This study was conducted as part of the Capital’s Funds for Health Improvement and Research (Grant No. CFH2022-3-7141). We employed a cross-sectional design in the Huairou District of Beijing, with data collected from August 1, 2022 to December 31, 2024. The sample size was determined through a power analysis (90% power), with reference to similar community-based stroke studies regarding key outcome proportions (14), and convenience sampling was used to recruit participants. A total of 481 community-dwelling stroke patients completed the survey.

All participants provided informed consent before the interviews. Each face-to-face interview was carried out by qualified clinical pharmacists, lasted approximately 20 min, and involved an electronic questionnaire. To ensure consistency and minimize bias in the assessment of DRPs, all interviews and subsequent evaluations were conducted by clinical pharmacists who had obtained certification and possessed over 3 years of work experience. The evaluation process followed standardized criteria based on drug labels, clinical guidelines, and expert consensus. All DRP classifications were independently reviewed by a senior pharmacist. Any discrepancies were resolved through discussion by a dedicated quality control panel. The study protocol was reviewed and approved by the Ethics Committee of Beijing Huairou Hospital (Approval No. (2022)-(016)-02).

Inclusion criteria: (1) age≥18 years; (2) had experienced a stroke within the past 6 months and were not currently hospitalized; (3) used three or more prescribed drugs daily; (4) willing and able to participate in face-to-face interviews conducted by clinical pharmacists.

Procedure of visits and collaboration: The initial pharmacist visit was conducted face-to-face at various settings including hospitals, community clinics, or patients’ homes. Follow-up contacts were carried out via face-to-face meetings, phone calls, or WeChat as needed. For recruitment, some patients were referred to the study by their neurologists or family physicians after an assessment confirming they met the inclusion criteria. During the subsequent intervention phase, pharmacists communicated directly with the patients’ neurologists and/or family physicians to discuss and collaboratively adjust medication regimens.

Exclusion criteria: (1) patients with severe or terminal illnesses; (2) those with mental disorders or severe cognitive impairment; (3) inability to complete questionnaires; (4) incomplete clinical data precluding statistical analysis.

The questionnaire was developed by a multidisciplinary expert group including pharmacists, clinicians, sociologists, and health management specialists. Its development process involved literature review, expert consultation, and multiple rounds of discussion. The final instrument consisted of two major sections: (1) Influencing Factors Questionnaire. Collected demographic information, health status, comorbidities, and drug-use history. (2) DRPs Assessment. DRPs were evaluated using the Pharmaceutical Care Network Europe (PCNE) Classification System (Version 9.1). The PCNE structure includes Problem (P), Cause (C), Planned intervention (I), Intervention acceptance (A) and Outcome (O). For the purposes of this study, only the first two components, problems and causes were analyzed. Pharmacists reviewed each case sequentially within the PCNE framework and selected corresponding classifications to achieve both qualitative and quantitative evaluation of DRPs.

All analyses were performed using SPSS 26.0. Descriptive statistics were used to summarize participant characteristics. Chi-square tests were conducted to examine associations between the presence of DRPs and individual variables. Multiple logistic regression was then used to identify independent factors associated with DRPs. DRPs (yes/no) served as the dependent variable. Independent variables included marital status, education level, monthly income, monthly drug expenditure, medical reimbursement type, number of chronic diseases, comorbidities, lifestyle behaviors, health status indicators, disease-related factors, and history of adverse drug reactions. Gender and age were included as covariates. Variables for the multivariate logistic regression model were selected based on clinical relevance and a univariate screening threshold of p < 0.05. A sensitivity analysis using a threshold of p < 0.10 confirmed no additional variables warranted inclusion. The final model was built using the Enter method to include all preselected variables simultaneously.

A total of 481 stroke patients were included in the analysis. Their basic demographic and clinical characteristics are summarized in Table 1. Among them, 295 (61.3%) were male and 186 (38.7%) were female. Most participants were aged 60 years or older (68.0%) and had a primary or junior high school education (68.6%). The majority reported a monthly income below 5,000 CNY.

Regarding lifestyle behaviors, about one-third reported current smoking or alcohol consumption, and most participants engaged in some form of regular physical activity. Comorbidities were common, particularly hypertension (70.9%), hyperlipidemia (55.7%), and diabetes (34.3%). Nearly 30% reported high intake of animal fat.

Polypharmacy was prevalent, more than 70% used five or more drugs daily. Almost half of the participants exhibited poor drug adherence, and around 18% reported symptoms of depression or anxiety. Continuous variables showed that the average BMI fell within the overweight range, and self-assessed health scores were moderate.

