Adherence to antihypertensive drug treatment in patients with apparently treatment-resistant hypertension in the INSPiRED pilot study

Abstract

Purpose: Drug adherence may be a major problem in the therapy of hypertension and in the diagnosis of therapy resistance. Adherence can be assessed by indirect methods or by direct methods like drug detection in urine with liquid chromatography-mass spectrometric methods.

Materials and methods: The current analysis included patients with apparently treatment- resistant hypertension (TRH) referred for renal denervation (RDN) and included in the the INSPiRED pilot trial (NCT 01505010). Adherence was repeatedly assessed by toxicological urine analysis over a time range of up to 17 months in a total of 18 patients.

Results: In the first urine samples of 18 patients the adherence rate (percentage of number of detected vs. prescribed medical drugs) ranged from 0 to 100% with a median of 73.2%. In further urine samples collected during the following up to 17 months every individual patient exhibited considerable changes in the adherence rate, neither a constancy nor a tendency could be deduced.

Conclusions: Urine analysis results exhibit variation over time and an assessment at a certain time point cannot be regarded as representative or predictor for future behavior. Therefore, it appears necessary to perform drug adherence testing repeatedly over time.

Trial registration: ClinicalTrials.gov identifier: NCT01505010.

Keywords:

• Compliance
• resistant hypertension
• toxicological analysis
• antihypertensive drugs
• renal denervation

Introduction

Resistant hypertension is defined as a permanently elevated blood pressure above 140/90 mm Hg, despite an antihypertensive drug regimen of at least 3 substances of different classes in optimal doses, preferably including a diuretic [1]. According to the US definition, patients with controlled blood pressure on medication with at least 4 drugs of different classes are also included in the definition of resistant hypertension [2].

Previous studies in patients with treatment-resistant hypertension have shown that a considerable percentage of the insufficient blood pressure adjustment is due to a lack of adherence [3–5].

Methods of adherence testing can be indirect by pill counting, prescription refill and electronic monitoring, but direct methods such as direct observation of therapy, determination of plasma and urine biomarkers or urine/plasma drug detection with liquid chromatography-mass spectrometric methods are considered more reliable [6–8]. The latter type of adherence testing has received increased attention, but such an analysis is not routinely available in most laboratories. This is due to the fact that no on-site tests are available for such a determination because of the large number of different substances. Therefore, a more complicated analysis has to be carried out. Such methods for detection of specific antihypertensive drugs in urine [3,9–13] as well as in serum have been described [4,11,14,15].

In the INSPiRED study [16,17] adherence testing was performed with liquid chromatography-mass spectrometric methods in urine samples collected over an observation period of up to 17 months. The results were evaluated and compared to give an impression on the reliability of adherence assessment.

Materials and methods

Study design

The INSPiRED study (trial registration: NCT01505010 on ClinicalTrials.gov) was designed to compare the blood pressure lowering efficacy and safety of renal denervation (RDN) vs. usual medical therapy in patients with apparently treatment-resistant hypertension (TRH). Details on patient selection, inclusion and exclusion criteria and the results are described by Jin et al. [16] and Jacobs et al. [17]. All patients were prescribed at least three or more antihypertensive drugs of different classes, preferably including a diuretic. The median number of prescribed drugs at the beginning was 6 (range 3–7). In the course of the study, however, the drug therapy could be adapted to new requirements.

Drug analysis in urine

The informed consent signed by patients enrolled in the INSPIReD pilot trial stated that a urine sample would be collected at every visit in order to check the presence of antihypertensive drugs in the urine. This was not reminded on subsequent visits. Urine samples of a total of 18 patients were collected for drug analysis at baseline (start of the study, 0 months), but additional samples were collected on the occasion of both planned and unplanned visits up to 17 months after start of the study.

