Introduction

Type 2 diabetes is a global health pandemic that threatens both public and national economic health because of the rapidly increasing number of cases and limited resources available for management (Ogurtsova et al., 2017). Indonesia, the world's fourth largest country, has a population of 272 million (Worldometer, 2019). Nearly four in every 100 (i.e., 10.7 million) Indonesians were living with a diabetes diagnosis in 2019 (International Diabetes Federation, 2019). Moreover, the prevalence of diabetes is expected to increase to 16.6 million people in 2045, with 90%–95% expected to have Type 2 diabetes (International Diabetes Federation, 2019). The high and rapidly increasing number of Type 2 diabetes cases aggregates the care burden and impacts the daily life of sufferers. Diabetes was the third leading cause of death in 2016, and about U.S. $576 million of the Indonesian health insurance (Jaminan Kesehatan Nasional) budget was spent treating this disease (Hidayat et al., 2022; Mboi et al., 2018), highlighting the importance of proper management to diabetes-related complication prevention and optimal health outcomes.

Proper self-care performance is key to preventing diabetes-related complications (Gode et al., 2022; Letta et al., 2022). Upon their diabetes diagnosis, patients in Indonesia are enrolled automatically in the chronic disease management program Prolanis, which is a program paid for by Indonesian Health Insurance from the Social Security Administrator for Health and is free for patients to attend monthly (Mahendradhata et al., 2017). Despite the Indonesian government providing money and manpower for diabetes management, the care outcome for patients has been unsatisfactory. A recent large population-based study in Indonesia surveyed 1,976 patients with Type 2 diabetes, finding only 30.8% of participants achieved the American Diabetes Association's recommended long-term serum glucose levels target (HbA1C < 7.0%), more than half reported not adhering to doctors' or nurses' recommendations, and nearly one in every three reported having at least one diabetes-related comorbidity (e.g., eye complications, cardiovascular complications, renal complications; Cholil et al., 2019). This indicates the importance of investigating and identifying the significant factors contributing to poor diabetes self-care performance to improving diabetes care outcomes.

Many factors influence self-care performance. Low knowledge levels (Mikhael et al., 2018) and depression (Alexandre et al., 2021; Shrestha et al., 2021) have been found to be associated with poor diabetes self-care performance. In addition, Ajzen (1991) proposed self-perceived behavioral control as potentially affecting related behaviors. Perceived behavioral control is defined as the perceived ease or difficulty associated with performing a particular behavior because of the presence of internal and/or external factors that can act as either inhibitors or facilitators (Ajzen, 1991, 2002). Higher perceived barriers and lower perceived control (Alexandre et al., 2021) have been found to be significantly associated with low levels of diabetes self-care performance.

The population of Indonesia has a relatively low average level of literacy (Asril et al., 2019). This, together with indigenous cultural mores related to the meaning and preparation of food, makes it difficult for Indonesians with diabetes to choose a healthy diet (Widayanti et al., 2020). Maintaining physical activity is another big challenge in Indonesia, particularly among patients with diabetes, because of modern work patterns, urbanization, technological conveniences, and gender-related expectations. Nearly one in two (45%) people with diabetes in Indonesia rarely or never completes the recommended amount of exercise (Cholil et al., 2019). A sedentary lifestyle is more evident in women than men, as women are more likely to be housewives and to work at home than men (Nurwanti et al., 2018). Previous studies addressed how urbanization relates to a sedentary lifestyle and how conflicts between traditional and Western health beliefs (healthy eating and regular follow-ups) pose challenges to preventing and managing Type 2 diabetes in developing countries (Fanany & Fanany, 2015; Fong et al., 2019). Although all of these barriers should be considered as potential factors affecting diabetes self-care performance, little is known regarding their effects on Indonesians.

