Latent profile analysis and influencing factors of engagement in medication safety among older adults with cardiometabolic multimorbidity: a cross-sectional study

Abstract

Objective

This study aims to explore the latent profiles and influencing factors of engagement in medication safety among elderly patients with cardiometabolic multimorbidity.

Design

A cross-sectional study.

Setting

The study was conducted at a class III hospital in Jiangsu, China.

Participants

The study included a sample of 316 older adult inpatients with cardiometabolic multimorbidity.

Outcome measures

Participants completed the Inpatients’ Involvement in Medication Safety Scale and the Multimorbidity Treatment Burden Questionnaire. Latent profile and multivariate regression analyses were used to identify subgroups and their associated factors.

Results

Latent profile analysis identified three distinct profiles: ‘passive participation’ (22.47%), ‘moderate participation’ (52.53%) and ‘active participation’ (25.0%). Multivariate logistic regression revealed that occupational status, marital status, medical payment method, daily medication type and treatment burden were significant independent factors distinguishing among these profiles (p<0.05).

Conclusions

The study confirms significant heterogeneity in medication safety engagement among older adults with cardiometabolic multimorbidity. The identified profiles and their specific influencing factors provide a basis for clinicians to stratify patients and develop targeted interventions, particularly for the vulnerable ‘passive participation’ group, to improve medication safety outcomes.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• A key methodological strength was the use of latent profile analysis to identify distinct subgroups of patient involvement in medication safety.

• The findings may lack generalisability as the sample was drawn from a single source.

• The cross-sectional design limits insights into how patient participation changes over time or across care settings.

Background

With the rapid ageing of China’s population, health challenges among older adults have become increasingly prominent. Recent reports reveal that approximately 69.1% of the older adult population in China is affected by at least one chronic disease, while 43.6% are living with two or more chronic conditions.¹ Multimorbidity is defined as the coexistence of two or more chronic diseases within an individual over an extended period.² Among the various forms of multimorbidity, cardiometabolic multimorbidity (CMM) represents one of the most prevalent, stable and recurrent comorbidity clusters in the older adult population.^{3 4} CMM is defined as the coexistence of two or more cardiovascular diseases or metabolic disorders in an individual, including hypertension, diabetes, dyslipidaemia and stroke.³ Given the chronic nature and complex aetiology of these conditions, the mortality risk associated with CMM is significantly increased.^{5 6} Research indicates that the mortality risk for patients newly diagnosed with CMM is nearly three times higher compared with those without any cardiovascular or metabolic diseases,³ posing a serious threat to public health.^{7 8}

Drug therapy is crucial for the secondary prevention and rehabilitation of patients with CMM.⁹ Adhering to long-term, standardised medication regimens as prescribed by healthcare professionals is essential for effective disease management. However, the current medication safety landscape among patients, particularly older adults, remains a significant concern due to numerous hidden risks. Age-related declines in physical, memory and cognitive functions¹⁰ often lead to challenges such as limited awareness of safe medication practices and poor adherence to prescribed treatments.¹¹ These factors substantially increase the risk of medication-related problems. Studies indicate that the rate of medication errors among older adults with chronic diseases is alarmingly high, reaching up to 75%,¹² which severely compromises patient prognosis and overall quality of life. Addressing these issues is essential to enhancing health outcomes and ensuring the well-being of this vulnerable population. Importantly, patients themselves are central to ensuring drug safety and play a vital role in promoting safe medication use.^{13 14} The theme of the fifth World Patient Safety Day,¹⁵ observed on 17 September 2023, is ‘Engaging Patients for Patient Safety,’ which aims to ‘elevate the voice of patients.’ This initiative underscores that ensuring medication safety is a shared responsibility, requiring not only the diligence of healthcare professionals and medical institutions but also active participation from patients themselves.

