Medication adherence is crucial for managing chronic diseases, especially among the older adults who are at an increased risk of polypharmacy and nonadherence. This study aimed to evaluate the patterns of medication adherence among older adult patients with diabetes in Korea and identify the factors affecting these patterns.

Using data from the 2020 Korea Healthcare Panel, we analyzed 984 patients with diabetes aged ≥ 65 years. Medication adherence was assessed in three dimensions: dosage, frequency, and timing. The independent variables included sociodemographic factors, health status, and healthcare perceptions. Latent profile analysis and logistic regression were used to identify adherence patterns and determinants.

The study population demonstrated high levels of medication adherence with average scores close to the ‘always adherent’ category across all dimensions. Two distinct adherence profiles were identified: “Adherent” (87.5%) and “Non-Adherent” (12.5%). Factors significantly influencing adherence included living alone, self-care ability, perceived stress, depression, and subjective health perception. Living alone, perceived stress, and positive health perception were correlated with higher adherence and self-care ability, and depression inversely affected medication adherence.

Older adult patients with diabetes in Korea show a high level of medication adherence. Medication adherence is multifactorial, highlighting the significant impact of non-medication factors in the older adult population.

Medication adherence is defined as the extent to which an individual’s behavior corresponds with the prescribed medication regimen from a healthcare provider [1]. The growing global aging population with comorbidities has intensified the prevalence of medication nonadherence among the older adults, a factor significantly influencing healthcare costs and patient outcomes [2]. Older adults, who are at increased risk for polypharmacy due to multiple comorbidities, exhibit a greater tendency for nonadherence compared to younger populations. Factors such as polypharmacy and absence of specific symptoms attributable to the chronic nature of the disease contribute to poor patient adherence. Notably, nonadherence among the older adults leads to adverse clinical outcomes, including mortality, hospitalization, and falls [3, 4]. Pharmacotherapy is vital in managing diabetes, where nonadherence can lead to significant treatment failures. Studies have shown that medication adherence in patients with diabetes is associated with cardiovascular and dyslipidemia outcomes. Furthermore, nonadherence contributes to physiological aging in the older adults, resulting in higher mortality rates than in younger individuals [4, 5].

Studies have been conducted on drug nonadherence in various chronic diseases related to drug factors [6, 7]. In diabetes management, the complexity of medication regimens, including the specific dosage forms, frequency, and instructions, has been considered a significant barrier to adherence, presenting opportunities for targeted interventions to simplify medication routines [8]. However, when considering medication nonadherence in the older adults, a comprehensive examination of various factors is essential, including cognitive function, personal beliefs, and health literacy. Drug factors, such as multiple medications, complexity of prescription, and experience of side effects, should be evaluated. Furthermore, socioeconomic factors and the quality of patient–healthcare provider interactions also play crucial roles in adherence [9, 10]. Research has focused on identifying the determinants of medication nonadherence in older adult patients and analyzing their ability to manage medication regimens effectively. This includes tasks such as monitoring medication quantities, interpreting labels, and maintaining a consistent medication intake schedule [11, 12]. Previous studies have highlighted the influence of factors such as regimen complexity, patient–healthcare provider relationships, and socioeconomic status on adherence; however, there is a lack of research focusing on the combined effect of these factors on older adult populations within specific cultural contexts.

Our study evaluated the multifactorial phenomenon of medication adherence by analyzing the patterns among the older adult population in Korea, focusing on medication use behaviors and population characteristics. Identifying medication use patterns and adherence levels among older adults is imperative for developing effective strategies to enhance adherence. By employing a holistic approach to evaluate these determinants, this study contributes to the existing body of knowledge by offering insights into the complexities of medication adherence in an aging society.

In Korea, nonadherence to diabetes medication poses a significant challenge, leading to adverse clinical outcomes. In addition, research on multifaceted drug use patterns related to medication adherence is lacking [13, 14]. Given these challenges, our study sought to address the critical gap in understanding medication adherence among the older adults in Korea. This study aimed to comprehensively analyze these patterns by employing a multifaceted approach to understand the factors affecting medication adherence. In doing so, this study aimed to provide insights that will inform interventions designed to improve health outcomes among the older adults.

