Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice J Hum Hypertens . 2026 Mar 23;40(4):333–341. doi: 10.1038/s41371-026-01135-8 Search in PMC Search in PubMed View in NLM Catalog Add to search How do patients with primary hypertension assess different endpoints of their treatment? a survey using analytic hierarchy process Charalabos-Markos Dintsios Charalabos-Markos Dintsios 1 Medical Faculty, Institute for Health Services Research and Health Economics, Centre for Health and Society, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany Find articles by Charalabos-Markos Dintsios 1, ✉ , Nadezda Chernyak Nadezda Chernyak 1 Medical Faculty, Institute for Health Services Research and Health Economics, Centre for Health and Society, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany Find articles by Nadezda Chernyak 1 Author information Article notes Copyright and License information 1 Medical Faculty, Institute for Health Services Research and Health Economics, Centre for Health and Society, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany ✉ Corresponding author. Received 2025 Nov 4; Revised 2026 Feb 25; Accepted 2026 Mar 5; Issue date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . PMC Copyright notice PMCID: PMC13068516 PMID: 41872346 Abstract Effects of antihypertensive therapy are estimated in clinical trials. There is a need to prioritize the endpoints according to patients’ preferences. 26 patients from two regions of Germany rated in 2019 their preferences regarding the importance of various endpoints of hypertension treatment (Mortality, Myocardial infarction, Stroke, Heart failure, and subdivided Adverse events) by a pairwise comparison of individual endpoints. Analytic Hierarchy Process (AHP), a multi-criteria decision analysis method was used to generate relative weights for each endpoint. The robustness of the results was defined by means of consistency. The elicitation yielded the following aggregated group weights: Stroke 0.320, Mortality 0.297, Myocardial infarction 0.202, Heart failure 0.119, and Adverse events 0.062, subdivided in Dyspnea, Pain, Edema, and Cough. The overall consistency reached for efficacy endpoints a consistency ratio below 0.1 (safety endpoints = 0.04) without exceeding established limits. In all sensitivity analyses but one, no rank reversal was observed, and Stroke was rated highest. Individual weights varied extensively. Some participants weighted Mortality (0.021–0.686) higher than Stroke (0.078–0.615) and Heart failure (0,021–0,469) higher than Myocardial infarction (0,047–0.431). Individual inconsistency exceeded the limits in almost half of the cases, with gender, therapy duration, and therapeutic scheme being explaining variables for inconsistency within binary logistic regression models. AHP can be used to obtain preferences of patients with primary hypertension for effectiveness and safety endpoints. Preference elicitation could provide important information for drug assessment (group weights) and shared decision-making (individual weights) following the concept of patient-centeredness at system and patient level. Subject terms: Health care, Renovascular hypertension Introduction Hypertension is the most significant controllable risk factor for cardiovascular diseases, accounting for an estimated 57% of all cardiovascular-related deaths [ 1 , 2 ]. Moreover, hypertension is strongly associated with an increased risk of Stroke , Myocardial infarction , and all-cause Mortality . The guidelines of the European Society of Cardiology (ESC) recommend different drug classes with evidence on clinical outcomes in the target population (ACE inhibitors, ARBs, dihydropyridine CCBs, diuretics, and beta-blockers) to achieve a target of blood pressure-lowering treatment of 120 – 129/70 – 79 mmHg [ 3 ]. Patient-centeredness in health policy is increasingly recognized as a major value policy makers wish to adhere to. It requires a solid evidence-base for patient-related issues in healthcare and patient-oriented assessment of health interventions. Patient preference studies may contribute to that evidence base. Among several methods that have been used to elicit patient preferences concerning endpoints , Analytic Hierarchy Process (AHP), a technique for multi-criteria decision analysis (MCDA) [ 4 ], fulfills the respective required features. It has been shown that the AHP works well to support a variety of health care decisions [ 5 , 6 ]. The AHP assists patients and health care professionals in grasping complex decisions by guiding them to prioritize and has been applied to facilitate a wide variety of health care decisions with multiple, even competing, outcomes [ 7 ]. Unlike available patient preference studies targeting clinical endpoints for chronic conditions such as rheumatoid arthritis [ 8 ], chronic hepatitis C [ 9 ] and diabetes mellitus [ 10 ], there are no studies eliciting endpoint preferences in patients with primary hypertension. The primary aim of the study was to investigate how patients suffering from high blood pressure prioritize efficacy and safety endpoints for the assessment of pharmaceutical interventions in primary hypertension treatment. Methods Trial design In this cross-sectional study, patients suffering from high blood pressure completed a self-reported AHP survey on efficacy and safety endpoints of pharmaceutical antihypertensive treatment. (Sub-)Criteria of the survey were gathered by systematic literature search and validated within a focus group. The survey was pre-tested by the same focus group. Participants and recruitment The inclusion criteria for participation were defined as i) a diagnosis of primary hypertension and ii) current use of medication for its treatment – regardless of the number of medications taken daily – as well as the presence of iii) sufficient language skills and iv) cognitive abilities. Furthermore, patients had to be v) of