The concept of patient engagement for medication safety refers to the active involvement of patients with autonomous decision-making capacity throughout the entire process of receiving regular medication, aiming to enhance the safety, rationality and effectiveness of medication utilisation. Involvement in medication safety management can improve the continuity of self-management during the care process, leading to better treatment outcomes and a reduction in medication errors and near misses.^{11 16} In recent years, the promotion of the ‘active health’ concept has encouraged more patients to take the initiative in participating in medication safety management, fostering a positive culture of patient safety. However, previous studies have shown that the extent of patient participation in medication safety is influenced by factors such as gender, educational level, residence and marital status. In addition, compared with other modes of comorbidities, CMM is significantly associated with increased disability and cognitive decline,^{8 17 18} which can lead to higher hospitalisation costs and more frequent hospitalisation experiences, placing a greater treatment burden on patients. Treatment burden refers to the total amount of healthcare-related tasks that patients must undertake and the consequent impact on their overall functioning and health.^{19 20} It includes economic factors, medication management, psychosocial challenges, the medical treatment process and so on. Such burdens can also significantly limit patients’ enthusiasm and their ability to participate in medication safety initiatives. Consequently, the level of participation in medication safety among elderly patients with CMM may vary based on individual characteristics and the extent of their treatment burden. Previous studies have primarily assessed patients’ medication safety engagement using total scores, overlooking individual differences. This approach has led to the failure to develop targeted interventions, resulting in suboptimal outcomes and inefficient use of medical resources.

Latent profile analysis (LPA) is a statistical method that identifies distinct subgroups within a population by categorising individuals with similar response patterns based on their scores on observed variables. This approach allows heterogeneous groups to be classified into homogeneous profiles. Therefore, this study will employ LPA to explore potential subgroups of older adults with CMM based on their engagement in medication safety behaviours. Additionally, the study will analyse relevant influencing factors, assess the impact of treatment burden on these behaviours, and ultimately provide a foundation for the development of tailored intervention programmes to improve medication safety in this population.

Methods

Study design and sample

This study employed a cross-sectional design and included a random sample of 316 older adult inpatients with CMM from a hospital in Nanjing City, Jiangsu Province, China. Patient recruitment took place from November 2023 to June 2024. The inclusion criteria for participants were as follows: (a) aged 60 years or older; (b) diagnosed with two or more cardiovascular metabolic diseases simultaneously, including hypertension, dyslipidaemia, diabetes, heart disease and stroke, where heart disease encompasses myocardial infarction, coronary heart disease, angina pectoris and heart failure²¹; (c) took chronic disease-related drugs for 3 months or more; (d) being conscious, without obvious cognitive impairment, and capable of understanding and speaking Chinese and (e) signed informed consent and agreed to participate in the study. The exclusion criteria included: (a) patients with severe dysfunction of the heart, liver, kidneys or other organs, as well as those with malignant tumours; (b) patients currently involved in other research studies. The sample size was calculated using G-power V.3.1, assuming a power of 0.80. The effect size was set at 0.25, and a two-sided test was conducted at the 0.05 significance level. Therefore, the estimated sample size for this study was at least 159, indicating that our sample of 316 participants was adequate to detect significant effects.

Procedures

Members of the research team collected information about the participants through face-to-face interviews. Before the investigation, the trained members of the research team explained the purpose and significance of the study, as well as the precautions for completing the questionnaire, to the participants. After obtaining informed consent from the patients, the questionnaires were administered on the spot and completed by the patients themselves. The questionnaire included a sociodemographic questionnaire, the Involvement in Medication Safety Scale and the Multimorbidity Treatment Burden Questionnaire for Elderly Patients. The total completion time was 15–20 min. For patients who had difficulty answering the questions, members of the research team assisted them by reading the questions aloud and completing a questionnaire based on their answers. Questionnaires were collected on the spot and checked for completeness.

Measure

Sociodemographic questionnaire

The questionnaire was jointly designed and completed by the members of the research group after literature review^{9 11} and discussion. Including gender, age, education level, working status, marital status, monthly per capita household income, residence, medical payment methods, number of concurrent chronic diseases, type of daily medication, history of adverse drug reactions, primary caregiver and the importance of conscious participation in medication safety.

Inpatients’ Involvement in Medication Safety Scale

The Inpatients’ Involvement in Medication Safety Scale was developed and validated by Wang Binghan et al in China.²² This 23-item instrument uses a 5-point Likert scale (scores range from 23 to 115, with higher scores indicating greater involvement) and comprises three dimensions: participation in decision-making (8 items), caring participation (10 items), and demands participation (5 items). The scale has demonstrated robust psychometric properties in its validation study: excellent internal consistency (Cronbach’s α=0.916 for the total scale; 0.777–0.858 for subscales), good test–retest reliability (r=0.742) and good content validity (content validity index=0.923). Exploratory factor analysis confirmed its three-factor structure, which accounted for 51.19% of the total variance. The scale was used with the written permission of the original authors.