This study used 2020 data from the Korea Healthcare Panel (Version 2.1), which is jointly organized by the Korea Institute for Health and Social Affairs and National Health Insurance. The Korean Healthcare Panel is a representative panel survey of Korean healthcare that produces basic data to establish healthcare policies by identifying the usage behaviors of healthcare service consumers and evaluating the effectiveness of policy implementation. The survey was conducted using a two-stage stratified cluster sampling method based on 2016 population census data. Of the original sample data from 2020, 984 individuals were identified as patients with diabetes aged ≥ 65 years. Medication adherence was evaluated for these individuals, with diabetes categorized as a chronic disease without differentiating between type 1 and type 2 in the responses.

This study delineates the variables into independent and dependent categories. These variables included sex (male = 0, female = 1), age (treated as a continuous variable), location of residence (large city = 0, small city = 1), level of education (no education = 1, elementary school = 2, middle school = 3, high school = 4, university or higher = 5), and living arrangements (living alone = 1, not living alone = 0). To accommodate the analyses, annual income was adjusted to reflect household size differences, using a normalization process where income was divided by the square root of household members and subsequently log-transformed to achieve a normal distribution. Moreover, we included binary-coded variables for disability status (none = 0; present = 1) and enumerated the presence of chronic diseases among the participants. The spectrum of chronic conditions surveyed encompasses various diseases, including but not limited to rheumatoid arthritis, herniated discs, diverse spinal disorders, cancers, cardiovascular diseases (angina, myocardial infarction), cerebrovascular accidents (cerebral hemorrhage, cerebral infarction), respiratory diseases (asthma, emphysema, chronic obstructive pulmonary disease, bronchiectasis), thyroid disorders (hypothyroidism, hyperthyroidism), mental health conditions (depression, bipolar disorder), dementia, chronic kidney disease, and end-stage renal disease.

Additionally, the study evaluated healthcare coverage of the participants (health insurance = 0, medical benefits = 1), self-care ability (inability to bathe or dress = 1, some difficulty = 2, no difficulty = 3), caregiving requirements (none = 0, required = 1), experience of medication side effects (none = 0, yes = 1), perceived stress (not at all = 1, a little = 2, a lot = 3, very much = 4), depression status (none = 0, present = 1), and subjective health perception (very poor = 1, poor = 2, middle = 3, good = 4, very good = 5).

The dependent variable was medication adherence, which was categorized into three aspects: dosage, frequency, and timing of medication.

The evaluation of adherence employed a four-point Likert scale, where 1 signified never adherent, 2 = rarely adherent’, 3 = mostly adherent’, and 4 = always adherent’, with higher scores indicating greater adherence. Adherence was assessed across three specific aspects: dosage, frequency of dosages, and administration time. Each aspect was scored through self-reported evaluations by individuals taking diabetes medication using a structured self-administered questionnaire provided in the Korea Healthcare Panel (Version 2.1) database. Participants responded to detailed items measuring their adherence behaviors. Higher scores represented greater adherence.

This study extracted the latent population using Mplus 8.0, to categorize the medication adherence of older adult patients with diabetes. SPSS 29.0 was used to verify the factors affecting medication adherence. The data analysis methods included frequency analysis and descriptive statistical analysis using the SPSS 29.0 program to identify the general characteristics of the study subjects and the characteristics of medication adherence. Latent profile analysis (LPA) was conducted using Mplus 8.0, to identify different types of medication adherence among older adult patients with diabetes. The use of LPA allowed us to explore the existence of subgroups within the older adult population with diabetes that exhibited unique adherence behaviors, offering a nuanced understanding of this phenomenon. Goodness-of-fit indices, including the p-values of AIC, BIC, SABIC, Entropy, BLRT, and number of cases per type, were used to determine the number of adherence types [15, 16]. As recommended by Muthen and Muthen (2000), lower values of AIC, BIC, and SABIC indicate better goodness-of-fit. Entropy is a measure of the average classification accuracy of a model and ranges from 0 to 1. The closer the value is to 1, the more accurate is the classification. The BLRT method compares the k-1 cluster model to the k-cluster model when the number of latent classes is k. If a comparison was statistically significant, a k-cluster model was adopted. Following Hill et al., [16] stratified analyses were conducted if each identified group constituted > 1% of the overall sample. Furthermore, logistic regression analysis was performed using the SPSS software (version 29.0) to examine the impact of medication adherence type on older adult patients with diabetes, providing insights into the behavioral aspects influencing medication compliance.

