Exploring the relationship between health knowledge, health literacy, self-care, self-efficacy, and glycemic control among adults with type 2 diabetes mellitus

Abstract

BACKGROUND

Adults with type 2 diabetes mellitus (T2DM) in Malaysia continue to have substantial comorbidities and struggle to achieve glycemic targets.

AIM

To comprehensively evaluate diabetes self-care and glycemic control using multiple self-reporting questionnaires.

METHODS

Adults diagnosed with T2DM attending the Seremban Health Clinic were recruited in this cross-sectional study. Eligible participants were recruited based on a consecutive sampling technique, first-come-first-served-basis if they fulfilled the inclusion and exclusion criteria. In addition to the usual sociodemographic, clinical, and laboratory data, the participants answered seven specific self-reporting questionnaires. This report was focused on six key variables: Glycemic control; self-care; self-efficacy; diabetes knowledge; health literacy; and medication adherence.

RESULTS

A total of 100 adults with T2DM participated. The proportions of participants achieving specific thresholds in the key variables were: Acceptable glycemic control 39.4%; adequate diabetes knowledge 59.6%; sufficient or higher health literacy 80.2%; and medication adherence 51.0%. The mean self-efficacy score was 110.6 (73.3% of maximum), and the mean self-care score was 30.7 (43.9% of maximum). A statistically significant linear correlation was observed for eight pairs of key variables with Pearson’s correlation values varying between 0.21 to 0.55. Self-efficacy had a relatively higher correlation while glycated hemoglobin was not correlated with other key variables. Path analysis was conducted to examine the relationships among diabetes self-efficacy (Diabetes Management Self Efficacy scale score), self-care behavior (Summary of Diabetes Self-Care Activities score), and glycemic control, but the model demonstrated a poor fit (χ² = 28.1, P < 0.001).

CONCLUSION

We found substantial suboptimal glycemic control and low self-care practices but acceptable levels of diabetes knowledge, self-efficacy, health literacy and medication adherence among the patients with T2DM. In spite of the correlations between self-care, self-efficacy, and medication adherence, it was surprising that self-care did not correlate with glycemic control. Prospective cohort studies are needed to explore whether these factors influence diabetes outcomes.

Key Words: Adult; Glycemic control; Health knowledge; Malaysia; Primary health care; Self-care; Self-efficacy; Type 2 diabetes mellitus

Core Tip: A comprehensive evaluation of multiple variables among adults with type 2 diabetes mellitus revealed substantial suboptimal glycemic control and low self-care practices but acceptable diabetes knowledge, self-efficacy, health literacy, and medication adherence. Self-care was associated with self-efficacy and medication adherence. Prospective cohort studies are needed to explore whether these factors influence diabetes outcomes.

INTRODUCTION

Type 2 diabetes mellitus (T2DM) has emerged as a major heath concern globally. The 2019 Global Burden of Disease Study reported a worldwide T2DM prevalence of 5282.9 per 100000 persons, a staggering 49% increase compared with the year 1990[1]. While high-income countries have experienced declines in cardiovascular complication rates and all-cause mortality in people with T2DM, the reverse is observed in low-income and middle-income countries[2]. In Malaysia the prevalence of diabetes mellitus in adults steadily increased with each national survey. The latest National Health and Morbidity Survey in 2019 reported that 18.3% of adults aged ≥ 18 years have T2DM[3]. The National Diabetes Registry 2019 reported substantial comorbidities and non-achievement of diabetes care treatment targets [e.g., in 2019 only 32.41% of patients with T2DM achieved glycated hemoglobin (HbA1c) ≤ 6.5%][4]. The standardized mortality ratio among adults diagnosed with T2DM in West Malaysia for the period 2010-2019 was 1.83 for males and 1.85 for females[5]. Unsurprisingly, T2DM has major health economic impacts in this country; Ganasegeran et al[6] estimated the total annual cost of diabetes in Malaysia as 600 million USD.

Social cognitive theory, developed by Bandura[7], emphasizes the role of observational learning, social experiences, and reciprocal determinism in behavioral change. According to social cognitive theory, self-efficacy is a central component that influences an individual’s ability to perform self-care behaviors and adhere to medication regimens. Higher self-efficacy leads to greater confidence in managing diabetes, which in turn promotes better self-care and medication adherence.