Two hundred fifty-nine stroke patients were detected having DRPs. Chi-square analysis indicated that several factors including monthly income, alcohol use, comorbid diabetes, high animal fat intake, polypharmacy, depressive/anxiety symptoms, and lower self-assessed health scores were significantly associated with the presence of DRPs.

Table 2 presents the multivariate logistic regression results. Several variables remained independently associated with DRPs after adjustment. Alcohol consumption significantly increased the likelihood of experiencing DRPs (OR = 1.770, p = 0.01). Comorbid diabetes was also an important determinant (OR = 1.818, p = 0.005). Higher animal fat intake (OR = 1.711, p = 0.016) significantly increased DRP risk. Depressive or anxiety symptoms were associated with the risk of DRPs (OR = 1.773, p = 0.037). Overall, the regression model suggests that behavioral, psychological, eating habits, and disease-related factors all contribute to DRP risk.

A total of 482 DRPs were identified in 259 patients according to the PCNE classification system (Table 3). The majority of DRPs were categorized under “other problems” (P3), primarily comprising unclear complaints or issues requiring further clarification (P3.2, n = 219). Problems related to treatment effectiveness were also frequent (P1; n = 197), with suboptimal treatment effect (P1.2; n = 188) being the most common subtype. A total of 52 treatment safety problems (P2.1) involved actual or potential adverse drug events.

Regarding causes, 482 DRPs can be attributed to 575 causes. Patient-related factors (C7) were the dominant category (n = 330). The most common issues included intentional underuse or non-use of drugs (C7.1; n = 198) and inappropriate timing or dosing intervals (C7.7; n = 88). Other notable causes included dosage-related problems (C3; n = 32), prescribing or drug selection issues (C1; n = 33), administration errors by healthcare professionals (C6; n = 23), and inadequate outcome monitoring (C9; n = 143).

Together, these findings indicate that DRPs among community-dwelling stroke patients primarily arise from suboptimal treatment effectiveness and patient-driven behaviors, particularly non-adherence and improper drug use.

Figure 1 shows the distribution of 482 DRPs across drug categories. The top four drugs most frequently associated with DRPs were antiplatelet agents, antihypertensives, antidiabetic drugs, and lipid-lowering drugs. The four drug categories with the highest number of DRPs are exhibited in Figure 2. For antiplatelet and lipid-lowering agents, the most common DRP type was “unclear problem/complaint,” often reflecting concerns requiring further clarification. For antihypertensives and antidiabetic agents, “suboptimal treatment effect” was the dominant issue.

Figure 3 shows the distribution of PCNE-classified causes for DRPs in the top four drug categories. Across all four categories, the leading cause was intentional underuse or discontinuation (C7.1), followed by “other causes” (C9.2) and inappropriate timing or dosing intervals (C7.7). A deeper examination suggested that antihypertensive and antidiabetic problems often emerged when blood pressure or blood glucose levels remained uncontrolled. In contrast, DRPs related to antiplatelet and lipid-lowering drugs frequently stemmed from patients discontinuing therapy due to concerns about potential adverse effects, thereby increasing their risk of recurrent ischemic events.

Table 4 shows the distribution of active ingredient of four drug categories with the highest number of DRPs. In the analysis of the four types of drugs with a relatively high incidence of DRPS, antiplatelet drugs are mainly aspirin, antihypertensive drugs are mainly CCBs and ARBs, hypoglycemic drugs are mainly biguanides, alpha-glucosidase inhibitors and insulin, and lipid-regulating drugs are mainly statins. The causes of occurrence are consistent with the overall distribution.

Based on the major determinants identified in this study, such as alcohol consumption, comorbid diabetes, higher animal fat intake, depressive or anxiety symptoms, and polypharmacy, as well as the distribution of PCNE-classified DRPs, several focused and feasible intervention strategies are suggested for community settings.

Overall, these targeted strategies support the development of a pharmacist-centered, closed-loop drug management system aimed at reducing DRPs and improving therapeutic outcomes among community-dwelling stroke patients.

Our study still has some limitations. First, it was conducted in a single suburban district, which may limit generalizability. Second, data on adherence and lifestyle behaviors were self-reported, possibly introducing bias. Thirdly, we employed a convenience sampling method, with patients primarily sourced from hospital outpatient departments, referrals by neurologists, and recommendations from family physicians. As a result, data from bedridden patients or those unable to attend hospital visits were not comprehensively collected, which constitutes a limitation of this study. Future studies could employ longitudinal designs, objective adherence measures, and multi-site recruitment to validate and extend these findings.