The general characteristics of the participants are listed in Table 1. There were slightly more females than males (445 males (45.2%) vs. 539 females (54.8%] ). The mean age was 73.45 years (SD = 5.56). A significant majority, 671 participants (68.2%) were living in large areas, contrasting with 313 participants (31.8%) residing in smaller urban areas.

In terms of educational attainment, 106 participants (10.8%) had not received any formal education, 379 (38.5%) had completed elementary education, 204 (20.7%) had completed middle school, 207 (21.0%) had graduated from high school, and 88 participants (8.9%) had a college degree or higher level of education. The majority of the study population, 756 individuals (76.8%), did not live alone, in contrast to 228 individuals (23.2%) who lived independently. The median annual income of the participants was $18,760.34 (SD = 16,267.93).

Regarding disability status, 832 participants (84.6%) did not have a disability, whereas 152 participants (15.4%) were living with a disability. Concerning self-care abilities, 837 respondents (85.1%) reported no difficulty in bathing or dressing. A large proportion of the cohort (912 individuals, 92.7%) did not require nursing care. The mean number of chronic conditions reported was 2.91 (SD = 1.27), underscoring the complexity of managing multiple health conditions in this age group.

Health insurance coverage was extensive within the cohort, with 914 (92.9%) participants receiving health insurance. The average perceived stress level is measured at 1.89 (on a four-point scale) (SD = 0.82) Regarding mental health, 883 participants (89.7%) reported no signs of depression, whereas 101 participants (10.3%) indicated experiencing mild depression. Subjective health status averaged 2.81 (on a five-point scale) (SD = 0.89).

Focusing on the primary variable of interest, medication adherence, the descriptive statistics demonstrated high levels of adherence among participants: in the four-point scale, the average adherence to medication dosage was 3.90 (SD = 0.32), adherence to the frequency of dosages was 3.87 (SD = 0.32), and adherence to the administration timing was 3.74 (SD = 0.47) (Table 2).

The number of adherence types was determined using the AIC, BIC, SABIC, Entropy, BLRT, and the number of cases per type. The results of the latent profile analysis showed that the AIC, BIC, and SABIC decreased as the number of types increased, indicating that the four types were the lowest AIC, BIC, and SABIC. The entropy had a higher value than the other types, with a p-value of 0.998 for both latent class models. The results indicated that BLRT was significant at p <.001 level for two, three, and four. However, the practical relevance of the three- and four-profile solutions was limited because each was represented in less than 1% of the total case pool. Consequently, this two-profile model was selected as the most representative schema for subsequent analyses and discussion within the study (Table 3). The two identified adherence profiles consist of an adherent group and a non-adherent group, as detailed in Fig. 1. Figure 1 illustrates the types of medication adherence determined through latent profile analysis. The analysis discerned two distinct groups: Type 1, referred to as the “adherent group,” consisted of 861 (87.5%) patients who mostly adhered to a dose, frequency, and time of day and achieved 4 points. Type 2, consisting of 123 (12.5%) patients, was termed the “non-adherent” because their medication adherence was mostly around 3 points, indicating less consistent medication adherence compared with type1.

Estimation of the types of adherence

Full size image

Table 4 presents the results of the logistic regression analysis that determined the factors affecting medication adherence. The study model was a statistically significant fit (χ2 = 29.058, p <.05), highlighting that classification of the type is based on several key determinants of adherence behavior base. In examining the factors influencing adherence patterns, logistic regression analysis identified living alone, self-care ability, perceived stress, depression, and subjective health perception as significant determinants. Specifically, living alone (B = 0.638, p <.05), perceived stress (B = 0.243, p <.05), and subjective health status (B = 0.257, p <.05) were significant positive determinants, while self-care ability (B = -1.150, p <.01) and depression (B = − 0.569, p <.05) were significant negative determinants. Individuals living alone, those with higher perceived stress, better subjective health, lower self-care ability, or without depression were more likely to belong to the adherence group.However, variables such as sex, age, region of residence, educational level, annual income, disability status, number of chronic diseases, healthcare coverage type, caregiving needs, and experience of adverse drug reactions did not significantly affect medication adherence.

