Abstract
Background Patients scored as ‘high-risk’ and with chronic conditions in primary care (PC) often face medication-related problems. A significant but underexplored problem is the patient’s medication concern, which may also impact treatment decisions and outcomes.
Aim This study aimed to describe the characteristics of patients scored as high-risk with chronic conditions in Madrid, Spain who have documented medication concern and to analyse the associated factors.
Design & setting Descriptive cross-sectional study, conducted within the Innovative Medicines Initiative Horizon 2020 (IMI-H2020) European BEAMER project, including all patients scored as high-risk with chronic conditions in the Madrid region.
Method Patients were identified by the Adjusted Morbidity Groups (AMG) stratification tool in PC electronic clinical records. Variables included sociodemographic, clinical, and pharmacological, along with self-reported medication concerns. Univariate and bivariate analyses and logistic regression models were performed with medication concerns as the dependent variable.
Results In all, 40 776 (28.56%) reported medication concern. Patients with medication concern were older (84.6 versus 72.7 years), had more chronic conditions (7 versus 6), and medications (9.7 versus 8.7). They were more likely to live alone (8.5% versus 2.7%) and showed higher dependency. The following were associated with medication concern: older age odds ratio (OR) 11.0 (95% confidence interval (CI) = 5.4 to 22.3; P<0.001), polypharmacy OR 3.2 (95% CI = 3.1 to 3.3; P<0.001), living alone OR 1.5 (95% CI = 1.4 to 1.6; P<0.001), Parkinson’s OR 1.3 (95% CI = 1.2 to 1.4; P<0.001); chronic pain OR 1.6 (95% CI = 1.2 to 2.1; P = 0.001), and cardiovascular risk factors OR 1.2 (95% CI = 1.1 to 1.3; P = 0.001).
Conclusion Age, polypharmacy, multimorbidity, and social isolation are predictors of medication concern in patients scored as high-risk with chronic conditions in PC. Addressing psychological barriers, simplifying treatment, and strengthening social support through tailored interventions could decrease medication concerns.
How this fits in
Patients with multimorbidity and polypharmacy may face medication concerns. GPs play a key role in managing these patients, providing medication counselling. This study addressed the associated factors to medication concern, highlighting the need for GPs to be aware of these factors when managing patients with chronic conditions. Patients with medication concerns were older, had more chronic conditions, and took more medications. They were more likely to live alone and had higher dependency according to the Barthel index. Parkinson’s disease, chronic pain, and cardiovascular risk factors were associated with medication concern. However, a previous history of cancer and a higher deprivation index were associated with a lower medication concern. PC providers must tailor interventions based on these factors to improve the patient’s beliefs.
Introduction
Improvements in living conditions and therapeutic options have contributed to an increase in life expectancy,1,2 and consequently, to a higher prevalence of multiple chronic diseases — often occurring simultaneously and defined as multimorbidity or multiple long-term conditions — which are associated with a substantial burden of morbidity.3 Among patients with multimorbidity, there is a subgroup of patients at high risk, defined as medically complex individuals with chronic conditions who have increased vulnerability to adverse health outcomes due to disease burden, frailty, disability, complicated treatment regimens, and greater healthcare utilisation.4–6 These patients are often identified using stratification tools such as the Adjusted Morbidity Groups (AMG) tool, which is commonly used in primary health care (PHC) in Spain to help GPs and nurses identify them, allowing for more tailored and proactive care.7
Patients with multimorbidity usually experience polypharmacy, the concurrent use of multiple medications to manage such conditions.8 Although many patients adapt to complex medication schedules,9 polypharmacy is frequently associated with unintended drug-related problems such as adverse drug reactions, drug interactions, and non-adherence,10,11 which often stem from factors related to a lack of communication by the healthcare professionals and the act of prescribing itself.12 Patients’ health literacy, lack of understanding, forgetfulness, and financial constraints have also been associated with drug-related problems.2,8,10 Although some of them may be preventable, these medication-related problems may harm patients and they have been linked to negative outcomes, including falls, frailty, cognitive decline, medication overdose, poor disease management, reduced quality of life, increased healthcare costs, hospitalisations, and mortality.2,13