Multimorbidity Treatment Burden Questionnaire for Elderly Patients

The Multimorbidity Treatment Burden Questionnaire for Elderly Patients is a validated instrument developed by Bai Dingxi et al in China to assess the perceived treatment burden among elderly patients with multimorbidity.²³ It comprises 33 items across seven dimensions: economic burden (4 items), self-management burden (6 items), access to healthcare burden (9 items), drug management burden (3 items), adverse drug reactions burden (3 items), social burden (3 items) and psychological burden (5 items). Responses are recorded on a 5-point Likert scale (scores 0–4), with anchors ranging from ‘no difficulty’ to ‘extremely difficult’ and ‘strongly agree’ to ‘strongly disagree’. The total score ranges from 0 to 132, with higher scores indicating a greater treatment burden. Comprehensive validation data from the original study support its robustness: The scale demonstrated excellent internal consistency (Cronbach’s α=0.895), high split-half reliability (0.938) and strong test–retest reliability (r=0.939, p<0.01). Content validity was satisfactory, with item-level Content Validity Indices (I-CVI) ranging from 0.833 to 1.000 and a scale-level average CVI (S-CVI/Ave) of 0.939. Convergent validity was supported by strong correlations between individual items and their respective dimension scores (r=0.522–0.897, all p<0.01). This scale was used with the written permission of the original authors.

Statistical analyses

Data processing and statistical analyses were completed jointly using SPSS V.25.0 and Mplus V.8.3 software after double data entry. The three dimensions of continuous-variable patients’ participation in medication safety behaviours were used as exogenous variables to construct a latent profile model using Mplus 8.3. The fit indices included Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), adjusted BIC (aBIC) and Entropy. Smaller values of the AIC, BIC and aBIC statistics indicate a better model fit. Moreover, classification accuracy is measured by entropy, which ranges from 0 to 1, with higher values indicating greater accuracy. Additionally, it is recommended that each profile be at least 5% of the total sample. The data were further processed using SPSS V.25.0 software. Qualitative data were expressed as frequencies and constitutive ratios; quantitative data that conformed to normal distribution were expressed as mean±SD; data that did not conform to normal distribution were expressed as median and IQR. The χ² test or one-way analysis of variance was used to compare social factors and general demographic differences across participation groups. Multiple logistic regression analyses were used to analyse the influencing factors. p<0.05 was statistically significant.

Results

LPA of engagement for medication safety in patients

The LPA analysis was based on three dimensions of the Patient Participation in Medication Safety Scale, and 1–5 latent profile models were established successively (table 1). The AIC, BIC and aBIC values showed an overall decreasing trend from the one to the four-profile solution. However, the final selection of the 3-profile model was based on two decisive criteria: (1) it achieved the highest entropy (0.777) among all models evaluated, indicating superior classification clarity, whereas the entropy of the four-profile model was lower (0.724) and (2) both the Lo-Mendell-Rubin Likelihood Ratio Test (LMR) and Bootstrap Likelihood Ratio Test (BLRT) tests were statistically significant for the three-profile solution (p<0.001). In contrast, the BLRT for the four-profile model was non-significant (p=0.139), failing to justify the added complexity. Therefore, the three-profile model was selected as the optimal solution for its optimal balance of fit, interpretability and parsimony.

Table 1. Potential categories of patient involvement in medication safety.

Model	AIC	BIC	aBIC	Entropy	P value		Categorical probability (%)
					BLRT	LMR
1	4565.826	4588.360	4569.330
2	4317.582	4355.140	4323.422	0.755	<0.001	<0.001	45.57/54.43
3	4242.155	4294.735	4250.331	0.777	<0.001	<0.001	22.47/52.53/25.00
4	4228.637	4296.240	4239.148	0.724	0.139	<0.001	28.48/17.72/16.46/37.34
5	4226.523	4309.149	4239.371	0.761	0.4169	0.4272	15.82/29.75/3.48/13.92/37.02

aBIC, adjusted BIC; AIC, Akaike information criterion; BIC, Bayesian information criterion; BLRT, Bootstrap Likelihood Ratio Test; LMR, Lo-Mendell-Rubin Likelihood Ratio Test.