This study investigated medication adherence among older adult patients with diabetes using data from the 2020 Korean Medication Panel. This reveals that the vast majority of the participants, who were community-dwelling older adults, were covered by the National Health Insurance, with a notable coverage rate of 92.9% (914 individuals). This extensive coverage highlights the representativeness of the sample in terms of health insurance penetration among the older adult Korean population. Additionally, the study reported a disability incidence of 15.4% (152 individuals) and a requirement for caregiving services of 7.3% (72 individuals), indicating the health and social care needs of the group. Most participants resided in urban areas, with a preference for living in communal households over living alone. This trend underscores the specific urban and social characteristics of older adult populations.

Quantifying medication adherence is challenging due to the absence of a universally accepted gold standard method. In this study, self-reported medication adherence across dosage, frequency, and timing showed high levels, with 87.5% of participants achieving optimal adherence. These results are inconsistent with other research results when compared with similar data; these discrepancies may be due to differences in research methodologies and adherence measurements [13, 17]. Despite high adherence across all metrics, the timing of medication intake was slightly lower than dosage and frequency. This pattern was consistent across the identified adherence profiles, indicating that individuals with high medication adherence consistently exhibited similar behaviors regarding dosage, frequency, and timing of intake. These insights highlight the nature of medication adherence behaviors among patients and underscore the importance of considering various dimensions of adherence in both clinical practice and research.

Medication adherence among the older adults is influenced by a broad spectrum of determinants, including patient characteristics, medication attributes, healthcare provider factors, healthcare systemic issues, and socioeconomic factors [10]. Numerous attempts have been made to evaluate the reasons for non-adherence among the older adults and produce recommendations for multifaced fatcors [18]. A significant concern is the challenge of polypharmacy driven by the multiple morbidities common to this demographic. Consequently, there has been considerable focus on the impact of medication use in the older adults. Several studies have demonstrated that medication-related factors, such as the number of medications, side effects, and regimen complexity, are associated with adherence in older adults [19, 20]. Our study explored the number of chronic diseases and experiences of drug side effects, which may contribute to the overall medication burden.

The findings revealed that the elderly patients in our study had a mean age of 73 years and an average of 2.91 comorbidities, a relatively lower figure compared to previous studies on similar age groups. This result may be attributable to the study design, which targeted community-dwelling older adults capable of participating in self-administered surveys. These participants appear to represent a population with fewer comorbidities and greater functional independence. Given that adherence can also vary depending on specific disease types and the overall medication burden, further studies should incorporate these factors to provide a more comprehensive understanding of adherence [21,22,23,24]. While medication-related factors are often highlighted, our results indicated that non-medication-related factors were significant predictors of adherence. Notably, variables such as living alone, perceived stress, and positive subjective health perceptions emerged as predictors of good medication adherence. Conversely, factors such as depression and the ability to bathe and dress independently were associated with poor adherence.

Regarding patient factors, previous studies have shown inconsistent results regarding the presence of a caregiver and living alone in terms of medication adherence among older adults [25, 26]. Our study suggests that living alone may enhance medication adherence. This suggests the potential for greater independence and self-management capabilities among those living alone, highlighting the importance of personalized patient assessments rather than generalized approaches. Furthermore, our study indicated that functional disabilities such as the inability to bathe and dress oneself underscore the need for assistance in medication management, potentially leading to good adherence. This suggests a complex interplay among personal autonomy, perceived health, and the practical aspects of medication management that requires further investigation.

This study found significant associations between daily stress perception, subjective positive health perception, and improved medication adherence. This finding is consistent with previous research on the impact of positive health expectations on good medication adherence [27].

Although the influence of stress perception may be constrained by the variability of its assessment, it highlights the significant impact of psychological factors on medication adherence, particularly among the older adult population. This area warrants further investigation to deepen our understanding of the relationship between psychological state and medication adherence outcomes. Consequently, there is a societal need for precise definitions and valid measures of psychological factors related to medication adherence.