Therefore, the beliefs influencing patients’ commonsense evaluations of prescribed medicines depend on the perceptions of personal need for treatment (necessity beliefs) and concerns about a range of potential adverse consequences,14 as mentioned before. Many patients are concerned about potential adverse effects of medications and this can lead to intentional non-adherence,15 reduced treatment effectiveness, poor health outcomes, and increased expenses.16,17 Negative expectations about medications can even trigger the nocebo effect, reducing the effectiveness of treatment.13
The way in which medication risks are communicated also plays a crucial role, for example offering information numerically (for example percentages) improves understanding.17,18 Furthermore, a lack of trust in healthcare providers or in the healthcare system can exacerbate medication concerns, as patients may doubt the safety or the necessity of prescribed treatments. This distrust can lead to reluctance in starting or continuing treatment, further contributing to non-adherence and worsening health outcomes.19,20
Some studies have examined medication concerns associated with specific types of medications, such as opioids for patients suffering from pain, in which patients associate use of the opioid with end-of-life care,21 or benzodiazepines for patients with mental health problems, in which case patients commonly report fears of dependency and sedation. A systematic review of qualitative studies reflected how medication concern rose consistently in different study settings.22
Even though previous mostly qualitative studies have addressed patients’ beliefs around medication concerns, there is limited evidence on the associated factors to medication concerns in the PC setting.
This study aimed to describe the characteristics of patients with multimorbidity in the Madrid region of Spain who were identified to have documented concern about medications in their electronic PC medical records and to analyse the associated factors. It was conducted under the BEAMER project, which is developing a disease-agnostic model that segments the population on the basis of actionable factors and predicts adherence behaviour as a means of identifying patients’ support needs; this will make it possible to provide personalised and targeted patient support that promotes better outcomes.
Method
Study design
Descriptive cross-sectional study within the IMI-H2020 European BEAMER project (BEhavioral and Adherence Model for improving quality, health outcomes and cost-Effectiveness of healthcaRe).23 The project seeks to develop an AI model to aid in the prediction of treatment adherence.
Population
The study included the entire population of patients aged ≥18 years who attended in PHC labelled as “high risk” according to the AMG stratification tool in the electronic health records,7 on April 30th, 2021 (approximately 5% of the Madrid population aged >18 years). Individuals receiving palliative care and those with an active diagnosis of dementia, autism, or any intellectual disability were excluded from this post-hoc analysis, as their preferences and willingness might be anomalous or might be reported by caregivers.
Variables
The primary outcome was self-reported medication concern, operationalised using the the ICPC-2 code “A13” (concern/fear of medical treatment) and defined as patients’ self-reported apprehension about their prescribed medications, including worries about side effects, potential drug interactions, adherence challenges, treatment complexity, and perceived impacts on overall health.
Explanatory variables
Sociodemographic: age (years), sex (male/female), deprivation index,24 and number of people in the household (living alone/not living alone).
Sociofunctional: the Barthel Functional Evaluation,25 which is performed by the nurses whenever patients reach the age of 60 years as part of their routine evaluations.26
Use of healthcare resources: number of appointments with physicians and nurses.
Clinical variables: total number of chronic diseases; cardiovascular risk factors (diabetes mellitus, hypertension, dyslipidaemia, obesity [defined as body mass index >30 kg/m2]); cardiovascular diseases (cerebrovascular disease, coronary heart disease, heart failure, valve diseases, arrhythmias); renal disorders (chronic kidney disease [CKD]); mental health disorders (anxiety, depression, severe mental health diseases [bipolar disorder, psychotic disorders]); musculoskeletal diseases (arthrosis, spinal hernia, osteoporosis); digestive disorders (cirrhosis, other hepatic diseases, inflammatory bowel disease); respiratory diseases (asthma, chronic obstructive pulmonary disease); ophthalmological problems (glaucoma); oncological diseases (including Hodgkin lymphoma, other lymphomas, leukaemia, melanoma, multiple myeloma, colon cancer, cervical cancer, stomach cancer, breast cancer, prostate cancer, pancreatic cancer, lung cancer, thyroid cancer, bladder cancer, central nervous system [CNS] cancer, and kidney cancer); neurological diseases (Parkinson’s disease and epilepsy); and thyroid diseases.
