Type 2 diabetes progression in an adult Ugandan population with new-onset diabetes: an observational prospective study

This observational prospective study was conducted in seven tertiary public and private not-for-profit mission or church-founded hospitals in Central and Southwestern Uganda between October 2019 and August 2021. These hospitals were selected because they serve urban, peri-urban, and rural populations, and have functional outpatient adult diabetes clinics.

A total of 207 adult patients with newly diagnosed type 2 diabetes initiated on oral hypoglycaemic agents (OHA) and enrolled in the main UDIP study that investigated the manifestation of diabetes in an adult Ugandan population with new-onset diabetes were randomly selected to join this study and then followed up for 12 months.

The participants recruited in the study were those aged ≥ 18 years with a recent diagnosis of diabetes (< 3 months since diagnosis), clinically stable (without evidence of metabolic decompensation necessitating hospital admission), and had been initiated on OHA. The diagnosis of diabetes in all of these participants was made by clinicians at the various outpatient clinics based on the World Health Organisation (WHO) guidelines for the diagnosis of diabetes [10]. Pregnant women were excluded from this study.

At baseline, information on relevant demographic (age at diagnosis and gender) and clinical characteristics (type of glucose-lowering treatment initiated at the time of diagnosis) were collected using a pre-tested case report form. This was followed by anthropometric (weight, height, waist circumference or WC, hip circumference, waist: hip circumference ratio or WHR, BMI) and resting blood pressure measurements.

A fasting blood sample was then collected for measurement of blood glucose (FBG), HbA1c, insulin, C-peptide, and lipid profile (total cholesterol or TC, high-density lipoprotein cholesterol or HDLC, low-density lipoprotein cholesterol or LDLC, triglycerides or TGL, and total cholesterol: high-density lipoprotein cholesterol ratio or TC/HDLC ratio). Insulin resistance (Homeostatic model assessment 2 for insulin resistance or HOMA2-IR) and the pancreatic beta-cell function (HOMA2-%B) were calculated using the online homeostatic model assessment-2 (HOMA2) calculator by the Diabetes Trial Unit of the University of Oxford, Oxford UK [11].

The study participants were then followed up every three months for 12 months. To maintain the participants in the study, the trained study nurses made periodic phone calls and also sent out phone messages to the participants reminding them of their scheduled appointments. These study nurses worked at the respective study sites and oversaw patient registration and diabetes education at the outpatient diabetes clinics.

Management of diabetes was determined by the attending clinicians following the WHO treatment guidelines which recommend the use of metformin as the first-line glucose-lowering drug and other oral agents like sulfonylureas or incretin therapies as add-on therapies [12]. At each time point, information on the type and dose of the OHA used, history of any hypoglycaemic episode regardless of severity and severe hyperglycaemia requiring hospital admission, and treatment adjustment (dose increase or addition of a new drug) was obtained. In addition, all participants were subjected to anthropometry and measurement of HbA1c, as a measure of assessing glycaemic control and response to treatment. At the 12-month follow-up time point, a fasting blood sample was drawn from all participants for measurement of FBG, HbA1c, insulin, C-peptide, and lipid profile.

Diabetes progression at the 12-month time point was defined as HbA1c ≥ 8% when on ≥ 2 OHA and/or treatment intensification (initiation of insulin therapy, and/or the dose of > 1 oral drug increased, and/or addition of a second oral agent) [2, 3].

The categorical and continuous variables describing all the study participants at baseline and 12-month time points of follow-up were expressed as proportions and medians with inter-quartile range (IQR), respectively. The HbA1c of the participants at each time point was expressed as a mean ± standard deviation. The changes in HbA1c based on the OHA used at each time point were determined and summarised in the form of a table and graph.

The study received ethical approval from the research ethics committee of the Uganda Virus Research Institute (GC/127/18/05/650) and the Uganda National Council of Science and Technology (HS 2431). All participating study sites offered administrative approval before the initiation of the study. All study participants recruited into the study provided written informed consent. The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki and the Good Clinical Practice guidelines of the International Conference on Harmonisation.

Compared with those at baseline, participants at the 12-month time point of follow-up had a higher median BMI (29 kg/m² [25.1–31.7] vs 27.5 kg/m² [24.3–31.6], p = 0.001). There was no statistically significant difference in most of the clinical and metabolic characteristics between both groups (Supplementary Table 1).

