Assessment of Patients' Views on Drug Benefits and Risks: An Interview Study with Cardiovascular Patients

Better and balanced information strategies supporting cardiovascular patients' adherence are required. Cardiovascular drugs have outstanding morbidity and mortality benefits. This can be counteracted by patients' perceptions of risks. Drug information should help the patient but not fuel unwarranted fears. We performed a cross-sectional survey of patients admitted to a cardiology ward. We evaluated (i) the patients' general benefit-risk estimation of their pharmacotherapy; (ii) views on benefits; (iii) views on risks; and (iv) information sources. Additionally, we assessed aspects of anxiety and depression with the Patient Health Questionnaire-4 (PHQ-4). (i) 67 patients (66%) rated expected drug benefits higher than potential risks. (ii) 72% of benefits motivated the patients to take their medication as prescribed. Patients more frequently mentioned surrogate markers as benefits than clinical benefits (p < 0.001). (iii) 56% of risks mentioned were perceived as bothersome and 35% as concerning. Risks were more often perceived as bothersome and concerning by patients with higher PHQ-4 scores (p=0.016). (iv) Physicians were the most frequent information source of benefits (92% of patients) and risks (45%), and pharmacy staff for 27% and 14%, respectively. Laymen or media served as sources of information on benefits in 39%, for risks in 40%, and package leaflets in 26% and 36%. 42% of the patients would like to receive more information on benefits versus 27% on risks. Our results suggest that knowledge of benefits motivates patients to take their drugs as prescribed. There is already good information on surrogate markers for process control with active patient involvement. However, a lack of knowledge still exists in relation to clinical benefits. Regarding risks, it has been shown that patients with higher PHQ-4 scores are more likely to be bothered or concerned. Both emphases on clinical benefits and individualization depending on PHQ-4 scores may be valuable resources for patient counseling to support adherence.


Introduction
Due to their high prevalence and mortality, cardiovascular diseases are of great interest in industrialized countries [1]. Teir corresponding pharmacotherapies often have outstanding efects on morbidity and mortality as they avoid myocardial infarctions and strokes. Tese efects are not immediately noticeable to patients. However, in order to achieve such positive efectiveness in the long term, they have to take their medication conscientiously over a long period of time. Terefore, it is not surprising that adherence decreases over time, which has been shown to especially afect patients with cardiovascular diseases [2]. In this regard, diverse forms of nonadherence occur, e.g., patients modifying their dosages or stopping their therapy altogether by themselves. While nonadherence represents a wellknown issue [3], better strategies for more balanced drug information could, on the one hand, provide the patient with knowledge about benefts and risks and, on the other hand, help to prevent unjustifed fears and support adherence.
Patients' attitudes toward drugs and, thereby, their adherence is infuenced by information. Te psychological mechanisms of nocebo and placebo need to be considered in this context [4][5][6]. For statin therapy, for example, it has been shown that fewer risks were reported in blinded than in open observational studies [7]. Studies have also shown that laymen's information can incite fears of risks [7,8]. Such anxiety is likely to decrease adherence, as shown for statins [9,10]. Due to the increasing access to digital information and patients' individual needs for information on drugs [11,12], healthcare professionals face additional challenges in information and counseling.
Adherence is primarily predicted by patients' attitudes toward their medicines [13]. It is further infuenced by two drug-related aspects. First, nonadherence is often caused by risk factors for events that have actually occurred or that the patients are afraid of [14][15][16], although they may never have experienced them themselves. Second, the patients should also be aware of the benefts of their pharmacotherapy. An imbalance between experienced or expected risks and benefts can negatively infuence adherence [13,17]. Strategies to support patients in their challenge to stay motivated should therefore include the patients' views on both drug risks and drug benefts. Furthermore, patients' individual characteristics were found to have an impact on adherence [4,11,18]. In this respect, depression and anxiety are of special interest as infuencing factors [19,20]. Tis is because depression represents frequent comorbidity in cardiovascular diseases and is associated with lower adherence and worse control of low cholesterol levels and blood pressure values [21,22]. One very short form for routine screening of the risk for depression and anxiety is the Patient Health Questionnaire-4 (PHQ-4).
We, therefore, investigated the patients' views on cardiovascular drug benefts and risks. Te relevant sources of information were reviewed and separated into professional and laymen sources. Tis way, this study should help to derive more tailored patient information strategies to support cardiovascular patients in their adherence.