Clinical variables were obtained from the ICPC-2 episodes of each patient’s electronic health record.
Pharmacological: total number of medications; polypharmacy (defined as ≥5 concurrent prescriptions); adverse effects due to drugs taken at the correct dose; previous treatment complications.
Type of medication: antithrombotic (acenocoumarol, new oral anticoagulants, acetylsalicylic acid, other antiaggregants [ticagrelor, ticlopidine, and clopidogrel]), opioids (tramadol, morphine, fentanyl, and tapentadol), CNS medications (serotonin reuptake inhibitors [SRIs], benzodiazepines, and injected medications: intramuscular cyanocobalamin and subcutaneous heparin.
Adherence to medication: measured on the basis of patients who have ≥80% days covered for the minimum recommended daily dosage during a one-year observation period according to dispensation records in electronic clinical records.
Data collection
Madrid’s healthcare system operates within the framework of the Spanish National Health System and provides universal health coverage to all residents. Data were collected from secondary sources registered in the Madrid PC electronic medical records database and in the medicine dispensation records within the Madrid region.23
Analysis
Descriptive analysis was performed to determine the baseline characteristics of the sample. Categorical variables were expressed as numbers and percentages. Continuous variables were expressed as means and standard deviations (SDs) if they followed a normal distribution, and as medians and interquartile ranges (IQRs) if their distribution was skewed or non-normal. The categorical variables were grouped into new multivariables as described in Supplement 1.
Bivariate analysis between patients who had no medication concerns and those who did was performed, using Pearson’s chi-square test for categorical variables, a t-test for continuous normally distributed variables, and a Mann-Whitney U test for continuous nonnormally distributed variables.
A logistic regression model was built to analyse the factors associated with medication concerns. The model included all sociodemographic variables, all variables that showed differences in the bivariate analysis, and all the variables that could act as possible confounders (sex, age, deprivation index, number of appointments with the nurse, family physician, previous adverse events, and previous treatment complications). The model included living alone, number of diseases, cardiovascular events, cardiovascular risk factors, arrhythmias, respiratory disease, anxiety, depression, severe mental health disease, cirrhosis, other hepatic disease, inflammatory bowel disease, spine hernias, chronic pain, arthrosis, osteoporosis, glaucoma, CKD, oncological background, Parkinson’s disease, thyroid disorders, number of medications, previous treatment complications, previous adverse events, polypharmacy, acenocoumarol, other anticoagulants, acetylsalicylic acid, clopidogrel, tramadol, codeine, morphine, fentanyl, tapentadol, antidepressants, benzodiazepines, intramuscular cyanocobalamin, statins, and subcutaneous heparin. We performed a background regression excluding the variables with the highest P values and looked for the best fit model by conducting the likelihood ratio test and Pseudo R2. The final model included the following variables: sex, age, deprivation index, previous adverse events, polypharmacy, living alone, number of diseases, cardiovascular events, cardiovascular risk factors, arrhythmias, chronic pain, arthrosis, osteoporosis, glaucoma, CKD, oncological background, and Parkinson’s disease.
The analyses were conducted using STATA version 18 in the Python programming environment (version 3.9.7) and used the following libraries: NumPy (v1.19.5), Pandas (v1.4.0), and Scikit-learn (v1.0.2).
Results
Among the 163 188 patients scored as high-risk with chronic conditions stratified by the AMG tool, 142 795 met the inclusion criteria (Figure 1). In total, 40 776 (28.6%) patients self-reported medication concerns.
The figure is a flow diagram showing the selection and classification of patients scored as ‘high-risk’. The starting group contains 163 188 high-risk patients with multimorbidity. From this group, patients are excluded if they are in palliative care, have dementia, have autism, or have an intellectual disability. The excluded groups are: 2706 patients in palliative care, 17 651 patients diagnosed with dementia, 9 patients diagnosed with autism, and 27 patients with intellectual disability.After these exclusions, 142 795 high-risk patients remain. This group is then divided into two categories according to whether ICPC-2 code A13, “Concern/fear of medical treatment”, was recorded. The larger group contains 102 019 patients with no recorded ICPC-2 code A13. The smaller group contains 40 776 patients with recorded ICPC-2 code A13.
