Direct assessment of adherence and drug interactions in patients with hypertensive crisis—A cross‐sectional study in the Emergency Department

Abstract

Though drug adherence is supposed to be low in hypertensive crisis (HTN‐C), there are no data available from direct adherence assessments. The aim of the present study was to evaluate adherence to prescribed antihypertensives and potential interactions of concomitant drugs and foods with prescribed antihypertensives in patients with HTN‐C by a direct evaluation via biochemical urine analysis. In the present cross‐sectional study, 100 patients with HTN‐C, admitted to the emergency department (ED), were included. A biochemical urine analysis using gas chromatography‐tandem mass spectrometry was performed. Out of 100 patients, 86 received antihypertensives. Urine analyses could be evaluated unambiguously in 62 patients. In 15 of these 62 patients (24%), a nonadherence could be demonstrated, and in 21 patients (34%), a partial nonadherence could be demonstrated. Patients with nonadherence or partial nonadherence showed a longer hypertension history (15[5‐22] vs 10[3‐15] years, P = 0.04) were prescribed more general medication (number 7.1 ± 3.4 vs 3.4 ± 1.8; P < 0.01) as well as antihypertensive drugs (number 2.8 ± 1.1 vs 1.5 ± 0.7, P < 0.01). A potential BP‐raising trigger by medications or food interaction was frequently detectable, predominantly with nonsteroidal anti‐inflammatory drugs (NSAIDs; n = 38), glucocorticoids (n = 8), antidepressants (n = 10), and licorice (n = 10). Nonadherence and partial nonadherence to prescribed antihypertensives might play a crucial role for the occurrence of HTN‐C. However, further case‐controlled studies are needed to confirm the present findings. Ingestion of concurrent over‐the‐counter drugs such as NSAIDs but also prescribed drugs as well as aliments may lead to critical BP elevation. In order to prevent HTN‐C, the present findings emphasize the importance for clinicians to pay attention to the issue of adherence and co‐medication.

Abbreviations

BMI

body mass index

BP

blood pressure

CI

confidence interval

ED

emergency department

eGFR

estimated glomerular filtration rate

GC‐MS

gas chromatography‐coupled mass spectrometry

HTN

arterial hypertension

HTN‐C

hypertensive crisis

IQR

interquartile range

NSAID

nonsteroidal anti‐inflammatory drugs

OR

odds ratio

SBP

systolic blood pressure

SD

standard deviation

1. INTRODUCTION

Hypertensive crisis (HTN‐C) is defined by a severe and abrupt increase in blood pressure (BP) with impending or progressive acute end‐organ damage.1 It is classified into hypertensive emergency and urgency based on the presence or absence of acute end‐organ damage. Among the HTN population, about 1%‐2% of the patients develop HTN‐C.2 HTN accounts for as many as 0.9% of all emergency department (ED) visits,3 HTN‐C of 0.5%4 including about 0.2% with hypertensive emergency.5 In patients with hypertensive emergency, the mortality is higher (4.6%) compared with patients with hypertensive urgencies (0.8%)6 with a 30‐day mortality of 4% in patients who require parenteral antihypertensive therapy for HTN‐C7 and 90‐day readmission rate with the same diagnosis in 29% of cases who initially required hospitalization for HTN‐C.8 Two different but interrelated mechanisms may play a central role in the pathophysiology of the hypertensive crisis. Firstly, a failure of the autoregulatory mechanisms in the vascular bed and secondly activation of renin‐angiotensin system, lead to further vasoconstriction. This generates a vicious cycle of continuous injury and subsequently ischemia.2 Furthermore, withdrawal of certain BP medications, specifically drugs acting on the sympathetic nervous system, can cause a sudden increase in BP, called rebound hypertension.9 Moreover, a series of different risk factors have additionally been described to be associated with HTN‐C including female sex, the grade of obesity, the presence of hypertensive or coronary heart disease, the presence of a somatoform disorder or a higher number of prescribed antihypertensive drugs.10 The most important risk factor besides the above mentioned was, however, nonadherence to medication. All studies investigating adherence in HTN‐C showed several methodological shortcomings, like retrospective design or indirect adherence measurement (eg, assessment by clinicians, questionnaires/interviews, pill count, prescription refill, measurement of pharmacodynamics parameters). To the best of our knowledge, the present study represents the first study overcoming these limitations by using a prospective design and a direct biochemical measurement to assess adherence data in patients with HTN‐C admitted to the ED.