On stratification in two groups based on HbA1c cut-offs and number of OHA being used at baseline, compared with those with HbA1c ≥ 8% and on ≥ 2 OHA, participants with HbA1c < 8% and on < 2 OHA had a higher prevalence of pre-existing hypertension (51.2% vs 25%), a higher median (IQR) systolic blood pressure (139 [127–152] vs 125 [113–133] mmHg), and HOMA2-%B (70.8 [53.1–99.8] vs 38.7 [23.4–65.6]). No differences were noted in the anthropometric and most metabolic characteristics between both groups (Supplementary Table 2).

The HbA1c changes of all study participants, based on the HbA1c cut-offs at baseline and OHA used at each time point are summarised in Supplementary Tables 3 and 4, and Fig. 1.

Both participants on metformin monotherapy and metformin-sulfonylurea combination experienced an initial sharp decline in HbA1c between the baseline and 3-month time points. However, this decline was maintained at the 6-and 9-month time points in participants on combination therapy, with a steady rise in HbA1c later between the 9- and 12-month time points (Fig. 1 and Supplementary Table 3).

Participants with HbA1c < 8% and on < 2 OHA at baseline had a steadily maintained median HbA1c level throughout the follow-up period while those who had HbA1c ≥ 8% and on ≥ 2 OHA experienced a gradual decline in the median HbA1c level (Supplementary Table 4).

Tables 2 and 3 summarise the treatment adjustments at each time point and the independent predictors of short-term diabetes progression, respectively.

Of the 116 participants (56%) who completed the 12-month follow-up study period, diabetes progression was observed in 64 participants, corresponding to a prevalence of 55.2% (95% CI 45.7–64.4). An HbA1c ≥ 8% on ≥ 2 OHA and treatment intensification were noted in 44.8% and 29.3% of participants, respectively.

Of the 34 participants who had treatment intensification at the 12-month time point, 10 participants had a new drug added to their treatment regimen (sulfonylurea and pre-mixed insulin were added to the treatment regimen of nine participants and one participant, respectively).

Treatment adjustment at each time point of follow-up was documented more in participants initiated on metformin monotherapy at baseline (Table 2). Conversely, an HbA1c of ≥ 8% on ≥ 2 OHA was observed more in participants initiated on metformin and sulfonylurea combination at baseline compared with those initiated on metformin monotherapy (n = 34, 65.4% vs n = 18, 34.6% respectively).

On multivariate analysis, only the female gender (adjusted odds ratio or AOR 3.2, 95% CI 1.1–9.2, p = 0.03) was noted to independently predict diabetes progression at the 12-month time point of follow-up (Table 3).

Supplementary Table 5 summarises the demographic, clinical, anthropometric, and metabolic characteristics of participants who progressed and those who did not at the 12-month time point of follow-up.

Compared with the non-progressors, participants classified as progressors had a higher HbA1c level (69 mmol/mol [55–85] vs 47 mmol/mol [40–59]), HOMA2-IR (1.7 [1.2–2.5] vs 1.2 [0.9–2.0]), and a lower HOMA2-%B (39.8 [26.1–59.9] vs 90.3 [46.5–120.7]). No marked differences were noted in BMI, WHR, lipid profile, fasting insulin, and C-peptide concentrations between both groups.

On comparing the baseline characteristics of participants who completed the follow-up period and those who were lost to follow-up, the latter group had more females (62.5%) and had a lower median (IQR) HbA1c (9.3 [8.2–11.9] vs 11.4 [8.8–12.9] %, respectively) (Supplementary Table 6).

To our knowledge, this is the first study to prospectively follow up a cohort of unselected adult African patients with new-onset diabetes to understand, in a real-world setting, the frequency and predictors of short-term progression of type 2 diabetes. Despite this strength, our study had some limitations. Its short follow-up period may explain why we did not observe an association between diabetes progression and commonly reported phenotypic characteristics like HbA1c, BMI, WC, and some lipid profile parameters. We also had a very high rate of loss to follow up through the study. This is because the study was mainly performed during the COVID-19 pandemic which significantly affected retention in clinical care due to the frequent and prolonged lockdown periods and challenges in accessing public transport for most participants. Additional factors associated with diabetes progression like lifestyle or environmental factors and specific genetic variants were not assessed in this study.