Determinants of glycemic control among persons living with type 2 diabetes mellitus attending a district hospital in Ghana

Study design

A cross-sectional descriptive study was conducted to determine factors associated with glycemic control among Type 2 diabetes mellitus patients attending the out-patient unit of the Shai-Osudoku District Hospital from October to December 2022. Patients were sequentially enrolled on the study and face-to-face interviews were conducted. Data on socio-demographic characteristics, lifestyle and history of diabetes were collected from the patients using a structured questionnaire. Anthropometric and glycated hemoglobin measurements were also done. The patients’ clinic records were reviewed and data on previous HbA1c levels and medications being taken extracted.

Study site

The study was carried out at the Shai-Osudoku District Hospital (SODH), located in Dodowa, the capital of Shai-Osudoku District of the Greater Accra Region, Ghana. It is a 140-bed capacity hospital with both in- and out-patient facilities and laboratory services. The hospital has an average out-patient attendance of 60 000 annually and runs a chronic diseases care clinic for diabetic and hypertensive patients. It serves the district’s population of about 105,600 (according to Ghana’s 2020 population and housing census). The hospital also serves as the main referral centre for the lower-level health facilities in the surrounding peri-urban and rural communities. The district has four government health centres, 10 Community-based Health Planning and Services (CHPS) compounds, one private hospital, one mission hospital and four private clinics. The inhabitants are mainly subsistence farmers, fishermen, artisanal workers, craftsmen, and petty traders. There are a few civil servants, mainly migrant employees of government ministries, departments, and agencies.

Sample size estimation and sampling

Yamane’s formula [(n=N/1+Ne²) where n = the required sample size, N = the total population and e = the margin of error of 0.05] was used to estimate the required minimum sample size based on the population of 240 visiting the diabetic clinic regularly. Using a 5% margin of error and 5% non-response rate, a sample size of 158 patients was estimated. To increase the statistical power and allow for further sub-group analysis, however, all Type 2 diabetes mellitus patients who consented to be part of the study were sequentially enrolled.

Inclusion/exclusion criteria

All Type 2 diabetes mellitus patients who visit the study facility regularly and consented to be part of the study were eligible for participation except those who were seriously ill.

Data collection tool

A questionnaire was developed specifically for this work, incorporating the Perceived Dietary Adherence Questionnaire (PDAQ) for persons living with Type 2 diabetes mellitus and the Morisky Green Levine medication adherence scale. The questionnaire had four main sections. The first section captured data on the background characteristics of study participants including age, sex, marital status, employment status, level of education, place of residence, household size, sex of head of household, and lifestyle characteristics. The second section captured data on medical history including duration of diabetes, existence of comorbidities and type of antidiabetic medication being taken. The third section collected data on the patient’s perceived dietary adherence. This section consisted of nine questions structured to capture a nutritional therapy guideline that was adapted to the food items in our local settings consisting primarily of carbohydrates, fats and oils, fruits and vegetables. The fourth section captured data on medication adherence based on the Morisky Green Levine medication scale. Information collected included whether patient sometimes forgot to take his or medications, whether in the past two weeks, there was any day patient did not take his or her medications and whether patient took his or her medications the day before the study.

Data collection procedures

Data management and analysis

All completed questionnaires were checked for missing information and errors daily. The data were entered and cleaned using Epi Info Version 7. Statistical analyses were done using STATA version 17. Descriptive statistics were used to present data as frequencies with their corresponding percentages. The proportion of patients with good glycemic control (HbA1c <7%) was determined and its distribution across the independent variables was presented in cross-tabulations.

The weight and height were used to calculate the body mass index (BMI) of patients. Weight in kilograms was divided by the square of the height in meters. BMI was categorized as underweight (<18.5kg/m²), normal weight (18.5-24.9kg/m²), overweight (25.0-29.9kg/m²), and obese (≥30.0kg/m²). Medication adherence was assessed based on four main questions with Yes or No responses. The questions were negatively coded. The lower the total score, the more adherent the patient was to the medication regimen. This was categorized as low (score of 3-4), moderate (score of 1-2) and high (score of 0) adherence.