LPA revealed three distinct participation profiles (figure 1).

Figure 1. Reflected the mean values of the three profiles of participation levels in terms of the scores of each dimension. The three profiles were named according to the mean values of their dimension scores. Profile 1 had a lower mean value of each dimension score, so it was named ‘passive participation group’; profile 2 had a dimension score between categories 1 and 3, so it was named ‘moderate participation group’; category 3 had a higher dimension score, so it was named ‘active participation group’. The ‘passive participation group’ accounted for 22.47% of all subjects, the ‘moderate participation group’ for 52.53%, and the ‘active-level participation group’ for 25.00%.

General data about the participants

In this study, 320 questionnaires were distributed, and 316 were valid, yielding a valid recovery rate of 98.75%. The means of the total score for inpatients’ involvement in medication safety behaviour (77.84±7.14), and the average scores of the three dimensions were (26.37±3.26) (participation in decision-making), (36.77±3.13) (caring participation), and (14.70±1.88) (demands participation). Treatment burden score was (68.00 (56.00, 76.00)). Other general information is given in online supplemental table 1.

Analysis of factors influencing patient engagement in medication safety latent profiles

The results of univariate analysis showed that there were statistically significant differences in gender, education level, working status, marital status, monthly per capita household income, residence, medical payment methods, number of concurrent chronic diseases, daily medication type, primary caregiver, and the importance of conscious participation in drug safety, seven dimensions of treatment burden among the three groups (p<0.05), as shown in online supplemental table 1.

Multivariate logistic regression analysis was performed, with variables showing statistical significance in the single-factor analyses as independent variables and the potential patient categories involved in drug safety as dependent variables. The results showed that working status, marital status, medical payment methods, type of daily medication, economic burden, self-management burden, access to healthcare burden and social burden are important influencing factors across different potential categories (p<0.05), as shown in table 2.

Table 2. Logistic regression analysis of factors influencing medication safety behaviour in older adults with cardiometabolic multimorbidity (n=316).

Variables	β	SE	Waldχ2	P value	OR	95% Cl
C1:C3
Employed	1.748	0.65	7.221	0.007	5.74	1.605 to 20.534
Spouse	−18.814	0.65	837.931	<0.001	6.75	1.890 to 2.410
Medical insurance for urban employees	−1.721	0.855	4.051	0.044	0.179	0.033 to 0.956
Type of daily medication <3	1.811	0.892	4.123	0.042	6.119	1.065 to 35.156
Economic burden	0.596	0.164	13.283	<0.001	1.815	1.317 to 2.501
Self-management burden	0.452	0.145	9.77	0.002	1.571	1.184 to 2.086
Access to healthcare burden	0.268	0.072	13.782	<0.001	1.308	1.135 to 1.507
C2:C3
Employed	1.364	0.486	7.869	0.005	3.912	1.508 to 10.148
Self-management burden	0.437	0.114	14.756	<0.001	1.548	1.239 to 1.934
Access to healthcare burden	0.148	0.05	8.945	0.003	1.16	1.052 to 1.278
Social burden	0.416	0.177	5.522	0.019	1.516	1.071 to 2.145

Discussion

This study found that patient engagement in medication safety was moderate. Using LPA-derived dimension scores, the study identified distinct profiles of medication safety participation among older adults with CMM. Following the identification of these profiles, a comparative analysis was conducted to examine the influencing factors, including sociodemographic characteristics and treatment burden, across the different profiles.

The results showed that the score for patients’ medication safety engagement was (77.84±7.14), which was lower than that of hospitalised patients reported by Wang and Qin (89.64±15.90)²² and elderly hospitalised patients reported by Yongping et al (85.26±16.70).²⁴ This may be attributed to the fact that the subjects in this study are CMM and that the diseases interact closely. For example, hypertension, hyperlipidaemia and diabetes are risk factors for each other.^{25 26} As a result, the relevant control indicators are more stringent and must be considered alongside multiple factors, such as blood pressure management and lipid and glucose control. Furthermore, the current landscape of drug development and treatment mechanisms for cardiovascular metabolic diseases is constantly evolving, and strategic adjustments may be needed during drug therapy,²⁷ which places higher demands on patients’ participation in medication safety management. At the same time, this study focused on older adult patients, whose cognitive and comprehension abilities may be limited, resulting in reduced engagement in communicating with healthcare providers about their conditions and discussing medication treatment plans.