adult age. Exclusion criteria were defined as i) other forms of hypertension and ii) further (multi-)morbidity in the cardiovascular and cerebrovascular area, to obtain statements on patients’ endpoint weighting triggered solely by primary hypertension . The patients should iii) not have suffered a cardio- or cerebrovascular event, i.e. not be treated with antihypertensives for secondary prevention, to avoid biased endpoint weightings based on their own experience of the event. This did not apply to Adverse events , which are medication-related and thus within the experience of any patient receiving anti-hypertensive medication. The survey setting for the application of the AHP questionnaire as part of the structured interview consisted of three pharmacies (one in North-Rhine-Westfalia and two in Bavaria) to guarantee regional representativeness. Face-to-face interviews were conducted in a separate customer area, the existence of which is common in German pharmacies. A convenience sampling was followed from March to June 2019. Sample size There is no established minimum sample size for AHP in the literature [ 11 , 12 ]. A review of different AHP survey applications in healthcare depicted a range of one to 1300 participants [ 7 ]. We referred to a sample size in AHP surveys with healthcare relevance [ 13 – 18 ]. A target sample size of at least 25 participants was set to ensure a robust dataset for the AHP. Analytical hierarchy process The Analytic Hierarchy Process (AHP) is a structured decision-making method designed to analyze complex problems by systematically incorporating subjective preferences into the decision-making process [ 7 , 19 ]. AHP is classified as a compositional method within Multi-Criteria Decision Analysis (MCDA). To implement AHP, an overarching goal is defined first, followed by the hierarchical structuring of decision criteria. Participants perform pairwise comparisons to evaluate the relative importance of criteria, using a standardized numerical rating scale (ESM Fig. 1 ). These comparisons provide the basis for calculating numerical weights for each criterion. Finally, the weighted scores are aggregated by maximum eigenvalue of the matrix of pairwise comparisons to determine the rankings of the criteria. For more details on the matrix algebra implemented in AHP see Dolan et al. 1989 [ 20 ] and Supplementary Material (ESM Text 1 ). The main criteria are categories of outcomes that are conceptually related (Fig. 1 ). Since only outcomes were prioritized by patients and hypertension treatments were not compared, the hierarchical levels only include in this case the objectives and (sub-)criteria without treatment alternatives. Endpoints were identified from a systematic literature review about different antihypertensive drugs as first line therapy in patients with primary hypertension [ 21 ] and from the interviews with patients when testing the feasibility and use of the developed specific AHP questionnaire in a respective focus group. The endpoints were selected according to the following rationale in respect to the methodological constraints: the endpoints needed to be mutually exclusive, clear, comprehensive, and to be of importance within the same order of magnitude. The literature review led to the selection of the following endpoints : Mortality, Myocardial infarction, Heart failure, Stroke and Adverse events , consistent with established clinical endpoints in hypertension research [ 3 ]. Fig. 1. AHP structure. Open in a new tab The figure shows the AHP decision structure with two levels: the 1st level includes the efficacy and aggregated safety endpoints, whereas the 2nd level refers to the individual safety endpoints. Statistical analysis The analysis comprised four phases. In the first phase the values of the weighted efficacy endpoints (first hierarchy level) and safety endpoints (second hierarchy level) were derived by the principal right eigenvector approach for each participating patient after completion of the AHP survey instrument. Thereafter, the consistency ratios were calculated to check for potential inconsistencies due to violation of one or more of the quality criteria for AHP, i.e., reflexivity, reciprocity, and transitivity on an individual level. In the next phase the aggregated weights were calculated again with the right eigenvector approach by averaging the individual results through geometric means and consistency was checked on an aggregated group level. In the third phase based on different stratification criteria (consistency and health related quality of life) subgroup analyses were implemented to check for robustness of the aggregated weights (extent and hierarchical order) and potential rank reversals. In the final (fourth) phase, the explorative binary logistic regression model was calculated to gain more insights on potential factors impacting the consistency of the preference weights. Statistical significance of the model and its coefficients as well as explainable variance were determined. The binary regression model was chosen, since the dependent variable is per definition dichotomized based on a consistency threshold. Multicollinearity was tested by means of variance inflation factor (VIF). The principal right eigenvector was calculated using matrix multiplication including respective iterations and the generation of random matrices to derive the random index (R.I.) according to Dolan et al. [ 20 ] with MS excel [ 22 ]. The different regressions were conducted using the regression module of SPSS (version 16). Elicitation instrument The applied instrument consisted of a questionnaire developed specifically for the survey, which was administered to the participants using a structured interview. The structured interview primarily served to ensure compliance with the inclusion and exclusion criteria and to allow questions to the interviewer in case of (cognitive) overload of