Glycemic control and the occurrence of diabetes-related complications are influenced by various modifiable and non-modifiable factors, including genetic risk, sociodemographic factors, health knowledge, health literacy, self-efficacy, self-care, and medication adherence[8,9]. Self-care refers to the persistent practice of seven behaviors that promote better glycemic control, namely healthy eating, being active, medication intake, monitoring, risk reduction, problem solving, and healthy coping[10]. Health literacy is a set of “skills that enable individuals to obtain, understand, appraise, and use information to make decisions and take actions that will have an impact on health status”[11]. Self-efficacy refers to “people’s beliefs about their capabilities to produce designated levels of performance that exercise influence over events that affect their lives”[12]. Health literacy and self-efficacy promote better glycemic control but mostly via the mediation of optimal diabetes-specific self-care practices[9,13-15]. Medication adherence “refers to the act of conforming to the recommendations made by the healthcare provider with respect to timing, dosage, and frequency of medication taking”[16]. The systematic review by Capoccia et al[17] reported that higher medication adherence in patients with T2DM was associated with improved glycemic control.

Self-efficacy, self-care, and glycemic control (measured using HbA1c) have been assessed together previously[18-20]. However, there is still a lack of studies that comprehensively assess multiple factors influencing outcome measure such as glycemic control or self-care. Such evaluation can explore the inter-relationship of multiple variables and their potential roles on the outcome measures and help to identify modifiable factors that can become targets of future diabetes care interventions.

MATERIALS AND METHODS

Cross-sectional study

Study site: Klinik Kesihatan Seremban (KKS, Seremban Health Clinic) is a large public primary care clinic located in Seremban, the capital of the state of Negeri Sembilan, Malaysia. It has daily attendances between 500 to 800 per day and is designated as a “type 2” Klinik Kesihatan (health clinic)[21].

Study participants: The inclusion criteria were adult patients with T2DM having follow up in the Non-Communicable Disease Clinic of KKS as well as being recorded in the National Diabetes Registry. Females with gestational diabetes, children, or adults with type 1 diabetes were excluded. The diagnosis and management of diabetes patients at KKS follow the recommendations of Malaysian Clinical Practice Guidelines for T2DM[22].

Participant recruitment and consent: Consecutive sampling technique was used for participant recruitment. Eligible patients were approached during their regular diabetes follow-up visits on a first-come-first-served basis as long as they fulfilled the inclusion and exclusion criteria. The study questionnaires were self-administered. However, for illiterate participants standard face-to-face interviews were conducted using their preferred language version. All participants provided written informed consent prior to participation in the study.

Study questionnaires included sociodemographic, clinical, and laboratory data with seven specific questionnaires measuring various aspects relevant to diabetes care (Table 1). Questionnaires in English and three other versions in local languages (Malay, Mandarin, Tamil) were obtained from the developers. The study questionnaires were administered during face-to-face interviews using the language version preferred by the participants. Patients with higher literacy levels were allowed to complete the questionnaire independently. The clinical and laboratory data were obtained via patients’ paper-based and electronic medical records (TelePrimary Care-Oral Health Clinic Information System) and laboratory database (Chemo Laboratory Information Management System).

Table 1 Description and psychometrics of study questionnaires[23-37].

