We conducted a cross-sectional study among patients with diabetes attending the outpatient diabetic clinic at Lira Regional Referral Hospital from July to December 2025. LRRH is located in Adyel Division, Lira City, approximately one kilometer from Lira town center, serving a population of about two million people primarily from the Lango sub-region. The diabetic clinic manages an average of 150 patients per month, providing essential diabetes care and education.

We included diabetic patients aged 18 to 65 years attending the clinic during the study period. Diabetes was defined as fasting capillary glucose ≥ 7.0 mmol/L or current treatment for diabetes mellitus. We excluded individuals with acute febrile illnesses in the preceding 48 hours, those with reported or documented electrolyte abnormalities within the preceding month (including potassium < 3.5 mmol/L or >5.5 mmol/L, magnesium < 0.7 mmol/L or >1.1 mmol/L, or calcium < 2.1 mmol/L or >2.8 mmol/L), those with other endocrine disorders (e.g., thyroid hormone disorders), and those actively taking medications known to affect cardiac rhythm or QTc interval (antiarrhythmic drugs classes I and III, antidepressants, or digitalis) in the prior 24 hours. We also excluded individuals with self-reported or documented history of cardiac disease (heart failure, ischemic heart disease), renal disease, or hepatic disease.

Sample size was calculated using the Kish-Leslie formula for cross-sectional studies [13]. Using a prevalence of abnormally long QTc of 29.8% from a previous Ugandan study [12], a 5% margin of error, and a 95% confidence interval, we calculated a minimum sample of 322 participants. After accounting for a 10% non-response rate, we targeted 358 participants. Consecutive sampling was employed until the sample size was attained.

We collected socio-demographic and clinical characteristics using a structured interviewer-administered questionnaire, including data on alcohol use, smoking, physical activity, and diabetes treatment history. Diabetes type was determined from the patients’ medical records.

2.4.1. Anthropometric Measurements

Weight was measured to the nearest 0.1 kg using a calibrated Seca weighing scale (Seca 762, GmbH & Co. KG, Hamburg, Germany) with participants wearing light clothing and no shoes. Height was measured to the nearest 0.1 cm using a stadiometer with participants standing upright. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared and categorized according to WHO standards: underweight (<18.5), normal (18.5 - 24.9), overweight (25.0 - 29.9), and obese (≥30.0). Waist circumference was measured at the umbilical level using inelastic tape to the nearest 0.1 cm at the end of normal expiration.

2.4.2. Blood Pressure Measurement

Blood pressure was recorded in a sitting position using an automatic sphygmomanometer (Omron HEM 705 LP, Omron Healthcare, Inc., Bannockburn, IL, USA). Mean arterial pressure (MAP) was calculated as diastolic blood pressure plus one-third of pulse pressure.

2.4.3. Laboratory Measurements

Blood was collected for glycosylated hemoglobin (HbA1c) measurement (Cobas Integra 400, Roche Diagnostics, Basel, Switzerland). Random blood sugar measurements were obtained using a Freestyle Glucometer (Abbott Diabetes Care Inc., Maidenhead, UK).

2.4.4. Physical Activity Assessment

Physical activity was assessed by asking participants about total weekly time engaged in physical activities (including walking, running, swimming, and ball games). Adequate physical activity was defined as ≥150 minutes per week according to WHO recommendations [14].

We performed 12-lead ECG recordings on all study participants using a portable ECG machine (Edan Instruments, Inc., Hessen, Germany). Participants rested in a supine position for 5 minutes before recordings. The QT interval was measured from the beginning of the QRS complex to the end of the T wave, where it crosses the isoelectric line. When a U wave was present, we measured the QT interval to the bottom of the angle between the T and U waves. QT was taken as the mean of QT from five consecutive cycles in lead V5.

We calculated the QTc interval using Bazett’s formula (QTc = QT/RR^1/2) for heart rates between 60 and 100 beats/minute and Fredericia’s formula (QTc = QT/RR^1/3) for heart rates < 60 or >100 beats/minute, as Bazett’s formula tends to over-correct at high heart rates and under-correct at low heart rates [15]. We classified the QTc interval as normal (<430 ms in males, <450 ms in females), borderline (430 - 450 ms in males, 450 - 470 ms in females), or prolonged (>450 ms in males, >470 ms in females) according to established criteria [15]. For analysis, borderline and prolonged categories were combined as “abnormally long QTc” to increase statistical power, consistent with previous studies [12][16].

Data were double-entered into Excel and exported to STATA version 17 (Stata Corp, College Station, Texas, USA) for cleaning and analysis. Continuous variables were expressed as mean ± standard deviation or median with interquartile range, and categorical variables as frequencies and percentages.

The prevalence of abnormally long QTc interval was calculated as the proportion of participants meeting sex-specific criteria for borderline or prolonged QTc (i.e., QTc ≥ 430 ms in males or ≥450 ms in females), with 95% confidence intervals using the exact binomial method.