the participants. Postal or digital transmission of the questionnaire integrated into the structured interview was omitted to avoid missing data and the completion by third parties (family members or life partners). The endpoints were described in layman’s language. Various public patient information websites served as sources, primarily including the German Institute of Quality and Efficiency in Health Care (IQWiG) website https://www.informedhealth.org/ . The description of the efficacy and safety endpoints is given in Supplementary Material (ESM Text 2 ). The interviewer was trained to convey consistent content in relation to the relevant descriptions. First, before using the survey instrument, it was tested in a focus group session with five affected patients regarding (i) required language skills, (ii) cognitive effort, (iii) provided background information, and (iv) the participants’ potential willingness to weigh up options. Second, to ensure a common understanding of the pairwise comparisons, participants were asked to repeat the introductory example and clarifying explanation were offered in case of misconceptions. Excerpts of the questionnaire used in the AHP survey, and the AHP rating scheme are presented in Fig. 2 . Fig. 2. Excerpt from the AHP survey on endpoints. Open in a new tab The excerpt includes exemplary one question asked to the participants regarding their pair-wise importance ratings when comparing respective endpoints. Results Demographics Of the 28 patients who gave their consent, only the data of 26 patients could be analyzed, as one patient failed for cognitive reasons despite the interviewer’s accompanying survey and another patient refused to participate in the corresponding endpoint assessments. 87 patients contacted refused to give their consent due to time constraints. The sociodemographic characteristics of the anonymized analyzed participants, as well as information on drug therapy and its duration, can be found in Table 1 . The average age is 61 ± 12 years (50% male) with a treatment duration of 9.8 ± 6.9 years (65% of those received combination therapy). Table 1. Baseline characteristics. Nr. Age (years) Gender Treatment duration (years) Treatment scheme VAS Utility (%) 1 72 m 20 mono 60 2 72 m 8 combi 60 3 69 m 18 combi 50 4 64 f 6 combi 75 5 86 f 21 combi 40 6 54 m 7 combi 80 7 67 f 20 combi 80 8 55 f 3 mono 95 9 45 m 1 mono 75 10 67 m 16 combi 70 11 58 m 20 mono 70 12 69 f 15 combi 50 13 53 f 10 mono 80 14 37 f 4 mono 70 15 77 m 22 combi 50 16 46 m 1 combi 65 17 66 m 7 combi 90 18 61 f 5 combi 90 19 71 m 11 combi 50 20 40 m 1 mono 80 21 43 f 2 mono 90 22 52 f 5 combi 80 23 63 f 10 mono 90 24 71 m 9 combi 90 25 69 f 10 combi 70 26 65 f 4 combi 80 Σ 61 ± 12 50%m 9.8 ± 6.9 65% combination 72.3 ± 15.3 Open in a new tab Health-related quality of life (HRQoL) was used as a discriminatory criterion. The visual analogue scale (VAS) from the validated and standardized generic index instrument EQ-5D-3Lwas used for this purpose. The average utility values (in %) of the analyzed survey population derived from the VAS was 72.31 ± 15.31 (Table 1 ). 23% (6/26) of the participants stated a VAS utility greater than or equal to 90, which is considered to be a very high utility (almost ceiling effect). On the other hand, 27% (7/26) stated a rather low VAS utility of less than or equal to 60. The patient characteristics indicate that participants were suffering from primary hypertension of varying severity, but all of them were event-naive regarding cerebrovascular and cardiovascular events. AHP survey The AHP procedure yielded the following aggregated group results for preference-based weighting of the endpoints surveyed (global weight) in descending order (Table 2 ): 1. Stroke (0.320). 2. Mortality (0.297). 3. Myocardial infarction (0.202). 4. Heart failure (0.119). 5. Adverse events of the therapies (0.062), subdivided again (global weight, local weight) in 5a. Pain (0.020, 0.316), 5b. Cough (0.008, 0.122), 5c. Dyspnea (0.024, 0.397), and 5 d. Edema (0.010, 0.165). Table 2. Individual and group weights and consistency. Nr . CR Endpoints CR Adverse Events Mortality Myocardial Infarction Stroke Heart Failure Adverse events Pain Dyspnea Cough Edema Individual-Wi 1 1.49* 0.16 0.164 0.187 0.200 0.284 0.165 0.659 0.076 0.171 0.094 2 0.13 0.15 0.510 0.146 0.266 0.036 0.042 0.394 0.408 0.061 0.137 3 0.13 0.23 0.290 0.131 0.507 0.036 0.037 0.610 0.260 0.039 0.090 4 0.14 0.08 0.417 0.125 0.395 0.031 0.031 0.698 0.182 0.060 0.060 5 0.12 0.11 0.046 0.166 0.615 0.085 0.088 0.673 0.152 0.076 0.099 6 0.78* 0.19 0.087 0.134 0.295 0.469 0.015 0.178 0.514 0.041 0.267 7 0.12 0.23 0.207 0.095 0.501 0.149 0.049 0.188 0.605 0.091 0.116 8 0.19 0.13 0.534 0.121 0.121 0.202 0.022 0.626 0.266 0.054 0.054 9 0.01 1.11* 0.212 0.240 0.240 0.282 0.027 0.366 0.283 0.327 0.024 10 0.00 0.00 0.686 0.078 0.078 0.080 0.078 0.250 0.250 0.250 0.250 11 0.00 0.22 0.667 0.083 0.083 0.083 0.083 0.427 0.218 0.137 0.218 12 0.28* 0.00 0.596 0.088 0.203 0.036 0.077 0.444 0.444 0.560 0.560 13 0.48* 0.50* 0.325 0.121 0.448 0.089 0.018 0.225 0.675 0.075 0.025 14 0.00 1.81* 0.243 0.243 0.243 0.243 0.027 0.355 0.190 0.166 0.289 15 0.19 0.19 0.087 0.158 0.432 0.272 0.051 0.294 0.173 0.071 0.462 16 0.20 0.12 0.574 0.121 0.203 0.075 0.027 0.150 0.509 0.070 0.271 17 0.19 0.44* 0.130 0.259 0.428 0.146 0.037 0.033 0.665 0.216 0.086 18 0.40* 1.23* 0.609 0.092 0.236 0.039 0.024 0.129 0.573 0.140 0.158 19 0.20 0.38* 0.067 0.381 0.397 0.035 0.120 0.223 0.656 0.085 0.036 20 0.33* 1.91* 0.622 0.088 0.229 0.021 0.040 0.285 0.337 0.018 0.359 21 1.21* 0.50* 0.296 0.322 0.347 0.023 0.012 0.075 0.225 0.025 0.675 22 0.72* 1.46* 0.021 0.431 0.213 0.152 0.183 0.299 0.398 0.056 0.246 23 0.37* 0.94* 0.262 0.143 0.282 0.105 0.208 0.146 0.354 0.212 0.288 24 0.57* 1.82* 0.041 0.240 0.231 0.212 0.275 0.173 0.314 0.210 0.303 25 0.50* 0.88* 0.592 0.246 0.102 0.042 0.018 0.109 0.208 0.628 0.054 