The urine samples were analyzed as described previously [9] using ultra-high performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS consisting of 1290 LC and 6460 Triple Quadrupole MS from Agilent, Waldbronn, Germany) in the selective and highly sensitive multiple reaction monitoring mode (MRM) for the following 54 antihypertensive drugs and metabolites: aliskiren, altizide, amiloride, amlodipine, atenolol, azilsartan, barnidipine, benazepril/benazeprilat, bendroflumethiazide, bisoprolol, bumetanide, candesartan, carvedilol, chlortalidone, clonidine, diltiazem, doxazosin, enalapril/enalaprilat, eprosartan, felodipine, furosemide, hydrochlorothiazide (HCT), irbesartan, lercarnidipine, lisinopril, losartan, metoprolol, minoxidil, molsidomine, moxonidine, nebivolol, nifedipine, nitrendipine, olmesartan, perindopril/perindoprilat, piretanide, prazosin, quinapril/quinaprilat, ramipril/ramiprilat, spironolactone/canrenone, telmisartan, terazosin, torasemide, triamterene, urapidil, valsartan, verapamil and xipamide.

If patients received drugs that were not covered by the toxicological analysis (i.e. hydralazine, labetalol, indapamide), they were not considered for the calculation of the adherence rate. However, the analysis results can be regarded as highly reliable because 97.4% of all the prescribed drugs that were expected in all the urine samples were detectable.

Results

Toxicological urine analysis was performed for 18 patients of whom there were in median 5 urine samples (range 2–10) obtained up to 17 months after the start of the study. In Table 1, the numbers of detected drugs in comparison to the total number of drugs prescribed at each sampling time are given in relation to months after start of the study. From these data, the adherence rates were calculated. It is obvious that marked changes in adherence rate did not only occur between subjects but also within subjects.

Table 1. Number of substances detected in urine with the number of prescribed medical drugs in parentheses. For patients who received a renal denervation treatment (RDN) timing was indicated to illustrate the effect on adherence. Urine samples from 18 patients were analyzed at different and irregular points in time. The median and range (min-max) of the adherence (%) in the individual patients is shown. Below a graphical representation of the range of adherence is given over all patients, the line indicates the medians.

At the beginning of the study, patients received in median 6 antihypertensive drugs (range 3–7). In the course of the study, the medication could be adapted to the therapeutic requirements, so that in the entire course of the study 2 drugs were prescribed at 3.1% of urine sampling times, 3 drugs in 12.2%, 4 drugs in 13.5%, 5 drugs in 24.5%, 6 drugs in 33.7%, 7 drugs in 12.2% and 8 drugs in 1%.

At baseline, median adherence among all patients was 73.2%. For the planned follow-up visits after 1, 3 and 6 months, the median adherence was 61.4%, 14.3% and 33.3% respectively. In addition to these planned visits, urine samples were collected from further follow-up visits, with median adherence varying between 0% and 85.7%. The range of variation over all points in time and patients was between 0% and 100%.

When looking at the individual patients in more detail, only 3 patients (#7, #13 and #14, Table 1) had a high or full adherence. Patients #4, #11, #15 and #17 showed very little adherence at all times. In patients #2, #12 and #16, adherence increased from start to end of study. Patients #3 and #10, on the other hand, exhibited good adherence at the beginning and stopped taking their medication at some time during the course of the study.

Discussion

Drug adherence is a major problem in the therapy of hypertension and in the diagnosis of therapy resistance. Previous studies have shown that adherence in patients with apparently treatment-resistant hypertension can be as low as only 50% [3] but in controlled studies higher rates are also observed [9]. Poor drug adherence and subsequent poor blood pressure control despite prescription of multiple antihypertensive drugs may lead to consider invasive and cost-intensive treatments such as renal denervation [18–20]. Therefore, adherence assessment is an important part in the diagnosis of therapy resistant hypertension.

There are several ways to assess adherence [6]. Indirect and direct methods are distinguished, and each method has its advantages and disadvantages. In the present study, adherence was assessed by toxicological urine analysis of a large range of substances as specific targets, which is, therefore, a direct method. The data obtained were from patients considered for inclusion in the INSPiRED pilot study [16,17].

For determination of longitudinal drug adherence, urine samples of 18 patients were analyzed. We expect that “white-coat adherence” [21] was limited because adherence testing took place during a long period and patients did know when adherence testing would be performed.