Ethics Vote. Te Ethics Committee of the Medical
Faculty of Leipzig University (223/18-ek) approved the study protocol. Tis study conforms to the ethical standards of the 1975 Declaration of Helsinki. All patients participated voluntarily, and written informed consent was obtained from all patients prior to inclusion. In case the medication review or the interview indicated potential patient safety aspects concerning risks or nonadherence, the attending physician was informed immediately.

Setting.
Our survey was performed from July to/until October 2018 in a cardiology department with 46 beds in a university hospital ofering tertiary care. Te department ofers interventional cardiology and the diagnosis and treatment of cardiac insufciency, coronary heart disease, arrhythmias, and all other cardiovascular diseases.

Participants.
We consecutively enrolled all patients admitted to the cardiology department who were more than 18 years old and fully contractually capable. Further inclusion criteria were that patients took their medication on their own responsibility, had sufcient German language skills, and had been prescribed cardiovascular pharmacotherapy for at least four weeks. Exclusion criteria were patients' refusal to participate; cognitive impairment; communication barriers (e.g., blind and deaf ); and the absence of one of the inclusion criteria. If the patient shared a room with another patient who had already participated in the study and an information exchange was expected, he or she would not be included.

Study Design.
We performed a cross-sectional study by conducting semistructured face-to-face interviews with inpatients. Te interview was supplemented by a subsequent written assessment of anxiety and depression using the PHQ-4 [23].

Development of the Interview
Guide. An expert panel consisting of fve pharmacists developed the interview guide. Te interview guide was pretested incrementally with pharmacists (n � 6) and medical laymen (n � 11) not involved in study conception and conduction. Pretest techniques were used including sorting, paraphrasing, comprehension probing, category selection probing, and confdence rating. Due to the pretest, changes in vocabulary have been made in order to meet the patients' level of understanding. Terefore, benefts were referred to as "positive efects" and risks as "negative efects." Te pretest was additionally used to test the practicability of the semistructured interview. None of the participants taking part in the pretests were involved in the main survey. Te results of the pretests were not included in the data analysis.
Te interview included 4 main themes: (i) Patients' general estimation of therapeutic benefts and risks. (ii) Patients' views on drug benefts and whether they motivate the patient to take a drug as prescribed. Te patient could mention the benefts actively (without suggestion) or subsequently passively based on a predefned list of suggestions for common benefts of cardiovascular therapy. (iii) Patients' views on drug risks and whether they were perceived as bothersome, concerning, or causing the patient to take medication diferently than prescribed. Bothersome and concerning risks were additionally assessed because these parameters could negatively infuence adherence. Te patients could mention the risk actively (without suggestion) or subsequently passively based on a predefned list of suggestions for common risks of cardiovascular therapy. (iv) Used information sources on cardiovascular therapy about benefts and risks. Additionally, we asked Benefts and risks included both already experienced events and those the patients had heard or read about. Adherence was addressed indirectly by asking the patients whether the beneft motivates them to take their medication as prescribed or whether the risk caused the patients to take their medication diferently than prescribed, respectively.
2.6. Data Collection. All patients who met the inclusion criteria were personally invited to participate. All interviews were performed by the same person (pretrained pharmacist) from July 20, 2018, to October 10, 2018. Te interview was carried out face-to-face in a semistructured way to enable the patient to speak in an unbiased manner.
Before the study interview, the patient's medication was reviewed for cardiovascular medication using patient chart documentation. Additionally, the patient was asked about his or her current medication.
At the end of the interview, the patient was asked for sociodemographic data. Following the interview, the patient was invited to fll out the PHQ-4.
Te interviewer documented the patient's answers in written form. To minimize recall bias, notes were reviewed immediately after the interview and answers were coded and entered into an electronic data sheet (based on Microsoft Ofce Excel 2016). For quality assurance, every digitalized interview was checked by a second person (pharmacist) from the study team in order to exclude transfer errors.