26 0.41* 3.33* 0.608 0.047 0.222 0.104 0.019 0.252 0.324 0.249 0.175 Endpoints Mortality Myocardial Infarction Stroke Heart Failure Adverse Events Pain Dyspnea Cough Edema Group-W i W all 0.297 0.202 0.320 0.119 0.062 0.316 0.397 0.122 0.165 W cons 0.303 0.182 0.351 0.104 0.060 0.359 0.373 0.121 0.147 Open in a new tab *) Consistency ratio > 0.2 CR consistency ratiom W i weight i, W all weight derived from all participants, W cons weight derived only from consistent participants. The observed group inconsistency remained below the limit of 0.1 at both hierarchical levels. Stroke emerged as the most important endpoint, followed by Mortality . Myocardial infarction and the endpoint Heart failure followed. Adverse events , including severe ones such as Dyspnea , played a less important role from an aggregated group perspective. At the second hierarchy level, where Adverse events were differentiated and compared individually, Dyspnea predominated, followed by Pain . Edema and Cough played a less important role among respondents (Fig. 3 ). Fig. 3. AHP results endpoint weights. Open in a new tab The hierarchy tree contains the global weights for efficacy and aggregated safety endpoints at the 1st hierarchy level and the global (overarching criteria) and local weights (sub-criteria) for the individual safety endpoints at the 2nd hierarchy level. On the other hand, individual weights varied extensively (Table 2 ). Some participants weighted Mortality (0.021–0.686) higher than Stroke (0.078–0.615) and Heart failure (0,021–0,469) higher than Myocardial infarction (0,047–0.431) deviating more or less from the aggregated group weighting order. Furthermore, a few participants weighted Adverse events (0.015 – 0.275) very highly, even exceeding the individual weights for Mortality or Stroke . At the second hierarchical level of the different Adverse events Pain (0.033 – 0.698), Cough (0.018 – 0.628), Dyspnea (0.076 – 0.675), and Edema (0.024 – 0.675) the results were heterogeneous with broad and overlapping ranges. As the participants were under antihypertensive treatment (mono or combination therapies), this mainly reflects their experience regarding Adverse events according to their feedback and the feedback of the focus group during the pre-test of the survey. According to Saaty [ 11 ], an evaluation matrix is sufficiently consistent if the CR ≤ 0.1. Other authors, however, suggest for complex hierarchies and those that are not amenable to immediate modification during the evaluation process, a limit of 0.2 [ 23 – 25 ]. This was also the situation in the present analysis, as the CR could not be determined in parallel during the interview, which would have disrupted the interview process and jeopardized the survey in the chosen form. Subgroup analysis When considering individually consistent weightings (CR ≤ 0.2), the following endpoint order resulted: Stroke 0.351, Mortality 0.303, Myocardial infarction 0.182, Heart failure 0.104, and Adverse events 0.060. The information on the respective individual consistency ratios and the comparison of the weights between the entire survey population and the consistent participants can be found in Table 2 . Overall, 61% (16/26) of participants showed inconsistencies that exceeded the threshold of a CR ≤ 0.2 at least at one or both hierarchy levels. 11% (3/26) showed inconsistencies only at the first hierarchy level, 15% (4/26) only at the second level, and 34% (9/26) at both levels. Assuming that the first hierarchy level is more important for the research question than the second (weighting of the Adverse events relative to each other), 46% (12/26) were inconsistent at least at this level. The order of weighting did not change in the results obtained, but Stroke proved to be even more important for patients with consistent ratings. In addition, subgroup analysis based on HRQoL was performed. Stratification regarding disease severity using HRQoL was undertaken based on the utility values from the VAS. The value of 75 on the VAS, which represents the median of the respondents’ results, was selected as the threshold for dichotomization into severe and less severe disease states. The analysis was carried out across all patients and subsequently exclusively across consistent patients (CR ≤ 0.2) at the first hierarchy level. Of the 26 patients included, 12 patients reported a VAS utility score of less than 75 (40–70). The remaining 14 patients showed a range of 75 to 95. The respective weights of the patients stratified according to the VAS utility scores are shown in Table 3 . Table 3. Preference weights stratified according to VAS utilities. Mortality Myocardial Infarction Stroke Heart Failure Adverse Events C.R. N VAS-Utility VAS-U <75 All 0.334 0.192 0.301 0.106 0.067 0.01 12 VAS-U <75 Consistent 0.300 0.192 0.335 0.104 0.069 0.02 9 VAS-U >75 All 0.286 0.193 0.335 0.132 0.055 0.03 14 VAS-U >75 Consistent 0.300 0.178 0.338 0.149 0.035 0.07 5 Open in a new tab C.R . consistency ratio. In 3 of the 4 constellations (VAS-U < 75 consistent, VAS-U > 75 all and VAS-U > 75 consistent) Stroke receives a higher weight than Mortality . Only when considering all patients with a VAS utility value of less than 75 Mortality (0.334) received a higher weight than Stroke (0.301), which in turn cannot be maintained when only consistent weightings of the patients are included. The reason for this is the three times higher rating of Mortality compared to Stroke in two of the three inconsistent participants with a VAS utility of less than 75, who, as outliers, distort the overall result in this direction. The order of the other endpoints does not change depending on HRQoL compared to the total sample. It can therefore be concluded that the stratification criterion of the median of the VAS utility shows no influence in the sense of effect modification on the weighting results. Thus, a relatively robust result can be assumed across