Number	Domain	Description	Psychometrics
1	Sociodemographic and clinical data	24 items, e.g., age, gender, ethnicity, blood pressure, body weight, diabetes complications, etc	Not applicable
2	Laboratory data	8 specific laboratory test results, e.g. fasting glucose, fasting lipids, HbA1c	Not applicable
3	Self-care	There are several versions of Summary of Diabetes Self-care Activities scale. It was developed by Toobert et al[23]. The self-care activities of patients with diabetes included diet, physical activity, blood glucose self-monitoring, foot care, smoking, and medication adherence. The 10-item version (excluding smoking and medication adherence) was validated in Malaysia by Bujang et al[24]. The version used in this study had 11 items (five diet items instead of four)	Cronbach’s α of subscales varied between 0.651 and 0.905 as reported by Bujang et al[24] (based on 10-item version)
4	Self-efficacy	Diabetes Management Self Efficacy Scale-15, a 15-item scale measuring the confidence of respondents in their ability to carry out actions that promote diabetes management. It has been validated in the United Kingdom by Sturt et al[25]. This scale was a shortened version of the original 20-item version developed by Bijl et al[26]	No prior validation had been published from Malaysia
5	Health knowledge	Michigan Diabetes Knowledge Test-14 was developed by Fitzgerald et al[27]. Some versions include another 9 items on insulin use (these were not included in this study). Threshold of adequate diabetes knowledge is a score ≥ 7	Cronbach’s α = 0.702 as reported by Al-Qazaz et al[28]
6	Health literacy	Health literacy is the ability to access, read, understand, and use health information to make appropriate healthcare decisions and follow instructions for treatment. Malaysian version of Health Literacy Survey is a shorter 18-item version developed from the European Health Literacy Survey[29]. Threshold for limited health literacy was set at transformed health literacy score ≤ 33 (hence, sufficient or excellent level is > 33)	Cronbach’s α = 0.906 as reported by Mohamad et al[30]
7	Medication adherence	Malaysia Medication Adherence Assessment Tool is a 12-item scale developed by Hatah et al[31]. Score ≥ 54 was considered as adherence	Cronbach’s α = 0.910 as reported by Hatah et al[31]
8	Diabetes distress	Diabetes Distress Scale is a 17-item scale assessing diabetes-related emotional distress. It was developed by Polonsky et al[32]. A mean item score of ≥ 3 was considered a level of distress worthy of clinical attention. Four subscales can be computed (emotional burden, physician distress, regimen distress, and interpersonal distress)[33]	Cronbach’s α for Malay version is 0.94 as reported by Chew et al[34]. Cronbach’s α for English version is 0.92 as reported by Chin et al[35]
9	Depression	Patient health questionnaire is a 9-item scale (PHQ-9) based on the diagnostic and statistical manual of mental disorders IV that aims to measure the level of depression. It was developed by Kroenke et al[36]. PHQ-9 score ≥ 10 was indicative of moderate levels of depression. Major depression can be diagnosed if the following criteria are satisfied: (1) Item 1 or item 2 scores ≥ 2; and (2) Five items with the following scoring: Items 1-8 scores ≥ 2 or item 9 score ≥ 1	Cronbach’s α = 0.67 as reported by N Azah et al[37]

HbA1c: Glycated hemoglobin.

Sample size estimation

Sample size estimation was based on guidance from a previous study[38]. The main objective of the study was to determine correlations between variables measured in numerical form (e.g., self-care and self-efficacy). Therefore, sample size was calculated using a formula for Pearson’s correlation test. A previous study recommended that an effect size of 0.3 is minimum to indicate a sizeable effect[39]. To estimate a magnitude of correlation of at least 0.3 with minimum power = 80% and α less than 0.05, this study needed to recruit at least 84 participants[40]. After adding a non-response rate of 10%, the sample size required was 94 participants.

Statistical analysis

Statistical analysis was conducted using IBM SPSS Statistical Software version 28 (IBM Corp. Released 2021. IBM SPSS Statistics for Windows, Version 28.0. Armonk, NY, United States) and R Studio (R Studio: Integrated Development Environment for R. Boston, MA, United States, Retrieved from http://www.posit.co/). Categorical variables were presented as numbers and percentages, and continuous variables were presented as means and standard deviations (SDs). The key variables in this paper were glycemic control, self-care, diabetes knowledge, self-efficacy, health literacy, and medication adherence. The internal consistency of specific questionnaires used in this study was reported as Cronbach’s α; a value of > 0.5 was considered to be an acceptable level of reliability[41]. Outcome variables were glycemic control and self-care. Independent variables were sociodemographic variables, diabetes knowledge, self-efficacy, health literacy, and medication adherence. Linear correlation was assessed using Pearson’s correlation. A path analysis using structural equation modeling was conducted to determine the impact of selected variables on glycemic control. Statistical significance was set at P < 0.05.