In this study, three potential categories of medication safety participation were identified through the analysis of various characteristics, which were classified into passive, moderate and active participation groups. This classification highlights the heterogeneity in patients’ participation in medication safety behaviours. These profiles align closely with the Patient Activation Model (PAM),²⁸ which conceptualises patient engagement along a continuum of knowledge, skill and confidence for self-management. Specifically, the passive group exemplifies low activation, characterised by dependent and reactive behaviours. The active group demonstrates high activation, indicative of proactive and collaborative partnership in care. The moderate group occupies an intermediate position within this PAM framework, showing emerging but inconsistent initiative. This PAM-based interpretation directly suggests a differentiated clinical approach: moving from directive support for the passive/low-activation group, through skill-building and barrier reduction for the moderate/intermediate group, to collaborative partnership with the active/high-activation group. Notably, the passive participation group accounted for 25.00% of the sample, primarily comprising patients with low education levels, low income and high treatment burden. It is likely that their limited educational background results in an insufficient understanding of medication information, hindering their ability to evaluate the medication regimens or medication instructions provided by healthcare professionals.29,31 Additionally, their low economic status and high treatment burden may further objectively restrict their engagement in medication safety behaviours. In addition, all three patient groups demonstrated lower scores in the demands participation dimension, including those categorised as having high participation. This indicates that patients were weak in the awareness of participating in the identification and notification of drug safety problems. This phenomenon may be linked to patients’ trust in medical staff or a fear of offending healthcare providers by asking questions.³²

Logistic regression results indicated that occupational status, marital status, medical payment method and type of daily medication were significant factors influencing patients’ participation in medication safety.33,35 Notably, employed patients were more likely to be classified as passive and moderate participants. This may be attributed to competing work demands, which can limit the time and cognitive resources available for proactive health management. This observation finds indirect support in the literature, which suggests that transitioning out of the workforce (retirement) can be associated with improvements in subjective health and well-being.³⁶ It is therefore plausible that sustained workforce engagement presents a contextual barrier to intensive self-management activities such as medication safety. Patients taking fewer than three daily medications were classified as having a passive level of participation. This finding contrasts with established literature indicating that polypharmacy is a key risk factor for medication confusion and non-adherence.^{34 37} To reconcile this, we propose a testable hypothesis: the relationship between medication burden and engagement may be nonlinear (U-shaped). While high burden overwhelms patients, very low burden may fail to provide sufficient daily cues for active self-management, potentially leading to passive disengagement. This hypothesis shifts the clinical implication from merely reducing pills to tailoring support based on a patient’s specific burden context. For instance, patients on minimal regimens may need interventions to increase health salience and build self-management habits, moving beyond a one-size-fits-all approach. Consequently, they showed low levels of participation in medication safety behaviours. Patients who have a spouse and those covered by urban employee medical insurance are more likely to be classified as active participants in medication safety. This pattern is consistent with the literature underscoring the role of social support and economic resources in enabling health management.^{34 38 39} The presence of a spouse likely provides essential psychosocial and practical support, which has been linked to better self-management, while comprehensive medical insurance reduces financial barriers to care, a known determinant of adherence and engagement. Therefore, in clinical practice, priority should be given to identifying and intervening in passive patients who lack spousal support and adequate medical insurance. The intervention should focus on compensating for these key resource gaps: by proactively involving alternative supporters (eg, adult children, community health workers) in education, assisting them in accessing medication financial assistance programmes and co-creating simplified medication regimens, thereby directly reducing their barriers to engagement and treatment burden.

In addition, this study found that financial burden, self-management burden, access to medical services burden and social burden are hindrance factors for patients participating in medication safety. A higher self-management burden was significantly associated with a greater likelihood of belonging to the passive engagement profile (OR=1.571, p=0.002). This burden, encompassing lifestyle changes, self-monitoring and medication knowledge,²³ presents a substantial challenge. Drawing on existing literature, we posit that a potential mechanism underlying this association is that high self-management demands may contribute to negative affective states, such as fatigue and anxiety, which are known barriers to active health engagement.^{40 41} Over time, these states may undermine motivation and capacity for the very behaviours (eg, condition monitoring, medication learning) required for safety.⁴¹ To address these challenges, integrated clinical strategies are needed that directly reduce the management burden and systematically compensate for their gaps in social and economic resources. This entails proactively building support networks for those lacking assistance, helping them access medication financial aid to alleviate cost barriers, and reducing daily management complexity through co-created simplified regimens and visual aids. Such a multi-pronged approach can enhance medication safety engagement from multiple dimensions.