In addition to the barriers to preparing/choosing healthy food and exercise, psychological well-being also influences diabetes self-care performance. Depression decreases the ability of patients to solve their health problems and reduces their motivation, interest, and concentration, all of which lowers diabetes self-management ability (Shrestha et al., 2021). Ajuwon and Love (2020) reported that about 25% of African American patients with Type 2 diabetes reported having depressive symptoms (Ajuwon & Love, 2020). However, depressive symptom status is less reported and recognized in Indonesia because the assessment fee is not covered by insurance or Prolanis (Mahendradhata et al., 2017).

The cultural and psychological characteristics of the unique health practices employed by Indonesians present challenges to those with diabetes who attempt to achieve a good glycemic outcome. In addition, the theory of planned behavior (Ajzen, 2002) suggests all relevant factors should be considered together to obtain a more comprehensive understanding of self-care performance. However, previous studies conducted in Indonesia on patients with Type diabetes 2 only explored the relationship of each factor with self-care performance and did not consider how these factors may influence each other. Therefore, it is important to elucidate using a well-developed theoretical approach the influences of knowledge, perceived barriers, and depression on Indonesians with diabetes. In light of the above, this study was designed to address the influence the abovementioned factors have on the self-care performance of Indonesians with Type 2 diabetes.

Methods Design, Participants, and Sample Size

This cross-sectional study was conducted at two hospital-affiliated outpatient departments and four public health service centers in Makassar, Indonesia, from July to August 2017. Men and women aged 20 years and above (the legal age for signed informed consent in Indonesia) who had been diagnosed with Type 2 diabetes, had received medical treatment for at least 1 year, had managed their diabetes independently, and did not have a diagnosed cognitive impairment were eligible. The sample size was estimated using G*Power 3.0.10, with the alpha level set at .05, the power set at 0.8, and the effect size set at 0.15 (Cohen, 1988), and 14 predictors. The minimum sample size was estimated at 135. However, 20% was added to buffer the possibility of missing data, giving a final required sample size of 162.

Ethical Considerations

The study protocol and the informed consent form were approved by two university institutional review boards (N201706022 and 361/H4.8.4.5.31/PP36-KOMETIK/2017). A detailed introduction and explanation of the study were given to the participants, and written informed consent was obtained from each individual. The participants' understanding of the study and their voluntary participation were both confirmed before securing informed consent.

Measures

Eligible patients with Type 2 diabetes completed the self-report questionnaires described below. The entire process took about 30–40 minutes.

Participant characteristics

Demographic (age, gender, marital status, educational level, and income), diabetes history (years with a diabetes diagnosis and their type of diabetes treatment), and obesity status data were collected from the participants. Diabetes treatment type was distinguished into three categories: oral medication, insulin, and both oral medication and insulin. Body mass index (BMI) and waist circumference were used to determine obesity status. BMI was categorized based on the World Health Organization's report for Asians, with BMI < 18.5 kg/m2 identified as underweight, between 18.5 and 22.9 kg/m2 identified as normal weight, between 23 and 24.9 kg/m2 identified as overweight, and ≥ 25 kg/m2 identified as obese (World Health Organization Western Pacific Region, 2000). On the basis of the same report, the waist circumference cutoff points of 90 cm for men and 80 cm for women were considered optimal to identify those with a BMI ≥ 25 kg/m2 and for predicting metabolic syndromes and chronic diseases. Monthly family income was estimated based on Indonesia federal poverty guidelines (World Bank Group, 2021) and categorized into four categories: poor (monthly family income ≤ US $114), lower middle (monthly family income US $114–$196.2), middle (monthly family income US $196.2–$453), and upper middle (monthly family income > US $453).