2. METHODS

2.1. Study design and inclusion/exclusion criteria

Our study was designed as a prospective cross‐sectional study including 100 patients with HTN‐C admitted to the emergency department of the University Medical Center Göttingen. Patients who agreed to participate in the study were included. After getting information about the study, all screened patients had the option of not participating. All included patients provided informed consent before the initiation of protocol‐mandated procedures. This study was approved by the local ethics committee (#6/6/14) and was performed according to the principles outlined in the Declaration of Helsinki.

Adult patients with a HTN‐C, that is, BP ≥180 mm Hg systolic and/or ≥120 mm Hg diastolic determined in the ED, were included. If primary reason for ED admission reason were head injury, or cerebrovascular emergency, patients received stem cell or bone marrow transplant, prior enrollment and/or pregnancy were considered as exclusion criteria.

2.2. BP Measurement in the ED

BP measurement oscillometric method was performed by trained research staff before administration of antihypertensive medications in the ED. BP measurement was obtained in a standard fashion using an appropriately sized BP cuff at heart height with patients in semi‐recumbent position on an ED stretcher.

2.3. Assessment of risk factors

The following potential risk factors for HTN‐C were assessed: sociodemographic data (age, sex, marital status, profession), body mass index (BMI), systolic and diastolic BP, secondary causes of HTN, duration of antihypertensive treatment, number and name of the prescribed antihypertensive(s), concomitant medication, intake of combination pills, presence of relevant cardiovascular risk factors and diseases, alcohol or nicotine abuse and relevant concomitant diseases.

Moreover, self‐reported ingestion of foods potentially interfering with antihypertensive drugs (ie, grapefruit, garlic, broccoli, other types of cabbage, cinnamon, wild garlic, St. John's wort or licorice) until 48 hours prior to ED admission were documented.

2.4. Laboratory tests and Assessment of adherence

Urine specimens were taken for medication adherence tests. Systematic biochemical analysis was performed by gas chromatography‐coupled mass spectrometry (GC‐MS) using a GC‐MS system from Agilent Technologie (Waldbronn, Germany), consisting of a gas chromatograph 66890N with split/splitless injector and autosampler (7683B) coupled with a quadrupole mass spectrometer (MSD 5975C). The urine samples were extracted natively alkaline with an organic solvent mixture. Parts of the sample were subjected to cleavage of metabolic adducts by hydrochloric acid hydrolysis. After extraction and derivatization with acetic anhydride, the sample parts were combined and analyzed by gas chromatography. For the detection of hydrophilic drugs, for example, ACE inhibitors and diuretics, a third part of the sample was subjected to an extractive alkylation before gas chromatography. The extracts obtained were analyzed by gas chromatography/mass spectrometry in the full scan mode. The identification of the drugs was carried out by mass spectral comparison with reference libraries.11

Among the entirety of the antihypertensive medication taken by all patients, 23 prescribed antihypertensive medications could be detected by the GC‐MS (ramipril, enalapril, benazepril, valsartan, irbesartan, lorsartan, metoprolol, bisoprolol, propranolol, lercanidipine, amlodipine, felodipine, nitrendipine, verapamil, urapidil, torasemid, furosemide, piretanid, hydrochlorothiazide, xipamid, triamteren, eplerenone, and spironolactone). Limitations of the method related to antihypertensive substances were detection of nebivolol, carvedilol, aliskiren or moxonidine only in high concentrations. Lisinopril, telmisartan, candesartan, doxazosin, and isosorbide dinitrate were not covered by the method. For the assessment of adherence solely, the 23 well‐detectable antihypertensive drugs were used. Patients were classified as adherent when all of the prescribed antihypertensive drugs were found in the urine. Otherwise, they were classified as nonadherent. Within the group of nonadherent patients, a subgroup of partial nonadherent (if at least one prescribed antihypertensive drug was found in the urine) and complete nonadherent (none of the prescribed antihypertensive drugs were detected) were defined.