A higher burden related to healthcare access was significantly associated with a greater likelihood of passive engagement (OR=1.308, p<0.001). This burden primarily involved challenges with time, distance and resource availability.²³ Building on literature that documents the digital divide faced by older adults42,44 and systemic access barriers in healthcare settings, we interpret this association as reflecting how both digital and physical barriers to care can erode a patient’s capacity and motivation for proactive medication safety management. When navigating care is persistently difficult—whether due to complex online systems or inefficient offline processes—it can displace attention and energy from day-to-day self-management tasks. These insights equip frontline clinicians with actionable strategies to mitigate access barriers within their scope of practice. When encountering a patient from a high-burden profile, clinicians should proactively assess and acknowledge specific difficulties in accessing care. They can adjust communication and education by providing clear, written, step-by-step instructions. For care coordination, they can proactively engage social work resources, help plan more efficient visit schedules to reduce logistical burdens, and formally designate a family member or caregiver as a ‘care navigator’ for ongoing support. Concurrently, co-creating a simplified medication regimen can reduce dependence on frequent clinical visits. By adapting these specific communication, coordination and educational strategies, clinicians can directly alleviate access burdens within the existing system, thereby improving their engagement in medication safety.

This study revealed that social burden—the perceived interference of illness management with everyday social and relational life—was a significant factor associated with the moderate engagement profile (OR=1.516, p=0.019). This finding can be explained through the lens of ‘competing demands’. This burden likely acts as a chronic psychosocial stressor that depletes patients’ emotional and cognitive resources,⁴⁰ leaving insufficient capacity for the initiative and collaboration required in proactive medication safety management, thereby constraining engagement to a moderate level. This internalised relational barrier extends the current understanding of chronic disease self-management, which has focused more on external, objective obstacles.⁴⁵ To address this, clinical practice requires targeted adjustments in communication, education and coordination: proactively screening for and empathetically validating social struggles in communication; providing specific education on ‘social relationship adaptation’ skills, such as planning low-burden activities or practising communication strategies; and systematically connecting patients to peer support networks while fostering understanding within their core support circle through coordination. This integrated psychosocial support directly mitigates the social burden, paving the way for patients to transition towards a more active engagement pattern.

Limitations

Several limitations of this study should be acknowledged. First, the findings are derived from a single tertiary hospital in Jiangsu Province, China, limiting the generalisability of our results to other healthcare settings, as patient engagement may be influenced by specific sociocultural factors, institutional norms (eg, healthcare delivery models, doctor–patient relationship dynamics), and patient expectations that vary across regions and countries. Second, the cross-sectional design precludes causal inferences and cannot capture how patient participation and its determinants evolve over time or across different care transitions, justifying the need for longitudinal research. Third, while we focused on treatment burden, other critical factors relevant to older adults—such as cognitive function, health literacy, family support and prior healthcare experiences—were not assessed. Their omission may mean our model does not fully capture the complexity of engagement in medication safety.

Future research should prioritise multicentre and longitudinal designs to enhance external validity and establish temporal relationships. Including the unmeasured variables would provide a more comprehensive understanding. Finally, intervention studies are warranted to test strategies to reduce treatment burden and enhance patient engagement based on the subgroup profiles identified here.

Conclusions

This study reveals significant heterogeneity in medication safety engagement among older adults with CMM, identifying distinct patient profiles based on their level of participation. A combination of socioeconomic, social support and treatment burden factors significantly influenced the profiles. These findings underscore the necessity for a shift from a one-size-fits-all approach to tailored patient management. Early identification of at-risk profiles can guide healthcare professionals in developing targeted strategies—focused on mitigating treatment burden and addressing structural and social determinants of engagement—to improve medication safety and health outcomes in this vulnerable population.

Data availability statement

Data are available on reasonable request.