Data Analysis.
Benefts were separated into surrogate markers and clinical benefts. To assess diferences between surrogate markers and clinical benefts, we frst counted the patients with the respective drugs (e.g., patients with antihypertonic medication for the surrogate marker "lowering blood pressure"). We assessed whether the patients mentioned the corresponding beneft of their medication, whether they mentioned it actively, and whether the mentioned beneft motivated the patients to take their medication as prescribed. We then performed a chi-square test with these data.
Mann-Whitney U tests were performed to assess differences in the PHQ-4 between patients with regard to their knowledge of risks and benefts. Te threshold for statistical signifcance was set at p < 0.05.
For each beneft and risk mentioned, patients were asked where they got this information from. Multiple answers were possible. Te analysis was carried out at the patient level. Once a patient mentioned a source of information, this was considered a "source of information used." No further weighing was carried out, taking into account the frequency of the mentions within an individual patient interview.
Te confdence interval of 95% was determined for each value.

Participants.
Of 554 patients admitted to the department of cardiology during the study period, 102 (18%) participated in the interview, and the response rate to an invitation to participate was 41%; a further 24% agreed to participate but could not be interviewed due to organizational problems ( Figure 1). Table 1 shows the demographic data of the participants.

Patients' Views on Drug
Benefts. 101 (99%) of the patients mentioned at least 1 drug beneft. In total, the patients mentioned 637 benefts within the median (Q25/ Q75) 6 (4/8) per patient. Patients associated 354 metioned benefts (56%) with a particular cardiovascular drug. As shown in Figure 2, the most frequently named beneft was "lowering blood pressure" (87% of the patients with antihypertensive medication). Patients more frequently named surrogate markers than clinical benefts (p < 0.001, Figure 2). 460 (72%) of 637 mentioned benefts motivated the patients to take their medication as prescribed. Te benefts motivated the patients to take their medication with a range of 68% (anticoagulation) to 86% (myocardial infarction International Journal of Clinical Practice prevention) of the respective patients. We found no difference in motivation between surrogate markers and clinical benefts (Figure 3, p � 0.781). Te beneft of "lowering blood pressure" was also mentioned as the most frequently observed cardiovascular beneft by 37% of the patients with corresponding medication (Figure 2). Patients most frequently associated the cardiovascular beneft "anticoagulation" with the respective drug (76% of the patients with anticoagulation medication). Additionally, to the predefned benefts, which are shown in Figures 2 and 3, patients mentioned further cardiovascular benefts: diuresis (28 patients; 19 motivated), supporting the cardiovascular system (21; 18), heart rate regulation (8; 6), protection from consequences of atrial fbrillation (4; 4), improvement of cardiac arrhythmia (3; 3), improvement of heart failure (2; 0); and symptom control in angina pectoris (2; 1).

Patients' Views on Drug
Risks. 91 (89%) of the patients reported at least 1 risk that they either had experienced or were aware of. In total (actively and passively), the patients reported 575 risks within the median (Q25/Q75) 4 risks (1/8) per patient. 143 (25%) of these risks were associated with a specifc cardiovascular drug by the patients. Te fve most frequently named risk categories were "gastrointestinal disorders" (95 risks with 54 patients afected), "mouth and throat complaints" (56 in 32 patients), "cardiac disorders/ hypertension/hypotension" (44 in 29 patients), "skin and subcutaneous tissue disorders" (40 in 32 patients), and "psychiatric disorders" (38 in 23). Te risk category most frequently associated with a particular drug was "increased bleeding tendency/hematoma/epistaxis" (86%). 56% of the mentioned risks were perceived as bothersome, 35% as concerning, and 7% caused the patients to take their medication diferently than prescribed. Out of the 10 most frequently reported risks, "psychiatric disorders" (particularly addiction, sleep disorders, and depression) were the only risk category that was perceived as concerning in more than 50% (Figure 4; Table 2). Patients who perceived risks as bothersome or concerning had higher PHQ-4 scores than patients who did not (p � 0.016; Table 3).