all patients regarding the order of the endpoint weighting. On the other hand, the relatively high proportion of inconsistent patients seems remarkable, which is why an attempt was made to analyze the individual inconsistencies using regression calculations despite the relatively small survey group. Regression analysis The relatively strong inconsistency at both the primary and secondary hierarchical levels of the AHP led to the consideration of identifying explanatory variables for this phenomenon. Inconsistency at the individual and group levels arises within the AHP because of the respondents’ failure to comply with transitivity. In the binary logistic regression model patient characteristics such as age, gender, duration of medication and treatment regimen as well as the VAS utility values were included (Table 4 ). The Variance Inflation Factor (VIF) indicated a low correlation among the independent variables included in the models. Table 4. Logistic Regression Model. Logistic Regression Model 1 st AHP Hierarchic Level (Hypertension Endpoints) Model Summary Omnibus Test Variable Regression Coefficient SE Significance OR CI 95% -2 Log-Likelihood Cox - Snell R² Nagelkerkes R² X² df Significance VAS Utilities -0.045 0.038 0.233 0.956 0.887-1.030 30.495 0.187 0.250 5.395 5 0.370 Gender 1.005 0.885 0.256 2.733 0.482-15.486 Age -0.030 0.069 0.661 0.970 0.847-1.111 Duration 0.040 0.112 0.721 1.041 0.836-1.295 Scheme -0.503 1.271 0.693 0.605 0.050-7.306 Intercept 4.612 5.009 0.357 100.718 Logistic Regression Model 2 nd AHP Hierarchic Level (Adverse Events) Model Summary Omnibus Test Variable Regression Coefficient SE Significance OR CI 95% -2 Log-Likelihood Cox - Snell R² Nagelkerkes R² X² df Significance VAS Utilities -0.035 0.044 0.429 0.966 0.887-1.053 24.519 0.358 0.477 11.524 5 0.042 Gender 0.607 1.022 0.553 1.834 0.247-13.597 Age -0.058 0.079 0.464 0.944 0.808-1.102 Duration 0.233 0.146 0.111 1.262 0.948-1.680 Scheme -1.234 1.492 0.408 0.291 0.016-5.422 Intercept 3.987 5.340 0.455 53.980 Open in a new tab As a result, due to the small number of participants ( n = 26), the corresponding odds ratios are not statistically significant and can therefore be interpreted more as trends. Among the different calculations, the highest R-squared value (calculated here as Nagelkerke’s R-squared for logistic regressions) was achieved in the logistic regression model for consistency in Adverse events (second hierarchy level) with 0.477. Thus, it can explain a variance of the dependent variable (consistency) across all independent variables combined (gender, age, duration of therapy, treatment regimen, and VAS utility values) of almost 50%. In the regression model for the first hierarchy level ( hypertension efficacy endpoints ), a R-squared = 0.250 was achieved. Nevertheless, the R-squared is relatively low, and thus the model can explain little of the variance. Regarding the assessment of model quality using the omnibus test of the model coefficients, the regression model for the first hierarchy level is not significant (0.370). Ultimately not all selected variables influence the consistency as a dependent variable, whereas the model for the second hierarchy level achieves significance (0.042). In the model for the first hierarchy level, only gender (OR = 2.733) and the treatment regimen (OR = 0.605) show a strong influence. When looking at the model for the second hierarchy level, in addition to gender (OR = 1.834) and the treatment regimen (OR = 0.291), the duration of treatment also impacts consistency (OR = 1.262), albeit to a lesser extent than the first two variables. This can be hypothetically explained by the fact that with longer therapy more experience has been gained regarding drug-induced Adverse events . Furthermore, interactions may also occur with combination therapies. Hence, more consistent results are obtained in the survey at the second hierarchy level, although the treatment regimen already had a strong influence at the first hierarchy level. For both hierarchy levels, a more consistent response tendency depending on the male gender is recognizable – albeit less pronounced at the second hierarchy level. The regression analysis was completely exploratory in nature and aimed only at identifying potentially explanatory variables. Evaluation of the AHP survey instrument Just over half of the participants (14/26, 54%) reported the AHP survey to be ‘rather easy’ to complete whilst the remaining participants reported that they found it to be either ‘moderately difficult’ (7/26, 27%) or ‘very difficult’ (5/26, 19%) to complete. Apart from the two participants not included in the AHP survey, all other participants reported no further problems of understanding regarding the survey method, the description of the (sub-)criteria to be considered at the first hierarchy level and at the second hierarchy level, and the assessment of HRQoL. The mean completion time reached 30.1 ± 6.4 min, and thus unexpectedly a relatively homogeneous result. Discussion Stroke occupies a distinct role in the prioritization among endpoints when compared to Mortality . In a best-worst scaling by Aschmann et al. [ 26 ] Stroke was the most important outcome to avoid among patients with multiple chronic conditions, reflecting its severe impact on HRQoL. However, in this study Mortality was excluded. Using a standard gamble approach, Solomon et al. [ 27 ], found that a severe Stroke was weighted nearly equivalently to death. In contrast, our study did not subdivide the endpoint Stroke to ensure comparability with other defined endpoints . To our knowledge, this is the first AHP study to examine patient preferences on efficacy and safety endpoints in the pharmaceutical treatment of high blood pressure including next to Morbidity Adverse events and Mortality. Though AHP can produce robust and valid results for smaller samples [ 12 ], our study with 26 evaluated