For factors associated with abnormally long QTc, we first compared characteristics between participants with normal and abnormally long QTc using chi-square tests for categorical variables and t-tests or Wilcoxon rank-sum tests for continuous variables. Modified Poisson regression with robust variance was used to estimate prevalence ratios (PR) and 95% confidence intervals [17]. Variables with p < 0.2 in bivariate analysis were entered into multivariable models using stepwise backward elimination with entry and removal probabilities of 0.05 and 0.1, respectively. Given the clinical overlap between mean arterial pressure, hypertension history, and antihypertensive regimen, we assessed collinearity using variance inflation factors (VIF) and found no evidence of multicollinearity (VIF < 2.0 for all variables). We retained MAP as a continuous measure of blood pressure control and included hypertension history and antihypertensive regimen as separate variables to capture both disease status and treatment effects. Age and diabetes duration were forced into the model a priori based on clinical relevance. Statistical significance was set at p < 0.05.

Ethical approval was obtained from the Lira University Research Ethics Committee (approval number: LUREC-2025-404). Administrative permission was granted by the hospital director of LRRH and the person in charge of the diabetic clinic. This study adhered to the Declaration of Helsinki. All participants provided written informed consent before enrollment. Participants who could not write provided consent with a thumbprint. Each participant received an incentive of 10,000 Ugandan shillings for transport reimbursement, and those requiring further evaluation were referred to appropriate care.

Of 480 individuals attending the diabetic clinic during the study period, 120 were not recruited (94 did not meet inclusion criteria [age > 65 years, n = 48; age < 18 years, n = 12; acute febrile illness, n = 15; active medications affecting QTc, n = 19], and 26 declined participation). Of the 360 recruited, 10 (2.8%) did not return their questionnaires. Of the 350 questionnaires returned, 12 (3.4%) were excluded due to incomplete or inappropriate completion, leaving 338 participants for analysis (response rate: 93.9%).

Baseline characteristics are summarized in Table 1. The mean age was 52.4 years (SD ± 10.2), and 63.0% (213/338) were female. The majority had type 2 diabetes (315/338, 93.2%), with 6.8% (23/338) having type 1 diabetes. Mean HbA1c was 9.2% (SD ± 2.8), and mean duration of diabetes was 6.2 years (SD ± 5.7). The majority of participants (70.7%, 239/338) had a history of hypertension.

Table 1. Baseline characteristics of participants by QTc status.

Note: *BMI: body mass index; BP: blood pressure; MAP: mean arterial pressure; RBS: random blood sugar; PA: physical activity; Data presented as n (%) unless otherwise specified; p < 0.05; *Abnormally long QTc includes borderline and prolonged categories combined.

Out of the 338 participants analyzed, 76 had abnormally long QTc intervals, giving a prevalence of 22.5% (95% CI: 18.3% - 27.3%) as shown in Figure 1Figure 1.

Figure 1. Prevalence of abnormally long QTc interval among study participants.

In multivariable analysis, the factors independently associated with a higher prevalence of abnormally long QTc were elevated MAP (aPR = 1.16, 95% CI: 1.05 - 1.28, p = 0.002) and use of furosemide-containing antihypertensive regimens (calcium channel blocker + loop diuretic) (aPR = 1.32, 95% CI: 1.11 - 1.57, p = 0.002) (Table 2). Adequate physical activity (≥150 minutes/week) remained protective (aPR = 0.85, 95% CI: 0.77 - 0.94, p = 0.002). Compared to injectable-only anti-diabetic regimens, use of oral agents alone (aPR = 0.62, 95% CI: 0.55 - 0.70, p < 0.001) and combined oral and injectable regimens (aPR = 0.57, 95% CI: 0.50 - 0.66, p < 0.001) were associated with lower prevalence of abnormally long QTc.

Table 2. Multivariate analysis of factors associated with abnormally long QTc interval.

Note: aPR: adjusted prevalence ratio; CI: confidence interval; ref: reference category.

The prevalence of abnormally long QTc interval (22.5%, 95% CI: 18.3% - 27.3%) observed in our study is comparable to findings from other African settings. A study from Tanzania reported a prevalence of 32% among type 2 diabetic patients at Kilimanjaro Christian Medical Center [18], while research from southwestern Uganda found a prevalence of 29.8% [12]. Our estimate is slightly lower but within the range reported from international studies (15% - 67%) [9]-[11]. A recent systematic review of QTc prolongation in diabetes mellitus reported pooled prevalence estimates ranging from 17% - 35% across various populations, with our findings falling within this range [5].

The prevalence we observed is substantially higher than that reported in the general population in Uganda (11.7%) [19], reinforcing that diabetes itself confers increased risk for QTc prolongation. This phenomenon is hypothesized to result from persistent hyperglycemia, which increases intracellular calcium, promotes free radical production, alters cardiac sympathovagal balance, and diminishes nitric oxide production [20]. Low nitric oxide levels impair primary active transport mechanisms in cardiomyocytes, prolonging myocardial repolarization [21].