Overall, the mean age of the included patients was 76.1 years (SD: 11.8) and 75 715 (53.0%) were men. The overall deprivation index was 0.1 (SD: 0.9). Table 1 shows the sociodemographic and clinical characteristics of the sample, comparing patients who were taking medication with those who were not. Patients with medication concerns were older (84.6 [SD: 6.8] versus 72.7 [SD: 11.6] years) and belonged to the lowest quartile of the deprivation index (78.5% within the 1st–3rd quartiles versus 21.5%). More men had documented medication concerns than women (24 569 [60.3%] versus 16 207 [39.7%]). Patients with fear lived alone more frequently (3462 [8.5%] versus 2756 [2.7%]) and were dependent according to the Barthel scale (11 385 [27.9%] versus 13 283 [13.0%]); they also had a greater number of appointments with the doctor and the nurse (22 [IQR: 13, 34] versus 19 [IQR: 11, 30]).
With respect to the clinical variables, patients with medication concerns had a median of 7 [IQR: 6, 8] diseases compared with those without, 6 [IQR: 5, 8], and a greater number of medications, 9.7 (SD: 3.5) versus 8.7 (SD: 3.9); they were also more adherent to treatment (15 247 [37.4%] versus 6141 [35.4%] [Table 2]). Among patients with medication concerns, 4119 (10.1%) had previously experienced adverse effects of drugs taken at the correct dose.
The results from the logistic regression model included the variables shown in Table 3. Being in the oldest group (81–108 years of age) was associated with the highest risk of medication concern, with an OR of 10.40 (95% CI = 6.5 to 16.7; P<0.001). The following were associated with medication concerns: polypharmacy, OR 3.13 (95% CI = 3.02 to 3.25; P<0.001); living alone, OR 1.46 (95% CI = 1.37 to 1.54; P<0.001); having Parkinson’s disease, OR 1.31 (95% CI = 1.22 to 1.41; P<0.001); having chronic pain, OR 1.58 (95% CI = 1.22 to 2.05; P = 0.001); and being diagnosed with 1 or more cardiovascular risk factors, OR 1.18 (95% CI = 1.09 to 1.28; P = 0.001). On the other hand, the association between being in the 4th quartile of the deprivation index and having medication concern had an OR of 0.66 (95% CI = 0.63 to 0.68; P<0.001) and the association of a previous history of cancer with having medication concern had an OR of 0.83 (95% CI = 0.80 to 0.87; P<0.001).
Discussion
Summary
This post hoc analysis described a subsample of all patients with multimorbidity scored as high-risk by the AMG in the Madrid region (Spain) within the BEAMER project. Our results revealed that self-reported medication concerns were associated with the oldest age groups (>66 years), polypharmacy, living alone, and living with cronic diseases such as cardiovascular risk factors, cardiovascular risk events, arrhythmia, chronic pain, arthrosis, osteoporosis, glaucoma, CKD, and Parkinson’s disease; however, a higher deprivation index and previous history of oncological disease were not strongly associated.
Strengths and limitations
This large cross-sectional study utilised data from the entire population of patients scored as high-risk with chronic conditions in Madrid, Spain. While cross-sectional studies have inherent limitations that prevent causal inferences, the large number of patients included and the high representativeness (as 82.57% of Madrid’s residents are public healthcare users) increased the reliability of our findings and decreased selection bias.27
Nevertheless, the use of secondary data obtained from electronic health records, while valuable for large-scale analyses, may introduce variability in the classification of variables owing to differences among healthcare professionals in documentation practices and prescribing patterns and might potentially lead to information bias. In the Madrid region, GPs may record diseases or symptoms in the electronic health record using ICPC-2 codes and include additional information as free text within the episode notes.
Previous research has highlighted the complexity of coding practices among GPs. For instance, a study in Wales observed how GPs sometimes differed in their preferences on what diseases and prescribing patterns to code, and when to code them.28 A previous systematic review also highlighted barriers of coding practices, which included difficulties when matching patients’ symptoms and codes.29 Coding practices were particularly relevant to the main outcome of this study, medication concerns, which are reported by patients to their GPs or nurses. Some professionals may choose to document patient concerns using free text, a modality that cannot yet be systematically analysed, whereas others may not record these concerns at all.