Spearman rank correlation was used to determine the relationship between glycemic control and continuous variables. The Chi-square test was used to determine association between glycemic control and categorical variables. One-way analysis of variance was done for categorical variables with more than two groups. Fisher’s exact test and Wilcoxon rank-sum test were used as non-parametric measures. Multivariate logistic regression was conducted to determine the strength of association between the individual-level factors and glycemic control at 95% level of confidence.

Ethical consideration

The study protocol, data collection tool and consent form were reviewed and approved by the Ethics Review Committee (GHS-ERC) of the Ghana Health Service with Protocol ID GHS-ERC 064/09/22. Written informed consent was obtained from all study participants before enrollment into the study, the confidentiality of patient identity and information was respected. The study was conducted under Good Clinical Practices and Helsinki Declaration for biomedical research on human subjects.

Demographic and clinical characteristics of patients

A total of 230 Type 2 diabetes mellitus patients visited the clinic during the study period and consented to participate in the study. Out of the 230 patients screened, 227 (98.70%) met the inclusion criteria and were enrolled and included in the analyses. The mean age of the enrolled patients was 60.76 (± 12.12) years with the majority of them being female (77.97%). Most of the patients (62.11%) were married and 77.54% had completed secondary or higher education. One hundred and fourteen (50.22%) of them were employed and 69.16% lived in urban parts of the district. The majority of the patients (71.11%) were either overweight or obese. Almost 60% of the patients have been living with diabetes mellitus for five or more years and over 74% do not exercise at all or exercise only once in a while. Over 85% were on oral hypoglycemic agents alone and more than half of the patients were adhering to medication based on the Morisky Green Levine’s medication adherence score (Table 1).

Level of glycemic control among T2DM patients

Most of the patients (61.23%, 139/227) had poor glycemic control (HbA1c ≥ 7%). The measured HbA1c of patients were positively skewed with a median HbA1c of 7.5% (IQR: 6.5% - 9.4%) (Fig. 1). There was no statistically significant difference in the median HbA1c among males and females using the Wilcoxon rank-sum test (p=0.2402). There was an inverse significant relationship between HbA1c and age of patients from the Spearman rank correlation (r=−0.2205, p <0.001). There was a statistically significant difference comparing HbA1c by the type of anti-diabetic agents (p<0.001) based on one-way analysis of variance (ANOVA).

• Download figure
• Open in new tab

Fig. 1: Graph showing the distribution of glycated hemoglobin of patients

Association between glycemic control and socio-demographic characteristics

A higher proportion of patients (70.75%) aged below 60 years had poor glycemic control compared to those aged 60 years and above (52.89%). A lower proportion (49.54%) of patients who usually eat healthy meals had poor glycemic control compared to 67.09% and 82.05% of those who only sometimes eat healthy meals and those who did not often eat healthy meals respectively. Using the perceived dietary adherence scale, 62.22% of those with low level of adherence had poor glycemic control compared to 57.14% of those with high level of adherence.

There was a statistically significant association between glycemic control and the age of patients (p=0.006), marital status (p=0.032), the type of anti-glycemic agent (p <0.001), eating healthy meals (p <0.001), and having a previously good glycemic level (p <0.001) based on Chi-square test (Table 2).

Multivariable analysis of significant independent factors

The multivariable logistic regression analysis revealed that patients who usually eat healthy meals (aOR: 4.78, 95% CI: 1.65, 13.88), are on oral hypoglycemic agents alone (aOR: 15.71, 95% CI: 1.90, 129.44) and those with previously good glycemic control (aOR: 4.27, 95% CI: 2.16, 8.43) had good glycemic control in this study (Table 3).

Limitations of the study

Glycated hemoglobin levels could be affected by anaemia and hemoglobin variants like sickle cell traits and Glucose-6-phosphate dehydrogenase deficiency. These were not assessed and therefore not controlled for in the analysis. Information on dietary and medication adherence was assessed using adopted tools which could be affected by recall bias. This is a cross-sectional study and thus causal relationship could not be established. These limitations notwithstanding the findings of this study can serve as a guide in the management of T2DM patients.