Determination of the actual antihypertensive medication was performed by reviewing the medication plan issued by the general practitioner in combination with an interview for self‐reported adherence in which patients were asked for (a) the duration of antihypertensive treatment, (b) the name of the used substances, (c) the number of antihypertensive drugs, (d) the use of fixe‐dose combination, and (e) if patients took the medication according to their medication list within at least for the last two days.

2.5. Statistical analysis

The analysis was carried out using the statistics program Statistica 13 and Microsoft Excel 2013. Data were reported as number and percent or as mean and standard deviation (SD) or as medians with inter‐quartile range (IQR). Based on the urine analysis, we classified patients as adherent opposed to nonadherent. Patients were classified as adherent if all prescribed antihypertensives were detectable by mass spectrometry. The adherent vs nonadherent group were next compared regarding the core variables of interest by independent t‐test or Mann‐Whitney U‐test for continuous variables where appropriate, chi‐squared test (used for categorical variables) and Fisher's exact test (used for dichotomous variables). For effect size estimates odds ratio (OR) and Cohen's d, subsequent converted to OR, were calculated for significant variables. Only to estimate OR in cells with zero, OR was calculated by adding 0.5 to each value (Haldanes‐correction).12 The threshold for statistical significance was chosen to be P < 0.05.

3. RESULTS

3.1. Patients

From October 2014 until June 2015, 120 patients could be screened and 100 patients were included into the study. Mean BP was 201 ± 18/106 ± 18 mm Hg. Hypertensive emergency was present in 13 patients (acute coronary syndrome [n = 9], neurological deficit [n = 3], pulmonary edema [n = 1]). Patients were admitted to the ED for a wide range of symptoms. In the ED, 70 patients (70%) had cardiovascular symptoms (eg, dyspnea [21%], (pre‐) syncope [5%], palpitation [11%] and angina pectoris [33%]) and in 53 patients (53%) neurological symptoms (headache [13%], tinnitus [2%], sight disorder [8%], vertigo or dizziness [30%]) could be detected. Overall, 46 patients required admission and inpatient treatment. HTN had been diagnosed a median of 11 (IQR 3‐20) years ago. In addition to the present HTN, 68 patients had further cardiovascular risk factors or disease. A number of 21 patients (21%) were active and 40 patients (40%) were former smokers. The sociodemographic as well as relevant anamnestic data are summarized in Table 1. Figure 1 shows the flowchart for study inclusion.

Table 1.

Patients’ characteristics at baseline

Patients (n)	100
Female n (%)	54 (54%)
Age (y)	67 ± 13
BMI (kg/m2)	29.1 ± 6.9
BMI ≥30 kg/m2, n (%)	37 (37%)
Duration of HTN (y)	11 (3‐20)
Initial BP (mm Hg)
Systolic	201 ± 18
Diastolic	106 ± 18
Concomitant diseases n (%)
Diabetes mellitus	22 (22%)
Hyperlipoproteinemia	29 (29%)
Coronary artery disease	27 (27%)
Congestive heart failure	6 (6%)
Occlusive artery disease	5 (5%)
Apoplex	14 (14%)
Chronic kidney disease	10 (10%)
History of smoking, n (%)	61 (61%)
History of antihypertensive treatment n (%)	86 (86%)
ACE‐inhibitora	39 (45%)
Angiotensin receptor blockera	38 (44%)
Renin inhibitora	2 (2%)
Calcium channel blockera	32 (37%)
β blockera	56 (65%)
Thiazide diuretica	36 (42%)
Loop diuretica	13 (15%)
Potassium‐sparing diuretica	2 (2%)
Aldosterone antagonista	3 (4%)
α blockera	5 (6%)
α‐2‐agonista	11 (13%)
Nitratea	2 (2%)

Values are mean ± SD, n (%), or median (IQR).

^aNumber of patients (in percent) under reported antihypertensive therapy (n = 86). BMI indicates body mass index.

Figure 1.