Comparison with existing literature
Our study revealed that older patients had more medication concerns than did younger patients. This might be related to the significant increase in the number of medications prescribed with increasing age and to the complexity of the treatments received, as shown in other studies.5,6,30 Previous studies suggest people in the oldest age category may find information provided by packaging leaflets and the different routes of application, timing of administration, and dietary restrictions associated with medication use overwhelming and complex, which could raise medication concerns.22,30,31
The results suggested that living alone, regardless of feelings of loneliness, was linked to a greater likelihood of medication concern. Another study conducted in South Korea found high rates of patients with medication concerns among patients living alone.32
Older people and people who are dealing with complex comorbidities often benefit from having strong social networks and having such networks might help alleviate the fears and concerns of people living with multimorbidity.31
Interestingly, our study revealed that non-adherent patients appeared to have a lower likelihood of medication concern. Patients with better adherence may be more prone to express their concerns to physicians and nurses, whereas non-adherent patients might express their concerns to healthcare professionals to a lesser extent.
Implications for research and practice
The world’s population is ageing and healthcare professionals must adapt to this situation and direct their efforts towards improving patient care, promoting shared decision-making, and in-depth understanding of the processes that lead patients to opt for one treatment or another.16 Future research should explore the complex relationship between medication concerns and factors such as polypharmacy, multimorbidity, and social isolation, particularly in older populations.
Strengthening social support networks, particularly for patients with complex comorbidities or those living alone, could further alleviate medication-related concerns and improve treatment outcomes in PC. These efforts will contribute to more effective, patient-centred care and better overall health outcomes.
Additionally, investigating the role of patient beliefs and attitudes towards medication, as well as the impact of healthcare providers' communication strategies, could improve medication concerns in this patient group. Given the central role of PC in managing chronic conditions, GPs and other PC professionals, such as nurses and pharmacists, are uniquely positioned to identify patients who are at risk, to foster trust, and to implement strategies that alleviate concerns about medication.
Our findings from this sub-study may provide PHC physicians with tools to find patients who may be prone to have medication concerns and better address their needs with medication.
Amongpatients scored as high-risk with multimorbidity, older patients, patients living alone, and patients with complex comorbidities are particularly vulnerable to medication concerns. Interestingly, patients considered non-adherent to treatment were less likely to express medication concerns, possibly indicating lower engagement or awareness of potential medication risks. This underscores the complexity of medication adherence and the need to address both psychological and practical factors when developing interventions.
Notes
Funding
This project received funding from the Innovative Medicines Initiative (IMI) 2 Joint Undertaking under grant agreement No. 101034369. This joint undertaking receives support from the European Union’s Horizon 2020 Research and Innovation Programme, the European Federation of Pharmaceutical Industries and Associations (EFPIA), and Link2Trials. This communication reflects the views of the authors, and neither the IMI nor the European Union, EFPIA, or Link2Trials are liable for any use that may be made of the information contained herein. This project received funding for English editing and publication from the Foundation for Biosanitary Research and Innovation in Primary Care.
Ethical approval
This study received ethics approval from the Hospital Universitario La Princesa Drug Research Ethics Committee and a favourable report from the Local Research Commission of the Primary Care Management of the Community of Madrid. Every method was performed in accordance with relevant guidelines and regulations. The requirement of informed consent from patients was waived, as established by the current legislation, because this study is retrospective and does not include individual personal information since the data were obtained from a secondary database with anonymised and dissociated information.
Provenance
Freely submitted; externally peer reviewed
Data
The data relied on in this article are stored in the Madrid Healthcare systems. The data are available under request.
Acknowledgements
The authors would like to express their sincere gratitude to their colleagues in the BEAMER consortium, FIIBAP, UPM, PredictBy, and the Research Unit of the Primary Care Management Directorate; their support throughout this work has been invaluable. Although they did not meet the criteria for coauthorship, their contributions and collaboration are deeply appreciated.
Competing interests
The authors declare that no competing interests exist.
- Received November 4, 2025.
- Accepted November 11, 2025.
- Copyright © 2026, The Authors
This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)
References
LinkedIn
Jump to section
More in this TOC Section
Related Articles
Cited By...
