigators were highly trained doctors and study personnel in the study clinics, all located within a university hospital. The number of investigated antihypertensive patients was high in the context of such chemical adherence testing, even if there was a relatively low number of non-adherent patients as earlier mentioned. Finally, we used a method to investigate drugs for non-adherence that is reliable, with low risk of false positive or negative test results [14].

We found fewer non-adherent hypertensive patients (6.9%) than we expected. However, we investigated a population-based sample of hypertensive patients on ≥ 2 antihypertensive drugs; patients who had previously received limited attention in this context. Our results on characteristics of adherence are in line with but also extend findings from other studies. As we see it emphasis should be put on information and close follow-up to ensure adherence especially in young patients and in patients of non-European ethnicity. The use of ARBs and single-pill combinations seems to increase adherence and could be preferred from this point of view when prescribing antihypertensive medication.