Flowchart. Adherence uncertainly assessable: permanent antihypertensive medication was applied in the emergency department prior to urine asservation

3.2. Antihypertensive treatment and adherence

Among the 100 included patients, 86 were pretreated with antihypertensive medication. Antihypertensive treatment was started in median 10 years prior to this event (IQR: 2‐20). On average, patients took 2.2 ± 1.5 antihypertensives. In 28 out of 86 patients (33%), fixed‐dose combinations were used. Relevant data regarding antihypertensive therapy regimens are summarized in Table 1.

At inclusion, 68 out of the 79 patients treated with antihypertensive medication (86%) indicated that they had taken their medications according to the prescription. There were 25 patients taking at least one antihypertensive drug which was not covered by the method in a sufficient way. However, in 11 of these 25 patients adherence could be assessed properly: 5 patients took fixed‐dose combination in which the combination ingredient was covered by the method, 6 patients were classified as complete nonadherent due to absent of all other detectable antihypertensive drugs. Therefore, there were 14 patients left in which adherence could not be assessed properly due to limitations of the method. These patients were excluded from analyses. In 24 cases, either the emergency physician prescribed antihypertensive medication in a time window that did not allow a valid assignment of the data of the urine analyzes to the intake behavior (n = 9) or due to the limitation of the method in detection of several antihypertensive drugs (cf. methods) (n = 14) adherence could not assessed certainly. Moreover, one patient did not have a medication plan and could not name the antihypertensive agent. Therefore, urine assessment data of 62 patients could be used for biochemical adherence assessment (Figure 1). Out of those 62 patients, 26 were completely adherent (42%) and 36 patients (58%) were nonadherent or partially nonadherent. Fifteen out of these 36 patients (41%) were completely nonadherent, that is, 24% of analyzed patients. The degree of adherence is shown in Figure 2A. The degree of adherence differed with respect to antihypertensive classes with highest adherence to β blockers and lowest to α blockers (Figure 2B). There was a high discrepancy between indirect and direct adherence measurement in the present study. A number of 68 out of 79 patients (86%) indicated complete adherence within the questionnaires, whereas urinary analysis revealed a number of 26 out of 62 patients (42%) to be adherent (P < 0.01). There were 4 patients (4%) in which antihypertensive drug could be detected which was neither listed on the medication plan from the practitioner nor indicated by the patients within the interview.

Figure 2.

Distribution of adherence level according to the percentage of medication detected. A, Degree of adherence in patients detected by mass spectrometry urine analysis. Percentage of prescribed drugs taken by patients. B, Degree of adherence regarding the particular antihypertensive classes (in percent of patients). ACEi: ACE‐inhibitors; ARB: Angiotensin receptor blocker; CCB: calcium channel blocker

3.3. Comparison of adherent and nonadherent patients

Nonadherent patients received treatment for significantly longer period for HTN (15 vs 10 years; P = 0.04), took more antihypertensive drugs (2.8 ± 1.1 vs 1.5 ± 0.7; P < 0.01) and received a larger total number of drugs (7.1 ± 3.4 vs 3.4 ± 1.8; P < 0.01) as well as were more frequently prescribed ≥3 antihypertensive drugs (20 patients vs 2 patients; P < 0.01). They differed from adherent patients with respect to antidepressants intake (6 vs 0; P = 0.04) and the prevalence of several co‐morbidities (diabetes mellitus [11 vs 0; P < 0.01], congestive heart failure [6 vs 0; P = 0.04], chronic kidney disease [6 patients vs 0 patients; P = 0.04]). Odds ratios for nonadherence are given for significant variables in Table 2.

Table 2.