participants is rather small, which limits robust statistical inference especially for regression models and reduces reliability of the subgroups carried out. Furthermore, subgroup analysis depending on socio-economic status would allow for a more extensive analysis on potential impact of this criterion on elicited preferences. The structured interviews as the chosen approach for the administration of the AHP questionnaires enabled personal interaction allowing for clarification of questions and high response rates due to personal touch. Some potential bias due to social desirability or potential interviewer bias cannot be avoided [ 28 ]. Another limitation refers to the fact that the study was conducted only in three pharmacies in two German regions and therefore the results may not be generalizable. AHP is primarily method focused on preference elicitation rather than statistical inference. Previous healthcare AHP studies have used comparable sample sizes (10-30 participants) [ 13 – 18 ] to derive stable group-level weights when consistency ratios are within acceptable limits. On an aggregated level consistency ratios showed acceptable magnitudes, whereas on an individual level consistency ratios partially exceeded consistency ratio thresholds. Methodological weaknesses of the AHP method are discussed in international literature. Regarding the rating scale used, it has been criticized for not reflecting every realistically possible assessment and that under certain conditions rank reversal can occur [ 29 ]. Yet, AHP is considered the most frequently implemented multi-criteria method worldwide [ 30 ] and there are recent applications in public-health relevant topics [ 31 – 33 ] and regarding patient-relevant endpoints [ 34 ] as well as different treatment strategies in specific conditions [ 35 ]. Shafir and LeBoef [ 36 ] show that individuals often violate rationality assumptions about logical conclusions, judgments, and choices in their decisions, deviating from the normative model of behavior. They support the criticism of rationality as an axiomatic form of behavior that is now emerging in the social sciences. Other well-established methods like Discrete Choice Experiments (DCEs) often present cognitive challenges for respondents, especially when the respective experiment involves many attributes or when respondents suffer from cognitive impairments. The AHP with its pair-wise comparison and weighting seems to cause less cognitive burden. The cognitive challenges in DCE stem from the cognitive effort required to evaluate parallelly trade-offs between different attributes and make choices, potentially leading to the use of simplifying heuristics and thereby even inaccurate responses [ 37 ]. Nonetheless, the high individual inconsistency shown in the present AHP survey may reflect cognitive burden or survey complexity, even though over half of the participants reported that the survey was ‘rather easy’ to complete. Aggregated patient preferences can be brought in for Health Technology Assessment (HTA) purposes in a positive way. In a recent exploration of the uses of patient preferences in HTA the authors concluded that patient preferences would add most value for informing what and how much matters to patients (e.g., endpoint selection and benefit-risk assessments) [ 38 ]. Furthermore, individual patient preferences can be used in the context of shared decision-making (SDM) as an alternative to traditional informed consent procedures [ 39 ]. SDM has been proposed as a model for clinical practice, where good communication skills of physicians are integrated with the use of patient decision support tools to explore preferences and make decisions [ 40 ]. In our case this translates to an individual approach for deriving patient preferences to decide on different strategies in the treatment of primary hypertension depending on endpoint weighting. AHP can be used on an individual level as a decision supporting tool [ 23 ] by modern digital technologies. AHP can be used to obtain preferences of patients suffering from primary hypertension for efficacy and safety endpoints . Preference elicitation could provide important information and challenge opinions on the importance of respective endpoints , as in the present study the avoidance of Stroke was similarly weighted as or even higher than Mortality . This highlights the importance of individual patient preferences to facilitate tailored therapies for improving benefit. Considering patient preferences fosters patient-centered approaches and can have even an impact on treatment success. Summary What is known about this topic: Stroke is the most important outcome to avoid among patients with multiple chronic conditions but is usually not compared against mortality in patient preference elicitation studies and medication adverse events are not included in these studies. What this study adds: The study examines patient preferences on efficacy and safety endpoints in the pharmaceutical treatment of high blood pressure including next to morbidity adverse events and mortality with stroke as an event to be avoided receiving the highest weight, followed by mortality and myocardial infarction. The results of the patients’ preference elicitation in hypertension can be used to tailor individual treatment within Shared decision-making or for Health Technology Assessment purposes as well. The Analytical Hierarchy Process (AHP) survey is feasible and accepted by patients with primary hypertension regarding the assessment of different endpoints of their treatment. Supplementary information ESM Figure (43.2KB, docx) ESM text 1 (28.6KB, docx) ESM text 2 (20.8KB, docx) Acknowledgements We would like to thank the participants in our AHP pilot study and the owners of the pharmacies in Bavaria and North Rhine-Westfalia where the participants were recruited. Author contributions CMD developed the concept, design