RESULTS

Demographic, clinical, and medication data

A total of 100 adults with T2DM participated in this study. Their mean age was 60.3 years (SD = 10.4, range: 34-88). The mean duration of diabetes was 10.2 years (SD = 7.5, range: 1-29). The demographic data of the study participants are described in Table 2 while the achievement of audit criteria for these 100 patients with diabetes is shown in Table 3. The mean HbA1c was 7.9 mmol/L (SD = 2.1, range: 4.6-17.7 mmol/L). Achievement of HbA1c ≤ 6.5% and HbA1c ≤ 7.0% were 24.2% and 43.4%, respectively. We did not find any statistically significant association of HbA1c with sociodemographic data. Almost all patients were on antidiabetic medications (97.0%) including metformin (85.0%), gliclazide (42.0%), vildagliptin (4.0%), and insulin therapy (30.0%) (the majority of these patients were also taking oral antidiabetic medications).

Table 2 Sociodemographic data of study participants.

Variables	Frequency (%)
Age group (n = 100)
< 50	16 (16.0)
50-59	27 (27.0)
≥ 60	57 (57.0)
Gender (n = 100)
Male	51 (51.0)
Female	49 (49.0)
Ethnic group (n = 100)
Malay	20 (20.0)
Chinese	33 (33.0)
Indian	45 (45.0)
Others	2 (2.0)
Education (n = 98)
Less than form 51	54 (55.1)
Form 5 or above	44 (44.9)
Income2 (n = 97)
B40	89 (91.8)
M40 or T20	8 (8.2)

¹Equivalent of year 11 of education.

²Defined based on percentile, B40 represents households in the bottom 40% income bracket, M40 includes those in the 41^st to 80^th percentile, and T20 refers to the top 20% of income earners.

Table 3 Achievement of clinical audit criteria among study participants.

Variables	Frequency (%)
Current smoking (n = 100)	13 (13.0)
Central obesity1 (n = 97)	78 (80.4)
Controlled BP2 (n = 100)	38 (38.0)
Good glycemic control3 (n = 99)	43 (43.4)
CKD stage 3 and above4 (n = 100)	15 (15.0)
Optimal LDL-C level5 (n = 97)	41 (42.3)
Urine proteinuria or microalbuminuria (n = 89)	28 (31.5)
Statin use (n = 100)	77 (77.0)

¹Male ≥ 90 cm, female ≥ 80 cm.

²Blood pressure < 140/80 mmHg.

³Glycated hemoglobin ≤ 7.0%.

⁴Epidermal growth factor receptor < 60 mL/minute.

⁵Low-density lipoprotein < 2.6 mmol/L.

CKD: Chronic kidney disease; LDL-C: Low-density lipoprotein cholesterol.

Statistics of key variables

Three out of five specific questionnaires in this paper had acceptable reliability [Malaysian version of Health Literacy Survey Questionnaire 18, Diabetes Management Self Efficacy scale-15 (DMSES-15), and Malaysia Medication Adherence Assessment Tool-12 (MyMAAT-12)]. Another two had Cronbach’s α less than 7 [Michigan Diabetes Knowledge Test-14 and Summary of Diabetes Self-Care Activities (SDSCA-11)]. Almost two-fifths of study participants had acceptable glycemic control (HbA1c ≤ 7.0%). However, HbA1c did not show any statistical associations with any sociodemographic variables. Adequate diabetes knowledge and sufficient/excellent health literacy were found in 59.6% and 80.2% of the patients, respectively. Higher education (Form 5 and above) was associated with diabetes knowledge and self-care. Chinese patients with T2DM showed significantly lower health literacy compared with other ethnic groups (P = 0.021). The mean self-efficacy score was 110.6 (73.3% of maximum), and the mean self-care score was 30.7 (43.9% of maximum). A higher self-care score was found among patients with education at Form 5 or above (Table 4).

Table 4 Statistical data of key variables.