In Table 1, the number of drugs detected in all urine samples with respect to the number prescribed, which might have changed through interventions, are detailed. The patients were mostly on up to 6 drugs where in 24.5% instances this were 5 drugs, most were on 6 drugs (33.7%), some on 7 (12.2%) and only one patient (#12) was prescribed 8 drugs at one time. Notably, only three patients exhibited full or high adherence over the whole time range (#7, #13 and #14).

In order to compare the patients and characterize their adherence, a classification is necessary. However, there is no clear or accepted classification of adherence by urine drug analysis. In the literature different adherence ratings can be found. In some studies, adherence is assumed when 100% of the prescribed drugs have been detected, all other findings representing non-adherence [4,11,15,22]. In other studies this was more graded like full-compliance (100% of the medication detected), partial compliance (at most 1 substance not detected) and total non-compliance (no medication detected) [10,14]. This was similarly used by Schmieder et al. [9] who assumed non-adherence if 2 or more substances were not detected in the analysis. Other authors [3] classify compliance in several categories of percentage of detected drugs (i.e. 0%, < 25%, 26–50%, 51–75%, > 75% and 100% detected drugs in comparison to the prescribed medication). A simple cut-off was proposed by Schmieder et al. [9] where detection of less than 80% of the medication drugs was regarded as non-adherent, which appears similar to the approach of de Jager et al. [18], who regarded 80% of detected drugs or less as non-adherence. This is equivalent to the assumption of non-adherence if 2 or more drugs of up to 9 prescribed drugs are not detected or if 1 or more drugs of up to 4 prescribed drugs are not detected. Only for 5 prescribed drugs there is a difference between both approaches, as a patient who would not take one of his five tablets (exactly 80% detection rate) would be regarded as adherent by the definition of de Jager et al. [18] but as non-adherent by Schmieder et al. [9]. In our study, this would affect 6 of 24 instances where 4 of 5 drugs were detected.

Therefore, using the percentage of drug detection as the adherence rate, the different patients may be compared. In the first urine samples of all 18 patients at the start of the study the adherence rate ranged from 0 to 100% with a median of 73.2%. During follow-up a number of samples were obtained in addition to the scheduled visits (1, 3 and 6 months after start of the study), and for more than a year afterwards. In every individual patient, considerable variations in the adherence rate over time can be noticed (Table 1), neither a constancy nor a tendency can be deduced. Patient #3 (Table 1) for example exhibits a 100% adherence to the 5 different prescribed drugs at the times of the first two urine samples, in the sample 3 months after study start none was detectable (0%), in the next month all 5 are again detectable but none in the samples 4 and 10 months later. This and other examples (e.g. #5 and #10, Table 1) show that adherence assessment via urine analysis is only representative for a selected time point and cannot be generalized to a larger time range, and that full adherence at a certain time does not guarantee a continued stable compliance with medication. It also does not appear that renal denervation (indicated by RDN in Table 1) drives patients to be more compliant or on the contrary leads to premature treatment interruption because patients hope to be cured, as was speculated in various reviews and editorials [23]. Current studies aim at elucidating psychologic factors to identify predictors of drug adherence and drug resistance [24].

Notably, most patients included in this study remained severely hypertensive despite prescription of a median of five antihypertensive drugs. Therefore, many of them met the definition of refractory hypertension [25]. Whether our findings may be extrapolated to more “standard” resistant hypertensive patients (i.e. mild blood pressure elevation on three antihypertensive drugs) or other subtypes of difficult-to-treat hypertension remains to be demonstrated.

Conclusion

Adherence was repeatedly assessed in the INSPiRED pilot study over a time range of up to 17 months using toxicological urine analysis. The results show that the drug detection rate exhibits marked fluctuations without an obvious constancy or trend. Therefore, even when assessed using objective, direct methods, adherence varies over time and an assessment at a certain time point cannot be regarded as representative or predictor for future behavior. At least in patients with apparently treatment-resistant hypertension, it appears necessary to perform drug adherence testing repeatedly over time.

Disclosure statement

The authors report no conflict of interest.