Identification of predictors of glucose control in a cohort of adult patients with diabetes mellitus at Kalafong Hospital

Abstract

Background and objectives of the study: Although it is known that good glycaemic control improves microvascular outcomes in diabetic patients, no local study has yet been undertaken to investigate the potential factors that influence poor or good blood glucose control. This research focused on the evaluation of blood glucose control as assessed by glycosylated haemoglobin (HbA1c) levels in diabetic patients. In addition, certain determinants which contributed toward poor control at Kalafong Hospital were studied in a cohort of adults with diabetes mellitus for the year 2008. The aim of studying these determinants was to identify patients with a high risk of disease morbidity and barriers that prevent these patients from meeting their goals of improved health outcomes. The specific objectives were to estimate HbA1c control of patients seen at the diabetic clinic at Kalafong Hospital Pretoria in 2008 and to assess any existing association between patient demographic characteristics and diabetes characteristics with HbA1c. Methods: The study was a retrospective cohort study. All diabetic patients aged 18 years and above, who had been registered in the 2008 dataset and who had come for at least one visit to the diabetic clinic and had at least one HbA1C measurement, were included in the study. Patients who did not meet the above criteria were excluded from the study. A total of 942 patients seen in 2008 were selected, 801 patients met these inclusion criteria. The outcome variable HbA1c was obtained by computing the mean of the two HbA1c values collected for each participant for the year 2008, and used as a continuous dependent variable in multivariate linear regression. For descriptive purposes, HbA1c values were categorised into good control (<7%), poor control (> or = 7&< or =10%) and very poor control (>10%). Data analysis was performed using Stata version 10. Statistical significance was established at a threshold of 95% (p < 0.05). Results: More than half of participants in the study were females (60.8%/39.2%). The mean age of participants in the study was 56 years (sd 14.1). With regard to race, the proportion of blacks was more than three quarters of the sample (93.1%/2.4%/2.4%). Our results showed that HbA1c level decreased with increasing age, (p = 0.016). These results also showed that for every 1 mmol/l increase in total cholesterol, there was a 0.178% increase in HbA1c, (p = 0.019; 95% confidence interval (CI): 0.030 - 0.327), suggesting that higher cholesterol was associated with poorer HbA1c control. In addition, for every 1 mmol/l increase in capillary glucose, the HbA1c increased by 0.276%, (p = 0.000; CI: 0.230 - 0.322) while for every one unit increase in BMI, the HbA1c reduced by 0.032%, (p = 0.017; CI: -0.057 to -0.006). Conclusion: These results suggest that patients with higher total cholesterol and patients with higher capillary glucose level are more likely to exhibit poorer HbA1c control, whereas, older patients and patients with a higher BMI are more likely to have better HbA1c control.