The revised diabetes knowledge test

The Revised Diabetes Knowledge Test, which includes 14 items used to test general knowledge and nine items used to test insulin usage knowledge, has been used to evaluate diabetes knowledge levels (Fitzgerald et al., 2016). All of the participants were required to answer the 14 items on the general test, whereas only patients taking insulin were required to answer the nine items on the insulin usage test. Each multiple-choice question has one correct answer. Diabetes knowledge was treated as a standardized continuous variable (number of correct answers divided by the total number of questions) to minimize the effect of the number of questions that should be answered. Thus, a score of 1 indicates perfect diabetes knowledge, and higher scores indicate a higher level of knowledge. The authors conducted a forward and backward translation of this instrument into Indonesian from the original English. Both the general and insulin usage tests have shown good reliability (Cronbach's α = .77 for the general test and .84 for the insulin usage test; Fitzgerald et al., 2016). In our pilot study, Cronbach's α values of .71 and .73 were obtained for the general and insulin usage tests, respectively. Five experts, including two endocrinologists and three diabetes care nurses, were invited to give their expert opinions. The content validity index was identified as .86 in this study.

Depression anxiety stress scale

The Indonesian version of the Depression Anxiety Stress Scale was translated from the original Lovibond and Lovibond (1995) Depression Anxiety Stress Scale and has shown good internal reliability for depression, anxiety, and stress level (Cronbach's α values of .91, .85, and .88, respectively; Damanik, 2011). In line with the purpose of this study, only 14 depression-level items were used. The participants were asked to rate their depressive symptoms for the past 2 weeks on a 4-point Likert-scale (0 = almost none, 1 = some of the time, 2 = a good part of the time, 3 = nearly all or most of the time). The total possible score range was 0–42, with higher scores indicating greater depressive symptom severity and normal = 0–9, mild = 10–13, moderate = 14–20, severe = 21–27, and extremely severe ≥ 28 (Damanik, 2011).

Perceived barrier questionnaire

The Perceived Barrier Questionnaire, previously modified from the Self-Care Inventory (Weinger et al., 2005), consists of 14 items that reflect the self-perceived level of difficulty regarding diabetes self-care and disease control. Each item was scored between 1 and 7, with 1 = extremely difficult and 7 = extremely easy. The total possible score ranged from 17 to 98, with higher scores indicating lower perceived barriers. In the pilot test for this study, the Cronbach's α of the questionnaire was identified as .72.

Self-care inventory-revised

The Self-Care Inventory-Revised (SCI-R) is used to evaluate diabetes self-care performance or capability to manage diabetes of the respondent (Weinger et al., 2005). The 12 items of the SCI-R are scored between 1 (never do this) and 5 (always do this as recommended without fail), with the aim being to show how often the respondent performs each behavior (e.g., choosing a healthy diet, self-monitoring blood glucose level, properly managing blood glucose, exercising regularly). The SCI-R total possible score ranges from 12 to 60, with higher scores indicating better self-care performance. The SCI-R has shown good reliability in a prior study (Cronbach's α = .87; Weinger et al., 2005) as well as in our pilot study (Cronbach's α = .72).

Data Collection

Doctors and nurses, who first see potentially eligible patients during routine outpatient visits, asked potential participants regarding their willingness to be interviewed before referring them to the research team. One of the researchers or Indonesian research assistants approached each potential participant, introduced the study, answered research-related questions, explained the voluntary nature of participation, and obtained informed consent before providing the self-report questionnaire. If a participant had difficulty understanding or completing the questionnaires, the researcher (or research assistant) read and explained the questions and helped them complete the form. All other data, for example, weight, height, fasting blood glucose, and HbA1c, were collected via a medical record review. The entire process took approximately 30–40 minutes.

Statistical Analysis

The data set obtained during this study was evaluated using IBM Statistics SPSS 21.0 (IBM Inc., Armonk, NY, USA). Descriptive analyses (frequency, percentage, mean, and SD) were used for demographic background information as well as for diabetes knowledge, depression, perceived behavioral control, and self-care performance. An independent t test and a one-way analysis of variance were used to investigate the differences in self-care performance scores among the various groups in terms of educational level and type of diabetes treatment. Pearson's r correlations were used for bivariate analysis to identify significant associations among the continuous variables. Statistically significant factors from the abovementioned inferential statistics were then entered into a regression analysis. Finally, a multiple linear regression analysis was conducted to determine the factors that were associated with self-care performance at a level of significance of .05 or higher.