Comparison between adherent and nonadherent patients

Parameter	Adherent n = 26	Nonadherent n = 36	P‐Value	Effect Size OR
Gender, n (%)			1.0
Female	16 (62%)	23 (64%)
Male	10 (39%)	13 (36%)
Age (y)	67 ± 11	68 ± 13	0.54
Marital status, n (%)			0.99
Single	1 (4%)	2 (6%)
Solid partnership	2 (8%)	2 (6%)
Married	15 (58%)	22 (61%)
Divorced	3 (12%)	4 (11%)
Widowed	5 (19%)	6 (18%)
Profession, n (%)			0.13
Independent	2 (8%)	1 (3%)
Employed	6 (23%)	5 (14%)
Unemployed	5 (19%)	2 (6%)
Retired	13 (50%)	28 (78%)
Systolic BP (mm Hg)	196 ± 19	200 ± 14	0.41
Diastolic BP (mm Hg)	109 ± 21	104 ± 18	0.28
Hypertensive emergency, n (%)	3 (12%)	4 (11%)	1.0
Duration of HTN (y)	10 (3‐15)	15 (5‐22)	0.04	2.7a
Hypertension ≥12 y, n (%)	9 (35%)	23 (64%)	0.04	3.3
Number of antihypertensive drugs	1.5 ± 0.7	2.8 ± 1.1	<0.01	11.8a
Fixed‐dose combinations, n (%)	10 (38.5%)	10 (27.8%)	0.42
Total number of drugs	3.4 ± 1.8	7.1 ± 3.4	<0.01	10.6a
≥3 Antihypertensive drugs	2 (8%)	20 (56%)	<0.01	15.0
Concomitant disease
Depression, n (%)	0 (0%)	6 (17%)	0.04	11.3
Diabetes mellitus, n (%)	0 (0%)	11 (31%)	<0.01	23.9
Coronary artery disease, n (%)	4 (15%)	11 (31%)	0.23
Congestive heart failure, n (%)	0 (0%)	6 (17%)	0.04	11.3
Chronic kidney disease, n (%)	0 (0%)	6 (17%)	0.04	11.3
Former apoplexy, n (%)	5 (19%)	5 (14%)	0.73
Hyperlipidemia, n (%)	5 (19%)	12 (33%)	0.26
BMI (kg/m2)	28.3 ± 7.5	29.9 ± 7.7	0.44
History of smoking, n (%)			0.24
Current smoker	7 (27%)	4 (11%)
Former smoker	8 (31%)	11 (31%)
Alcohol consumption, n (%)	5 (19%)	7 (19%)	0.98

Values are n (%), median (IQR) or mean ± SD. OR (Odds ratio) for nonadherence.

^aOR was converted from Cohen's d.

There were no differences with respect to gender, age, body mass index, smoking behavior, BP and severity of HTN‐C, treatment with fixed‐dose combination therapy, coronary artery disease, previous stroke or hyperlipoproteinemia. Data are summarized in Table 2.

3.4. Influence of concurrent medication, stimulants or nutritional components on HTN‐C

In 96 of 100 patients, medications other than antihypertensives, stimulants or relevant nutritional components were detectable. In 10 patients, antidepressants were found (selective serotonine reuptake inhibitor: n = 4; tricyclic antidepressant: n = 3; selective serotonine‐ and noradrenalin‐reuptake inhibitor: n = 1; noradrenergic and specific serotonergic antidepressant: n = 2; others (Opipramol): n = 1), in 8 patients glucocorticoids and in 33 patients nonsteroidal anti‐inflammatory drugs (NSAID: Ibuprofen: n = 20; Diclofenac: n = 8, Ketoprofen: n = 1; Naproxen: n = 2; Indometacin: n = 1; Etoricoxib: n = 1). Moreover, in five patients, intake of low‐dose aspirin could be demonstrated. In 24 of the 33 cases (73%), NSAIDs were over‐the‐counter medications and were not prescribed by a physician. Moreover, drugs with uncertain influence on BP such as metamizole (n = 8), metoclopramide (n = 6), paracetamol (n = 5), and diazepam (n = 3) were also detected.

As potential BP influencing stimulants or nutritional components we detected caffeine (n = 92), nicotine (n = 20) and licorice (n = 10).

4. DISCUSSION

Previous retrospective studies13, 14 as well as a one prospective study10 have reported that nonadherence to medication is an important risk factor for HTN‐C. Though it is known that indirect methods for assessing adherence, such as assessment by clinicians, questionnaires, pill count or prescription refill, do not provide reliable results, none of these studies investigated drug adherence by a direct method and all of them thereby failed to report an accurate prevalence. This is the first study showing a high prevalence (58%) of nonadherence in patients with HTN‐C using a direct biochemical adherence measurement. Taking into account that there were 18 patients who refused to participate, one might assume that the rate of nonadherence could even be higher. Of note, 89% of nonadherent patients had taken less than 50% of their antihypertensive drugs.