	Measurement or questionnaire	Cronbach’s α	mean ± SD (range)	Number (%) at specific threshold1	Association with sociodemographic data, mean ± SD
Diabetes knowledge (n = 99)	MDKT-14	0.564	7.3 ± 2.5 (1-13)	Adequate knowledge: 59 (59.6)	Higher MDKT score among patients with education at form 5 or above: 8.3 (2.4) vs 6.5 (2.4), P < 0.001
Health literacy (n = 96)	HLS-M-Q18	0.904	True score: 54.9 ± 9.7 (26.0-71.0); Transformed score: 38.1 ± 6.7 (18.1-49.3)	Sufficient/excellent health literacy: 77 (80.2)	Lower HLS score among Chinese patients compared with other ethnic groups: 51.5 (10.5) vs 56.4 (8.9), P = 0.021
Self-efficacy (n = 100)	DMSES-15	0.878	110.6 ± 26.0 (30-150)	Not applicable	No association with sociodemographic data
Self-care (n = 98)	SDSCA-11	0.601	30.7 ± 13.3 (7-64)	Not applicable	Higher SDSCA score among patients with education at form 5 or above: 33.6 (14.3) vs 28.1 (11.4), P = 0.045
Medication adherence (n = 100)	MyMAAT-12	0.909	50.8 ± 9.3 (15-60)	Adherent: 51 (51.0)	MyMAAT score higher in those aged ≥ 60 vs those aged < 60: 52.6 (8.6) vs 48.5 (9.8), P = 0.027

¹See Table 1. MDKT-14: Michigan Diabetes Knowledge Test-14; HLS-M-Q18: Malaysian version of Health Literacy Survey Questionnaire 18; DMSES-15: Diabetes Management Self Efficacy scale-15; SDSCA-11: Summary of Diabetes Self-Care Activities-11; MyMAAT-12: Malaysia Medication Adherence Assessment Tool-12; SD: Standard deviation.

As shown in Table 5, a statistically significant linear correlation was observed for eight pairs of variables: Self-care and health literacy; self-care and self-efficacy; self-care (diet subscale) and self-efficacy; diabetes knowledge and health literacy; diabetes knowledge and self-efficacy; health literacy and self-efficacy; and medication adherence with self-care (whole scale) and self-care (diet subscale). Pearson’s correlation values varied between 0.21 to 0.55 (Table 5).

Table 5 Correlation among self-care, diabetes knowledge, glycemic control, health literacy, self-efficacy, and medication adherence.

	Diabetes knowledge	Glycemic control	Health literacy	Self-efficacy	Medication adherence
Self-care	0.16 (0.128)	0.06 (0.565)	0.31 (0.003)	0.49 (< 0.001)	0.37 (< 0.001)
Self-care (diet subscale)	0.07 (0.494)	-0.03 (0.736)	0.15 (0.135)	0.32 (0.001)	0.41 (< 0.001)
Diabetes knowledge		0.07 (0.495)	0.31 (0.002)	0.21 (0.038)	0.02 (0.827)
Glycemic control			-0.16 (0.125)	-0.01 (0.898)	0.09 (0.376)
Health literacy				0.55 (< 0.001)	0.13 (0.197)
Self-efficacy					0.11 (0.286)

The result was presented using correlation coefficient (P value).

Figure 1 presented a path analysis to examine the relationships among DMSES score, SDSCA score, and glycemic control (HbA1c). The model demonstrated a poor fit: χ² (3) = 28.1, P < 0.001. DMSES score significantly predicted SDSCA score (β = 0.245, P < 0.001). However, neither SDSCA (β = 0.015, P = 0.437) nor DMSES (β = -0.05, P = 0.605) was significantly associated with HbA1c levels.

Figure 1 Path-analysis exploring association between diabetes self-efficacy (Diabetes Management Self Efficacy scale score), self-care behavior (Summary of Diabetes Self-Care Activities score), and glycemic control (glycated hemoglobin). DMSES: Diabetes Management Self Efficacy scale; SDSCA: Summary of Diabetes Self-Care Activities; HbA1c: Glycated hemoglobin.

DISCUSSION

Patient characteristics

We noted the proportion of Malay patients was lower than expected (only 20.0%) while the proportion of Malay patients recruited from an earlier study that accessed the state-level National Diabetes Registry was 64.9%[42]. The current patients’ mean age was 60 years, just over half of the patients had less than a high school education, and an overwhelming majority (91.8%) were in the B40 income category, highlighting that the patients with T2DM in the public primary care clinics were older individuals with a lower level of education and lower income.