Results Participant Characteristics

One hundred eighty-five patients with Type 2 diabetes were enrolled as participants in this study. The background data (demographics and diabetes history) of the participants are presented in Table 1. Most (n = 128, 69.2%) of the participants were women, the mean age was 59.5 (SD = 8.6) years, and 83.2% were married. Almost half were educated to the junior or senior high school level (n = 83, 40.9%), whereas slightly more than one third (n = 65, 35.1%) held baccalaureate or higher degrees. The median time since diabetes diagnosis was 8 (minimum = 1, maximum = 36) years. Oral hypoglycemic drugs were the most common form of medication (46.5%) prescribed, followed by insulin (36.7%) and a combination of the two (16.8%). Moreover, mean fasting blood glucose and HbA1c levels were 188.7 g/dl (SD = 71.7) and 8.1% (SD = 2.1), respectively.

Table 1. - Demographics, Diabetes History, and Obesity Status of the Participants (N = 185) Variable n % Age (years; M and SD) 59.5 8.6 Gender  Male 57 30.8  Female 128 69.2 Marital status  Married 154 83.2  Unmarried/widowed 31 16.8 Education  Elementary school or less 37 20.0  Junior and senior high school 83 40.9  Baccalaureate degree and above 65 35.1 Income (monthly) a  ≤ US $114 (poor) 56 30.3  US $114–$196.2 (lower middle income) 32 17.3  US $196.2–$453 (middle income) 64 34.6  > US $453 (upper middle income) 33 17.8 Family history of diabetes  No 121 65.4  Yes 64 34.6 Years since diabetes diagnosis (median and min–max) 8 1–36 Diabetes treatment  Oral medication 86 46.5  Insulin 68 36.7  Oral medication and insulin 31 16.8 Body mass index b (M and SD) 24.5 4.0  Underweight or normal 68 36.8  Overweight 47 25.4  Obese 70 37.8 Waist circumference (cm; M and SD) 91.9 9.5  Female waist circumference (n = 126)   < 80 cm 14 11.1   ≥ 80 cm 112 88.9  Male waist circumference (n = 59)   < 90 cm 14 23.7   ≥ 90 cm 45 76.3 Fasting blood glucose (g/dl; M and SD) 188.7 71.7 HbA1c (%; M and SD) 8.1 2.1

a US $1 = Rp. 14,152.33. b Body mass index = body weight (kg)/height (m2).

Diabetes knowledge, perceived barriers, depressive symptoms, and self-care performance values are shown in Table 2. The mean score for diabetes knowledge was 0.4 (SD = 0.2), whereas the mean score for perceived barriers was 54.9 (SD = 10.2). Almost one in four of the participants (25.4%) were identified as experiencing depressive symptoms. The mean for self-care performance was 38.7 (SD = 6.9), which indicated moderate self-care performance. The poorest mean score for the self-care performance subscale was “keeping food records,” followed by “recording blood glucose daily at home,” “recording blood glucose level,” and “reading food labels.” The major self-perceived barriers included physical activity, self-monitoring blood glucose, and eye checks.