Our prevalence is lower than that reported from India (42% among patients with cardiovascular autonomic neuropathy) [22], likely because our study population represented the full spectrum of diabetic patients rather than those with established complications. This suggests that our findings may be generalizable to routine diabetic care populations in similar resource-limited settings.

4.2.1. Mean Arterial Pressure

The independent association between elevated MAP and abnormally long QTc (aPR = 1.16, p = 0.002) aligns with previous studies demonstrating that blood pressure influences cardiac repolarization [23]-[26]. Chronic pressure overload from hypertension leads to left ventricular hypertrophy, which is associated with downregulation of potassium channels responsible for phase 3 repolarization [27]. Additionally, hypertension-induced myocardial fibrosis disrupts electrical coupling between myocytes, creating heterogeneous repolarization that prolongs the QT interval [28]. In diabetic patients, this effect may be amplified by concomitant autonomic neuropathy and microvascular damage.

4.2.2. Antihypertensive Medications

The finding that furosemide-containing regimens (calcium channel blocker + loop diuretic) were associated with a higher prevalence of abnormally long QTc (aPR = 1.32, p = 0.002) is clinically significant. Loop diuretics like furosemide can cause excessive urinary loss of potassium and magnesium, electrolytes critical for normal cardiac repolarization [29]. Hypokalemia reduces the driving force for potassium efflux during phase 3 of the action potential, while hypomagnesemia impairs sodium-potassium ATPase function and exacerbates potassium wasting [30][31]. However, it is important to note that we did not measure serum electrolytes concurrently with ECG recordings. The observed association may be partially explained by the underlying disease severity or other unmeasured confounders, and our findings should be interpreted as hypothesis-generating rather than causal. Nevertheless, our findings suggest that when furosemide use is unavoidable in diabetic patients, regular monitoring of serum electrolytes with prompt replacement may be prudent.

4.2.3. Antidiabetic Medications

Patients requiring injectable antidiabetic agents (primarily insulin) had a significantly higher prevalence of abnormally long QTc compared to those on oral agents alone (aPR = 0.62 for oral-only vs. injectable-only, p < 0.001). The association between insulin-only therapy and QTc prolongation likely reflects that insulin-treated patients have more advanced disease, longer diabetes duration, poorer glycemic control, and potentially greater end-organ damage, all of which are independent risk factors for QTc prolongation [32]. While insulin-induced hypoglycemia can acutely prolong the QT interval through catecholamine-mediated hypokalemia [33], we cannot disentangle the medication effect from the underlying disease severity in this cross-sectional design. Therefore, the observed association should be interpreted as a marker of higher-risk patients rather than a direct pharmacological effect of insulin itself. Patients on combination therapy (oral + injectable) also had a lower prevalence than injectable-only patients (aPR = 0.57), possibly because combination therapy allows lower insulin doses, reducing hypoglycemia risk while maintaining glycemic control.

4.2.4. Physical Activity

Adequate physical activity (≥150 minutes/week) was independently protective against QTc prolongation (aPR = 0.85, p = 0.002). Regular exercise improves autonomic balance by increasing vagal tone and reducing sympathetic drive, stabilizing myocardial repolarization [34]. Physical activity also improves glycemic control, reduces oxidative stress, and promotes favorable cardiac remodeling, all of which may preserve normal ion channel function [35].

Our findings have several practical implications for diabetes care in resource-limited settings. First, the high prevalence of abnormally long QTc (22.5%) supports routine ECG screening for diabetic patients, particularly those with multiple risk factors. ECG is relatively low-cost and feasible in district-level hospitals in Uganda. Second, stringent blood pressure control should be prioritized, as elevated MAP was a strong independent correlate of QTc prolongation. Third, when diuretic therapy is necessary, potassium-sparing agents may be preferred over loop diuretics, or close electrolyte monitoring should be implemented. Fourth, physical activity should be promoted not only for glycemic control but specifically for cardiac electrical stability. Finally, patients requiring insulin therapy warrant closer ECG monitoring, with careful attention to avoiding hypoglycemic episodes.

Several limitations should be acknowledged. The cross-sectional design precludes causal inference; we can identify associations but not determine directionality. Medication use, diabetes duration, and hypoglycemic history were based on patient records and self-report, introducing potential information bias. QTc interval was measured at a single time point, not accounting for day-to-day variability due to fluctuations in glucose, electrolytes, or autonomic function. We did not measure serum electrolytes concurrently with ECG recordings, which limits our ability to definitively attribute the furosemide-QTc association to electrolyte disturbances. We excluded patients with known cardiac, renal, or hepatic disease, which may have underestimated the true prevalence in the broader diabetic population. Diabetes type was not included as a covariate in the final model due to limited power for subgroup analysis, though we have acknowledged this as an unmeasured confounder. Finally, the single-center design may limit generalizability to other regions.