Information Sources.
Te interviewed patients most commonly received information about drug benefts (94 patients, 92%) and risks (46, 45%) from their physician ( Table 4). Te second most common information source among healthcare professionals was the pharmaceutical staf (27% vs. 14% of the patients). Te patients used laymen/ media and the package leafet frequently as sources of information on the benefts of drugs (39% vs. 26%). 42% of the interviewed patients wished for more information about the benefts of their medication and 27% about the risks (Table 4).

Discussion
Our study analyzed patients' views on the drug benefts and risks of cardiovascular therapy. We performed this study to gain information on how to better support patients with tailored information in their challenge to adhere to their cardiovascular therapy. In this regard, we found that clinical benefts were less frequently mentioned compared to surrogate markers. Both types of benefts motivate patients to take their medication as prescribed. In contrast to this, risks are commonly bothersome and concerning. Tis was particularly the case in patients with higher PHQ-4 scores. Patients most frequently used their physicians as an information source, particularly for information about the benefts of drugs. Tey expected to be better informed about the benefts of their therapy in particular. A stronger focus on clinical benefts should be considered by all healthcare professionals (i.e., physicians, but also pharmaceutical staf and nurses).
We decided not to tape-record the interviews in our study. Te prompt and thus nearly verbatim documentation of patients´responses were sufcient for the quantitative purposes of the study we aimed at. Condensation on relevant aspects of the evaluation was easily achieved. On the contrary, an anticipated tape recording could have decreased patients' willingness to participate or, in case of participation, could have led to bias and few free responses. Tis would have had a counterproductive efect on the results.

Need for Tailored Drug Risk Information. Patient in-
formation and counseling about drug risks are justifed by ethical and statutory provisions but are also a prerequisite for shared decision-making [30]. More than half of the mentioned drug risks were perceived as bothersome, with one-third as concerning. Tese alarmingly high rates of negative attributions of drug risks translate to a need for action. Terefore, our results support carefully individualized risk information, which is a challenge for health professionals. In two noticeable examples, we want to emphasize the identifcation of vulnerable patients or risks for particular diligent counseling and the patient's empowerment in dealing with a risk. First, patients with higher PHQ-4 Lowering blood pressure (n(total)=101) Surrogate marker Anticoagulation (n(total)=89) Surrogate marker Lowering blood glucose level (n(total)=39) Surrogate marker Lowering cholesterol level (n(total)=68) Surrogate marker Stroke prevention (n(total)=102) Clinical benefit Myocardial infarction prevention (n(total)=102) Clinical benefit Atherosclerosis prevention (n(total)=102) Clinical benefit

Myocardial infarction prevention (n=50)
Clinical benefit benefit motivates patient to take medication as prescribed n (%) benefit does not motivate patient to take medication as prescribed n (%) not specified n (%) Figure 3: Motivation of associated benefts to take medication as prescribed. n (total) refers to the respective patients who mentioned the corresponding beneft. Surrogate markers and clinical benefts did not difer signifcantly in their patients' motivation for adherence (p � 0.781).
scores, which go along with a higher tendency for anxiety and depression, were more likely to report risks as bothersome or concerning. In addition, the risk of "psychiatric disorders" was most frequently perceived as concerning. We did not fnd diferences in information demand in patients with diferent PHQ-4 scores. Nevertheless, our results support risk information strategies that identify vulnerable patients and adapt the information to meet individual requirements. Te PHQ-4 score seems to be a valuable short form to identify particular psychiatric vulnerabilities [31].
Second, "an increased tendency to bleed" was also one of the ten most frequently mentioned risks. Conspicuously, patients were hardly concerned about this risk, and in most cases, they were able to assign this risk to a specifc drug. Tis can be explained by the particular complex counseling procedures for anticoagulants because of their  status as high-risk drugs [28,30,32,33]. It can be assumed that healthcare professionals intensively discuss bleeding risks and their management with their patients. Te patients might then feel more competent in dealing with the risk. Tus, the risks of anticoagulants may be an example of information that is already balanced. It should be considered which strategies could be transferred to other drugs as well.   International Journal of Clinical Practice