Table 2. - Diabetes Knowledge, Perceived Barriers, Depression, and Self-Care Performance (N = 185) Item n % Depression a (median and [min, max]) 4 0, 36  Normal (0–9) 138 74.6  Mild (10–13) 15 8.1  Moderate (14–20) 27 14.6  Severe (21–27) 1 0.5  Extremely severe (≥ 28) 4 2.2 Item Mean SD Diabetes knowledge b 0.4 0.2 Perceived barriers c 54.9 10.2  Eye check 2.6 1.6  Self-monitoring blood glucose 3.1 2.1  Physical activity at least once in 20–30 minutes 3.7 1.9  Physical activity 3 times a week 3.7 1.9  Avoiding high-saturated-fat food 4.1 1.7  Smoke cessation 4.1 2.3  Having meal on time 4.6 1.6  Controlling the amount of meal 4.9 1.4  Looking for support system 5.0 1.3  Consuming healthy carbohydrate source 5.2 1.3  Consuming fiber source 5.3 1.3  Consuming protein source 5.3 1.3  Taking insulin 5.6 1.4  Taking oral medication 5.8 1.1 Self-care performance d 38.7 6.9  Blood glucose self-monitoring   Check blood glucose daily at home 1.8 1.2   Record blood glucose level 1.6 1.3  Regular exercise 2.7 1.5  Healthy diet   Keep food records 1.2 0.7   Eat meal/snacks punctually 2.3 1.1   Read food labels 2.4 1.5   Eat foods in correct portions 3.7 0.9  Medication   Take diabetes pills/insulin punctually 2.6 2.4   Take the correct dose of diabetes pills/insulin 2.8 2.3  Visit the clinic based on the appointed schedule 3.4 1.6  Maintain the blood glucose level 3.7 1.1  Take rapid-acting sugar to resolve low blood  glucose 4.6 0.8

a Indonesian version of Depression Anxiety Stress Scale. b The Revised Diabetes Knowledge Test. c Perceived Barrier Questionnaire. d Self-Care Inventory-Revised.

Factors Associated With Self-Care Performance

The relationship between different patient characteristics and diabetes self-care performance is presented in Table 3. Patients with higher levels of education (F = 13.1, p < .001) and with an upper middle level of income (F = 3.1, p = .02) had relatively higher self-care performance scores. Furthermore, diabetes knowledge (r = .2, p < .001) and perceived barriers (r = .6, p < .001) were shown to have statistically significant and positive relationships with self-care performance. In contrast, depression level (r = −.4, p < .001) shared a negative association with self-care performance.

Table 3. - Differences Patients' Characteristics in Relation to Self-Care Performance Variable Self-Care Performance Mean SD t/F p/Post Hoc Gender t = −1.5 .13  Male .6 .1  Female .6 .1 Married status t = −1.7 .08  Unmarried/widowed .6 .1  Married .6 .1 Educational level 13.1** < .001  ① Illiterate to elementary school .5 .1 ① < ②, ① < ③  ② Junior and senior high school .6 .1  ③ Baccalaureate degree and above .7 .1 Income (monthly) 3.1* .02  ① < US $114 (poor) .6 .1 ① < ③, ① < ④  ② US $114–$196 (lower middle income) .6 .1  ③ US $197–$453 (middle income) .6 .1  ④ > US $453 (upper middle income) .6 .1 Family history of diabetes t = −1.1 .25  No .6 .1  Yes .6 .1 Diabetes treatment  Oral medication .6 .1 0.5 .56  Insulin .6 .1  Oral medication and insulin .6 .1 Body mass index 0.8 .43  Underweight or normal .6 .1  Overweight .6 .1  Obese .6 .1

*p < .05 (two-tailed). **p < .01 (two-tailed).

The factors contributing to diabetes self-care performance is shown in Table 4. Educational level, income, depression, diabetes knowledge, and perceived barriers accounted for 40% of the total variance in self-care performance. Having an elementary school or lower level of education (ß = 4.6, p = .002), having a junior or senior high school education (ß = 3.0, p = .006), and having moderate depression (ß = −0.9, p = .04) as well as level of diabetes knowledge (ß = 0.1, p = .006) and level of perceived barriers (ß = 0.3, p < .001) were all found to be significantly associated with self-care performance.

Table 4. - Regression Analysis of the Patients' Characteristics and Their Self-Care Performance Variable Self-Care Performance ß Standard Error t p Adjusted R 2 Illiterate and elementary school 4.6 1.4 3.2** .002 .4 Junior and senior high school 3.0 1.2 2.8** .006 Poor income 0.0 0.7 0.0 .970 Lower middle income 0.2 0.7 0.2 .810 Middle income 0.0 0.6 0.0