The achievement of the following audit criteria (central obesity, HbA1c < 7%, blood pressure < 140/90 mmHg, low-density lipoprotein cholesterol < 2.6 mmol/L) were 80.4%, 43.4%, 38.0%, and 42.3%, respectively. These proportions were similar to the baseline data from a state-level National Diabetes Registry reported by Wan et al[43] (78.5%, 40.9%, 35.8%, and 35.2%, respectively).

Summarized data on key variables

The proportion of adequate diabetes knowledge in the current study was found to be 59.6%, which is similar to the aforementioned study by Al-Qazaz et al[44] (58.2%). They also found an association between diabetes knowledge and education level. A meta-analysis of the Malaysian Michigan Diabetes Knowledge Test-14 studies by Lai et al[45] based on six Malaysian studies among adult patients with diabetes reported adequate knowledge in 51.9% of patients. This meta-analysis did not find a consistent relationship between ethnicity and diabetes knowledge.

Health literacy as measured using the Malaysian version of Health Literacy Survey Questionnaire 18 in this study had a Cronbach’s α of 0.904, which was very close to the validation study reported by Mohamad et al[30]. However, the proportion of limited health literacy in our study was found to be much lower than the figure for the adults in Malaysia[46] (19.8% vs 35.0%). The apparent lower health literacy score among Chinese patients with T2DM in this study was not shown in the national study (which was conducted among the general population). Diabetes-specific health efficacy as measured using DMSES-15 in this study had a Cronbach’s α of 0.878, similar to another Malaysian study using DMSES-15 (Cronbach’s α = 0.9)[47].

Diabetes-specific self-care as measured using SDSCA-11 in this study had a Cronbach’s α of 0.601. In our SDSCA-11 question 1 was revised from “How many of the last seven days have you followed a healthful eating plan?” to “How many of the last seven days have you followed Malaysian Healthy Eating Plate with appropriate serving sizes?” We noted many respondents had difficulty understanding “Malaysian Healthy Eating Plate” and “serving sizes” despite showing a photograph of typical Malaysian Healthy Eating Plate. The difficulty faced by our respondents in answering this item was consistent with that of a national survey where four-fifths of Malaysians were not aware of Malaysian Healthy Eating Plate[48]. A meta-analysis of 11 SDSCA-10 studies conducted in Malaysia found the Cronbach’s α varied between 0.614 and 0.741 (only five of them had Cronbach’s α value of ≥ 7.0)[49]. The SDSCA-11 used in the current study needs more work before it can be used in future studies. We found better self-care associated with a higher education level, a finding that was consistent with the study by Sharoni and Wu[18].

Medication adherence as measured using MyMAAT-12 in this study had a Cronbach’s α of 0.909, similar to the reliability data reported by Hatah et al[31]. Our study showed the proportion of adherence to be 51.0%, higher than that of Hatah et al[31] (36.2% among adults with T2DM). We found a statistically significant correlation between MyMAAT-12 and SDSCA (r = 0.37, P < 0.001). Although this finding was the first report from Malaysia using these two scales, the correlation was expected since medication adherence can be considered within the self-care domain, and the medication adherence was also included in the longer version of SDSCA by Toobert et al[23].

Correlation between key variables and path analysis

Table 5 shows a statistically significant linear correlation for only eight pairs of key variables (out of 20 unique pairs) with Pearson’s correlation values varying between 0.21 to 0.55. There are limited studies assessing correlation of these variables in Malaysia. Tharek et al[20] did find a correlation between DMSES-20 and SDSCA-10 with a value close to the present study (0.54 vs 0.49). It is notable that we failed to show a linear relationship between HbA1c and SDSCA (in both whole scale and diet subscale). This lack of clear-cut linear relationship between glycemic control and self-care is in contrast with the systematic review by Brown et al[13] in which dietary adherence was found to be the most significant predictor of HbA1c. The same review suggested a strong relationship between self-care and self-efficacy as was seen in our study. The lack of linear correlation between self-care and glycemic control in our study while unexpected is not unusual in the literature. In a meta-analysis of Malaysian studies, only four out of eight such studies demonstrated statistically significant linear correlations[49]. We found a correlation between health literacy with both self-care and health knowledge only in the bivariate analysis (Pearson’s correlation, Table 5).