The proven nonadherence rate is a little higher compared to a recently published cross‐sectional study of patients with previously known HTN presenting in the ED for any reasons. However, due to lack of a comparator group within the present trial, the interpretation requires caution. Further case‐controlled studies are needed to confirm the relationship between nonadherence and the occurrence of HTN‐C. Based on blood assay results, 79% of patients prescribed <3 antihypertensives were completely adherent, and respectively 59% of patients who were prescribed ≥3 antihypertensives were nonadherent.18 However, the study did not differentiate between HTN and HTN‐C,18 whereas patients included in that study showed mostly controlled BP values with a mean SBP of 137 ± 23 mm Hg,18 in the present sample, all patients fulfilled the criteria of HTN‐C and showed a distinctly higher mean SBP of 201 ± 18 mm Hg.

The choice to use a method to detect adherence rather than another depends on multiple factors, including reliability, sensitivity to white‐coat adherence, educational value, local facilities, long‐term feasibility, patient profile, and financial resources. However, indirect methods are simple, cost‐saving, time‐efficient and they imply a reasonable workload. Because of poor sensitivity, they are heavily dependent on patient behavior, affected by social desirability and recall biases.19 However, using self‐reports as a simple way to assess adherence, showed in the present as well in previous studies a strong discrepancy to the applied objective method and clearly confirms the limitations of this procedure.20 Direct methods show a higher degree of reliability and accuracy than indirect methods, but are more expensive and need more resources.19 The most accurate objective method represents measurement of drug/metabolite levels in body fluids. In HTN patients, the most widely used material for adherence testing are blood and urine samples.19 Thereby high‐performance chromatography coupled with a sensitive detector such as mass spectrometry is considered to be the reference analytic technique.19, 21, 22 Regarding the healthcare cost for HTN‐C and its complications, the cost of about 60 € for mass‐spectrometry based adherence assessment appears manageable.20 In comparison with blood analysis, urine analysis has several advantages, whereas blood requires invasive sampling and accurate timing of blood drawing, especially for drugs with high clearance, urine samples provide a much larger detection window and a noninvasive sampling procedure. Considering the high sensitivity of the method, the nondetection of expected antihypertensive medication in urine is consistent with nonadherence to the prescribed antihypertensive lasting at least four half‐lives.23 Thus, for the most antihypertensive medications, the time window from the last intake is more than 24 hours.

In contrast to electronic drug monitoring, it allows to test adherence for a broad spectrum of drugs. Furthermore, the method allows for detection of potentially interfering drugs, stimulants or nutritional components. Nevertheless, this technique provides no exact information on the timing or dosing of drug intake, which is a major limitation of our study. We cannot rule out that individual doses were not taken (reduced persistence) or only in reduced amount (reduced quality of execution). Thus, the number of adherent patients could even be overestimated. A further major limitation of the study is the relative small cohort and white coat adherence cannot be excluded. On the other hand, mass spectrometry is not able to assess if the treatment is rigorously followed beyond the last days before admission to the ED. A further limitation is due to the clinical setting in which emergency medicine therapy lead to administration of antihypertensive medication according to the regular medication plan, making a reliable assessment of previous adherence impossible in several patients. Some nonadherent patients might furthermore have taken their medication when they first experienced symptoms of HTN‐C. Regarding that there was a high discrepancy between self‐reported adherence and adherence assessed by biochemical methods, it must be considered that patients in the ED presenting with HTN‐C are highly dependent on the health‐system at present and this might have affected the patients to answer in ways that make them seem more appealing to healthcare providers.

Considering the incidence of hypertensive crisis in the ED and the costs for direct adherence measurement, it is of interest to identify patients at risk for nonadherence and HTN‐C. The present data suggest that comorbidities such as diabetes mellitus, depression, chronic kidney disease and congestive heart failure may be associated with nonadherence as well as a number of ≥3 antihypertensive drugs per day and long‐lasting history of HTN. Thereby patients with those characteristics are of special interest for an adherence screening as well as nonadherence counseling.