Depression has been extensively studied as a determinant of medication adherence, especially in older adult populations [10]. Numerous studies have identified depression as a predictor of medication nonadherence [4, 26]. Our study also found a negative correlation between depression and nonadherence. In patients with diabetes, depression affects not only pharmacological treatment but also non-pharmacological management, resulting in negative clinical outcomes. This highlights the importance of managing depression in patients with medication nonadherence. Furthermore, the link between mental health status and medication adherence is a reminder of the biopsychosocial nature of health management in the older adults, advocating for comprehensive interventions that extend beyond mere medication reminders.

The findings of this study confirmed the significance of psychological self-regulation in promoting medication adherence among older adults. In our study, the findings highlight that healthcare providers and policymakers should consider these characteristics when managing medication adherence in community-dwelling older adults with fewer comorbidities. The results were obtained through a comprehensive review of the medication and non-medication variables. This study provides healthcare providers and public officials with insights into medication use among the older adult patients. To improve medication compliance, it is important to consider a holistic approach that includes addressing patient stress, health perceptions, and depression, potentially through the reinforcement of patient education on health awareness and literacy [12, 28, 29]. The multifaceted determinants of adherence identified in this study underscore the complexity of medication management among the older adults, necessitating a multidimensional approach to intervention design. Recognizing diversity within the older adult population is crucial for developing tailored strategies that address specific barriers to adherence.

However, this study has some limitations that should be considered. Self-reporting may have led to an overestimation of medication adherence, despite being a widely accepted and validated method for assessing medication adherence among the older adults [30]. Additionally, the cross-sectional nature of the study limited the ability to infer causality between the identified factors and medication adherence. Second, the secondary analysis of a nationwide survey had limitations in thoroughly examining medication-related factors in this demographic population. The lack of specific information on the complexity of medication prescriptions, including medication dosage, requires further research. As medication adherence is influenced by both the type and severity of disease, the absence of disease-specific and severity data could potentially skew the findings [7, 31]. Understanding the treatment environments of respondents (e.g., hospital or clinic outpatient settings) is crucial. However, survey data limitations restrict detailed analysis due to the limited scope of survey items. Future studies should address this by identifying treatment environments and incorporating comparative research with relevant variables.

Third, while LPA was employed to classify different adherence types, the small number of cases in certain profiles posed a methodological limitation. We adopted the criterion proposed by Hill et al. (2000), which considers comparisons between groups feasible if the sample size of each group exceeds 1% of the total sample. Accordingly, this 1% threshold was applied in our study. However, as noted in the results, the number of cases within each profile was limited, which may have affected the robustness of the classification. Future studies should explore strategies to enhance the practical applicability of LPA in similar contexts, possibly by utilizing larger datasets or alternative clustering approaches to ensure meaningful subgroup differentiation.

Lastly, mental health conditions and chronic diseases like dementia are closely linked to medication adherence and must be considered. In this study, the number of respondents with such conditions was insufficient for use as an independent variable, reflecting the characteristics of the sample. Future research should focus on these factors, particularly if data sources can be combined with survey results, to better understand their impact on adherence. Additionally, this study relied on self-reported depression measures, which are subjective. Future studies should adopt objective outcomes for more accurate assessment. The development of patient-reported outcome tools that integrate psychological and medication-related indicators will be essential for comprehensive adherence management.

AIC:

Akaike Information Criterion

BIC:

Bayesian Information Criterion

BLRT:

Bootstrap Likelihood Ratio Test

COPD:

Chronic Obstructive Pulmonary Disease

LPA:

Latent Profile Analysis

SABIC:

Sample-Size Adjusted Bayesian Information Criterion

Download references

S.L.: Participated in the study design, interpretation of data and paper draft. K-H J.: Participated in the study design, acquisition of data and interpretation of data. and reviewed the manuscript. All authors read and approved the final Manuscript.

Correspondence to Kyu-Hyoung Jeong.

All methods were performed in accordance with the Declaration of Helsinki. The Korean Medical Panel survey used in this study was conducted with approval from the Institutional Review Board of the Korea Institute for Health and Social Affairs. Every participant provided written consent prior to their participation in the study. This study received approval from the Institutional Review Board of Sunchon National University(IRB No; 1040173-202411-HR-048-02).

Not applicable.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, 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 you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. 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-nc-nd/4.0/.

Reprints and permissions