The path analysis indicated a poor model fit. Notably, both diabetes self-efficacy (DMSES score) and self-care behavior (SDSCA score) demonstrated weak associations with glycemic control (HbA1c). These findings may reflect limitations inherent in the cross-sectional design of the study that restricts the ability to infer temporal or causal relationships. For example, during recruitment some patients may have demonstrated high levels of self-efficacy and self-care as a reaction to recently learning about their poor glycemic control. Therefore, these findings should not be interpreted as diminishing the importance of self-efficacy and self-care. Rather, they highlight the need for future research employing prospective cohort designs to more accurately assess the influence of these factors on diabetes outcomes.

Interventions to improve glycemic control

The meta-analysis by Chen et al[50] showed that empowerment-based intervention compared with routine care was associated with reduced HbA1c (standardized mean difference: -0.20; 95% confidence interval: -0.31 to -0.08). Empowerment is a patient-centered approach “in which individuals gain the knowledge, skills, attitudes, and self-awareness necessary to influence their own behaviour, thereby improving responsibility and autonomy and obtaining power to make informed decisions”[50]. Although the medical literature showed that glycemic control can be influenced by health knowledge, health literacy, self-efficacy, self-care, and medication adherence, the impact of the latter sets of variables on glycemic control is heterogeneous and possibly indirect in some of them. Lee et al[51], using structural equation modelling, showed that improving self-care behaviors was most essential for improving glycemic control. However, to improve self-care behaviors (which can include medication adherence), healthcare providers should target improving self-efficacy, which in turn can be improved by health literacy and health knowledge. Sustained improvement of self-care practices requires multiprong approaches at the community and primary care settings using various approaches, including health education, patient support, and digital tools[52-54]. This would benefit from the incorporation of evidence-based strategies or programs that are culturally acceptable and tailored to the social/educational disadvantages of specific primary care diabetes subgroups[55].

Study limitations

The under-representation of participants of Malay ethnicity was a notable problem in the current study. Patients with T2DM of Malay ethnicity have been found to have poorer glycemic control[56] and are somewhat less likely to practice appropriate diabetes self-care[49]. The difficulty faced by the study participants with lower education (and possibly lower literacy) in answering cognitively challenging self-reporting questionnaires used in this study might have reduced the accuracy of their responses. This problem has been alluded to by Dowse[57] with regards to the use of a health literacy scale, but it is likely that a similar problem may occur with other scales using Likert responses. Krumpal[58] extensively discussed the social desirability response bias that can be seen in questionnaires assessing “sensitive” issues (i.e. sexual nature), but we cannot discount the possibility that many participants in this study may be keen to show researchers that they had “good” or “expected” behaviors regarding diabetes-related attitude and practices as well. The lack of association between our two outcome variables (HbA1c and self-care) with other key variables suggested that their relationship may be more complex and may not be easily revealed by a linear correlation via a cross-sectional study.

On another note, consecutive sampling was applied in this study. Although it is classified as a type of non-probability sampling, it is generally considered more robust than convenience sampling. Unlike convenience sampling, which often involves selecting participants who are easily accessible to the researcher, consecutive sampling involves recruiting all eligible patients who present during the study period on a first-come-first-served basis. This approach helps minimize selection bias and enhances the representativeness of the sample within the clinical setting[59].

CONCLUSION

We found substantial suboptimal glycemic control and low self-care practices but acceptable levels of diabetes knowledge, self-efficacy, health literacy, and medication adherence among patients with T2DM. In spite of the correlations between self-care, self-efficacy, and medication adherence, it was surprising that self-care did not correlate with glycemic control. Prospective cohort studies are needed to explore whether these factors influence diabetes outcomes.

ACKNOWLEDGEMENTS

We wish to thank medical students in the batch ME119 for helping with data collection. We wish to thank questionnaire developers for allowing us to use their questionnaires. We would like to thank the Director General of Health Malaysia for his permission to publish this article.