and implementation of the study. CMD and NC were involved in data collection. CMD analyzed the data, calculated the AHP and the regression model. CMD and NC wrote the manuscript together. Both authors approved the manuscript and act as guarantors of the article. Funding This study did not receive any external funding. The study was funded by the institute’s own resources. Open Access funding enabled and organized by Projekt DEAL. Data availability Data and supporting materials are available upon request. Competing interests CMD is next to his academic affiliation employed by Bayer Vital GmbH, Leverkusen, Germany. NC has no competing interests. Ethics approval and consent to participate The study was approved by the Institutional Review Board of the Heinrich-Heine-University Düsseldorf. Written informed consent was obtained from all participants. All methods were performed in accordance with the relevant guidelines and regulations. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41371-026-01135-8. References 1. Flint AC, Conell C, Ren X, Banki NM, Chan SL, Rao VA, et al. Effect of systolic and diastolic blood pressure on cardiovascular outcomes. N Engl J Med. 2019;381:243–51. 10.1056/NEJMoa1803180 [ DOI ] [ PubMed ] [ Google Scholar ] 2. Murray CJL, Aravkin AY, Zheng P, Abbafati C, Abbas KM, Abbasi-Kangevari M, et al. Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the global burden of disease study 2019. The Lancet. 2020;396:1223–49. 10.1016/S0140-6736(20)30752-2 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. McEvoy JW, McCarthy CP, Bruno RM, Brouwers S, Canavan MD, Ceconi C, et al. 2024 ESC guidelines for the management of elevated blood pressure and hypertension: developed by the task force on the management of elevated blood pressure and hypertension of the european society of cardiology (ESC) and endorsed by the european society of endocrinology (ESE) and the european stroke organisation (ESO). European Heart Journal. 2024;45:3912–4018. 10.1093/eurheartj/ehae178 [ DOI ] [ PubMed ] [ Google Scholar ] 4. Dolan JG. Multi-criteria clinical decision support: A primer on the use of multiple criteria decision making methods to promote evidence-based, patient-centered healthcare. The patient. 2010;3:229–48. 10.2165/11539470-000000000-00000 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Dolan JG, Bordley DR. Using the analytic hierarchy process (AHP) to develop and disseminate guidelines. QRB - Quality Review Bulletin. 1992;18:440–7. 10.1016/S0097-5990(16)30570-X [ DOI ] [ PubMed ] [ Google Scholar ] 6. Hummel MJ, van Rossum W, Verkerke GJ, Rakhorst G. Assessing medical technologies in development. A new paradigm of medical technology assessment. International journal of technology assessment in health care. 2000;16:1214–9. 10.1017/s0266462300103253 [ DOI ] [ PubMed ] [ Google Scholar ] 7. Schmidt K, Aumann I, Hollander I, Damm K, von der Schulenburg JM. Applying the analytic hierarchy process in healthcare research: a systematic literature review and evaluation of reporting. BMC Med Inform Decis Mak. 2015;15:112. 10.1186/s12911-015-0234-7 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Bywall KS, Kihlbom U, Hansson M, Falahee M, Raza K, Baecklund E, et al. Patient preferences on rheumatoid arthritis second-line treatment: a discrete choice experiment of Swedish patients. Arthritis Research & Therapy. 2020;22:288. 10.1186/s13075-020-02391-w [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Mühlbacher AC, Bridges JFP, Bethge S, Dintsios CM, Schwalm A, Gerber-Grote A, et al. Preferences for antiviral therapy of chronic hepatitis C: a discrete choice experiment. The European Journal of Health Economics. 2017;18:155–65. 10.1007/s10198-016-0763-8 [ DOI ] [ PubMed ] [ Google Scholar ] 10. Fischer AK, Sadler A, Mathey E, Mühlbacher A. Analysis of patients preferences in type 2 diabetes mellitus second-line drug treatment: A discrete choice experiment. PLoS One. 2025;20:e0329743. 10.1371/journal.pone.0329743 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Saaty TL. Group decision making and the AHP. In: Golden BL, Wasil EA, Harker PT, editors. The Analytic Hierarchy Process: Applications and Studies. Berlin, Heidelberg: Springer Berlin Heidelberg; 1989. p. 59–67. [ Google Scholar ] 12. Melillo P, Pecchia L, editors. What is the appropriate sample size to run analytic hierarchy process in a survey-based research. Proceedings of the international symposium on the analytic hierarchy process; 2016. 10.13033/isahp.y2016.130 13. Kitamura Y. Decision-making process of patients with gynecological cancer regarding their cancer treatment choices using the analytic hierarchy process. Jpn J Nurs Sci. 2010;7:148–57. 10.1111/j.1742-7924.2010.00147.x [ DOI ] [ PubMed ] [ Google Scholar ] 14. Hummel MJ, Volz F, van Manen JG, Danner M, Dintsios CM, Ijzerman MJ, et al. Using the analytic hierarchy process to elicit patient preferences: prioritizing multiple outcome measures of antidepressant drug treatment. The patient. 2012;5:225–37. 10.1007/BF03262495 [ DOI ] [ PubMed ] [ Google Scholar ] 15. Basoglu N, Daim TU, Topacan U. Determining patient preferences for remote monitoring. J Med Syst. 2012;36:1389–401. 10.1007/s10916-010-9601-1 [ DOI ] [ PubMed ] [ Google Scholar ] 16. Lambooij MS, Hummel MJ. Differentiating innovation priorities among stakeholder in hospital care. BMC Med Inform Decis Mak. 2013;13:91. 10.1186/1472-6947-13-91 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Kadohira M, Hill G, Yoshizaki R, Ota S, Yoshikawa Y. Stakeholder prioritization of zoonoses in Japan with analytic hierarchy process method. Epidemiol Infect. 2015;143:1477–85. 10.1017/s0950268814002246 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Mühlbacher AC, Juhnke C, Kaczynski A. [Patients’ priorities in the treatment of neuroendocrine tumours: an analytical hierarchy process]. Gesundheitswesen. 2016;78:651–7. 