In addition to nonadherence, we recognized ingestion of drugs, stimulants and/or food components suspicious for involvement in development of HTN‐C in a remarkable amount. Two drugs frequently detected in our cohort, that is, NSAIDs and glucocorticoids, can raise BP by volume retention.24 Regarding that 86% of the included patients had an established antihypertensive therapy and 73% of the detected NSAIDs were over‐the‐counter medication, the importance of exposure of concurrent NSAID treatment by direct methods in HTN‐C becomes clear. Glucocorticoids can have some mineralocorticoid activity resulting in mineralocorticoid receptor activation and increased sodium resorption leading to increased BP.24

Moreover, psychotropic medications are well‐known triggers of clinically relevant BP changes25, 26 which were frequently detected in our cohort. From those 10 detected psychotropic medication in our patients, the risk to develop HTN were classified in 4 as frequent (≥1/100 to <1/10), 3 as occasional (≥1/1000 to <1/100) and 3 as unknown.26 Therefore, it cannot be excluded that this co‐medication may have played in part a role in development of HTN‐C.

Regarding stimulants and food components two detected substances were especially remarkable. Licorice was detected in 10% of patients and caffeine in 92%. For both components, a BP increasing effect has been demonstrated.27, 28

There was a statistically significant increase in mean systolic blood pressure by 5 mm Hg and diastolic blood pressure by 3 mm  Hg after chronic ingestion of a product containing glycyrrhizic acid.27 The onset of HTN‐C after licorice ingestion, however, has been restricted to two case reports.29, 30 Caffeine can increase blood pressure through several mechanisms including rise of sympathetic activity with consecutive increases catecholamine release and antagonizes endogenous adenosine. In a meta‐analysis, patients who consumed 200‐300 mg of caffeine had an average rise in systolic and diastolic BP of 8 and 6 mm Hg, respectively.28 This increase was observed in the first hour after consumption and lasted >3 hours. However, regular consumption of coffee, obviously did not lead to a rise in BP.24 However, the consumption habits of our patients with regard to caffeine were not recorded in our study, so these data should be interpreted carefully. Despite missing difference in smoking behavior between adherent and nonadherent patients, detection of nicotine in urine was frequent with a prevalence of 20%. This is of special interest since smoking is shown to potentially rise blood pressure and negative survival outcome is associated with smoking in hypertensive emergencies.31, 32

5. CONCLUSION

Though there are several questionnaire‐based reports assuming insufficient adherence, this is the first study providing adherence data from direct adherence analysis suggesting that nonadherence to antihypertensive medication might have an impact to the occurrence of HTN‐C. Additionally, ingestion of co‐medication or nutritive factors with the potential of BP elevation might have contributed to the occurrence of HTN‐C. In order to prevent HTN‐C and concomitant end‐organ damage the present findings emphasize the importance for clinicians to pay attention to the issue of adherence, and co‐medication and nutritive factors.

Physicians in the ED should not automatically escalate medication in cases of HTN‐C with a pre‐existing antihypertensive medication. The high prevalence of nonadherence as well as the consumption of BP‐increasing drugs and food components should be an integral part of the therapy decision to avoid recurrence of HTN‐C with an effective long‐term BP control as well as a rapid BP lowering. This might help to reduce economic costs and contribute to a reduction in re‐hospitalization rate. However, larger studies are needed to confirm the present findings and to evaluate concepts to increase treatment adherence.

CONFLICT OF INTEREST

MW and MK have received a research grant from Novartis Pharma GmbH and CVRx. MK declares having received lecture fees from CVRx and Novartis and is member of the CVRx Barostim Hypertension Registry Steering Committee. CHL has received, lecture honoraria from Servier, Heel, and Novartis and an honorarium from Pfizer for serving on an advisory board. All honoraria have been unrelated to the topic of the present study. He also receives royalties from Hogrefe Huber publishers.

Wallbach M, Lach N, Stock J, et al. Direct assessment of adherence and drug interactions in patients with hypertensive crisis—A cross‐sectional study in the Emergency Department^a . J Clin Hypertens. 2019;21:55–63. 10.1111/jch.13448