Patient Information and Counseling in the Context of Laymen Information and Package Leafets.
Digitalization facilitates patient access to a wide range of diferent information sources. Unfortunately, it is often difcult to distinguish between professional and laymen information in the healthcare sector [34,35]. Laymen information has shown to incite patient concerns about risks [7,8]. Such concern can decrease adherence, as shown for statins [9,10,36]. It can be assumed that laymen media information more often focuses on risks and thereby contributes to the imbalance between risks and benefts estimation [8,37]. In this context, it is a reassuring result that the Internet is an information source for only one-ffth of patients, although we expect an increase in the future. Package leafets are another easy-access information source. Tus, they were used more often by our respondents, especially in cases of risk. Terefore, it is not surprising that package leafets can increase patients' concerns [23,38]. Tey are not the easiest source of information for the patient to understand. Even healthcare professionals misinterpret information on risk frequency and causality in package leafets [39]. Both information sources appeared unsatisfactory to the patients. Consequently, it can be expected that patients in general seem to appreciate more information [40]. Besides, certain patient characteristics could indicate a further need for intensifed or tailored information with respect to the individual patient [11,12]. We can further conclude that patients should not be left alone with drug information obtained from the Internet or package leafets. Healthcare professionals should be sensitized and acknowledge their role as a professional information source who can help to prevent misconceptions in their patients.
Tis applies not only to physicians but also to nursing staf and pharmacists who are involved in patient care.

Limitations.
Our results are restricted to inpatients of a single hospital who probably had rather good access to healthcare professionals. We evaluated the patients' selfassessed motivation to take their medication as prescribed and whether risks were perceived as bothersome or concerning or causing the patient to take medication diferently from the prescription. We thereby did not diferentiate by the respective extent. We can only conclude motivation to adhere but not the actual, objective adherence of the patient, which was not the focus of our study. We did not analyze the causality of the mentioned risks. Te lists with predefned suggestions of benefts and risks were tailored to cardiovascular drugs and covered frequent or important benefts or risks. Nevertheless, it could not cover all possible benefts and risks. Interpreting the results is further limited because the fnal interview was not validated; however, it was pretested in advance. Furthermore, it must be noted that due to the semistructured form of the interview, slight variations in the order of questions were accepted.

Conclusions
Cardiovascular patients rated the expected benefts of their drugs-in most cases based on known surrogate markers-higher than potential risks. Knowledge of benefts frequently motivates patients to take drugs as prescribed. In total, they are more interested in further information on benefts than on risks. Nevertheless, the provided insight Table 4: Sources of information: number of patients who use and request certain sources of information about benefts and risks concerning their drug therapy (n total � 102 patients). Multiple categories were possible.
Patients' answers regarding drug benefts n (n (%); 95%-CI) Patients' answers regarding drug risks n (n (%); 95%-CI) Where did you get the information from? into patients' views emphasizes the need for additional information on clinical benefts rather than on surrogate markers and, to a lesser extent, also on drug risks. More precisely, our fndings frstly indicate that especially clinical beneft information is not yet fully utilized by healthcare professionals as a valuable resource to motivate the patient to take their medication. Tey also suggest that the higher PHQ-4 score in patients who perceive drug risks as bothersome or concerning should be considered in order to individualize information and counseling about risks.

Data Availability
Due to ethical concerns regarding patient privacy, supporting data cannot be made available.

Conflicts of Interest
Te authors declare that there are no conficts of interest regarding the publication of this paper.