10.1055/s-0035-1548932 [ DOI ] [ PubMed ] [ Google Scholar ] 19. Saaty TL. A scaling method for priorities in hierarchical structures. Journal of mathematical psychology. 1977;15:234–81. 10.1016/0022-2496(77)90033-5 [ Google Scholar ] 20. Dolan JG, Isselhardt BJ, Cappuccio JD. The analytic hierarchy process in medical decision making:a tutorial. Medical Decision Making. 1989;9:40–50. 10.1177/0272989x8900900108 [ DOI ] [ PubMed ] [ Google Scholar ] 21. IQWiG. Different antihypertensive drugs as first line therapy in patients with essential hypertension: Executive summary of final report A05-09, Version 1.0. Cologne, Germany: Institute for Quality and Efficiency in Health Care (IQWiG); 2005. Available from https://www.iqwig.de/download/a05-09_executive_summary_antihypertensive_drugs_as_first-line_therapy.pdf [ PubMed ] [ Google Scholar ] 22. Padilla-Garrido N, Aguado-Correa F, Cortijo-Gallego V, López-Camacho F. Multicriteria decision making in health care using the analytic hierarchy process and Microsoft Excel. Med Decis Making. 2014;34:931–5. 10.1177/0272989x14534533 [ DOI ] [ PubMed ] [ Google Scholar ] 23. Dolan JG. Shared decision-making – transferring research into practice: the analytic hierarchy process (AHP). Patient Education and Counseling. 2008;73:418–25. 10.1016/j.pec.2008.07.032 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Ijzerman MJ, van Til JA, Snoek GJ. Comparison of two multi-criteria decision techniques for eliciting treatment preferences in people with neurological disorders. The patient. 2008;1:265–72. 10.2165/1312067-200801040-00008 [ DOI ] [ PubMed ] [ Google Scholar ] 25. van Til JA, Renzenbrink GJ, Dolan JG, Ijzerman MJ. The use of the analytic hierarchy process to aid decision making in acquired equinovarus deformity. Arch Phys Med Rehabil. 2008;89:457–62. 10.1016/j.apmr.2007.09.030 [ DOI ] [ PubMed ] [ Google Scholar ] 26. Aschmann HE, Puhan MA, Robbins CW, Bayliss EA, Chan WV, Mularski RA, et al. Outcome preferences of older people with multiple chronic conditions and hypertension: a cross-sectional survey using best-worst scaling. Health Qual Life Outcomes. 2019;17:186. 10.1186/s12955-019-1250-6 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Solomon NA, Glick HA, Russo CJ, Lee J, Schulman KA. Patient preferences for stroke outcomes. Stroke. 1994;25:1721–5. 10.1161/01.str.25.9.1721 [ DOI ] [ PubMed ] [ Google Scholar ] 28. West BT, Blom AG. Explaining interviewer effects: a research synthesis. Journal of Survey Statistics and Methodology. 2016;5:175–211. 10.1093/jssam/smw024 [ Google Scholar ] 29. Dyer JS. Remarks on the analytic hierarchy process. Management science. 1990;36:249–58. 10.1287/mnsc.36.3.249 [ Google Scholar ] 30. Wallenius J, Fishburn PC, Zionts S, Dyer JS, Steuer RE, Deb K. Multiple criteria decision making, multiattribute utility theory: recent accomplishments and what lies ahead. Management Science. 2008;54:1336–49. 10.1287/mnsc.1070.0838 [ Google Scholar ] 31. Byun DH, Chang RS, Park M-B, Son H-R, Kim C-B. Prioritizing community-based intervention programs for improving treatment compliance of patients with chronic diseases: applying an analytic hierarchy process. International Journal of Environmental Research and Public Health. 2021;18:455. 10.3390/ijerph18020455 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Gao T, Bernstein P, Zhou H. Interactive experience design evaluation of community emotional healing installations based on analytic hierarchy process method. Soc Sci Med. 2026;388:118731. 10.1016/j.socscimed.2025.118731 [ DOI ] [ PubMed ] [ Google Scholar ] 33. Li Y, Zhang Q, Fang X. Research on patient-centered design for post-stroke depression patients based on SEM and comprehensive evaluation. Front Public Health. 2023;11:1120596. 10.3389/fpubh.2023.1120596 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Mühlbacher AC, Bethge S, Kaczynski A. Treatment after acute coronary syndrome: analysis of patient’s priorities with analytic hierarchy process. International Journal of Technology Assessment in Health Care. 2016;32:284–91. 10.1017/S0266462316000428 [ DOI ] [ PubMed ] [ Google Scholar ] 35. Peters MEC, Icks A, Dintsios C-M. Women’s preferences in early pregnancy loss management: a focus on psychological considerations using the analytic hierarchy process. The Patient - Patient-Centered Outcomes Research 2025. 10.1007/s40271-025-00793-z. Online ahead of print. [ DOI ] [ PMC free article ] [ PubMed ] 36. Shafir E, LeBoeuf RA. Rationality. Annual review of psychology. 2002;53:491–517. 10.1146/annurev.psych.53.100901.135213 [ DOI ] [ PubMed ] [ Google Scholar ] 37. Veldwijk J, Marceta SM, Swait JD, Lipman SA, de Bekker-Grob EW. Taking the shortcut: simplifying heuristics in discrete choice experiments. The patient. 2023;16:301–15. 10.1007/s40271-023-00625-y [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Marsh K, de Bekker-Grob E, Cook N, Collacott H, Danyliv A. How to integrate evidence from patient preference studies into health technology assessment: a critical review and recommendations. International journal of technology assessment in health care. 2021;37:e75. 10.1017/s0266462321000490 [ DOI ] [ PubMed ] [ Google Scholar ] 39. Spatz ES, Krumholz HM, Moulton BW. Prime time for shared decision making. JAMA. 2017;317:1309–10. 10.1001/jama.2017.0616 [ DOI ] [ PubMed ] [ Google Scholar ] 40. Elwyn G, Frosch D, Thomson R, Joseph-Williams N, Lloyd A, Kinnersley P, et al. Shared decision making: a model for clinical practice. Journal of General Internal Medicine. 2012;27:1361–7. 10.1007/s11606-012-2077-6 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials ESM Figure (43.2KB, docx) ESM text 1 (28.6KB, docx) ESM text 2 (20.8KB, docx) Data Availability Statement Data and supporting materials are available upon request. Articles from Journal of Human Hypertension are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (721.5 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top