Materials and methods

Study design and setting

This was a cross-sectional study conducted at the diabetic clinic of Mbarara Regional Referral Hospital (MRRH), rural Southwestern Uganda, from July to December 2023. MRRH serves the whole of southwestern Uganda but also receives patients from the central region of Uganda and neighboring East African countries, especially Rwanda, Tanzania, and the Democratic Republic of Congo. This facility is the largest outpatient diabetic clinic in southwestern Uganda, and it has a patient census of approximately 1,500 patients annually. The facility also enrolls about four new cases of diabetes per week.

Study participants and sampling

The study participants were selected following this inclusion criteria: aged above 18 years: diagnosed with diabetes at least in the past one year and above; not having any other systemic diseases; not having any history of diabetic complications, such as neuropathy, nephropathy, retinopathy, etc.; not using drugs, such as phenytoin, nifedipine, etc.; not undergoning any periodontal treatment since last one year; and willingness to participate in the study.

The study excluded participants who had diabetic emergencies, such as diabetic ketoacidosis or hypoglycemia. The study had a sample size of 336 respondents. This was determined using the Kish & Leslie formula of 1965, commonly applied for prevalence and cross-sectional studies. We used a prevalence of 32.3%, which was obtained from the MRRH clinical report of 2022. We used a 95% confidence level, 5% margin of error, and 10% non-response rate. A consecutive sampling method was used to select the participants at the MRRH clinic. Every respondent who met the inclusion criteria was enrolled until the sample size was obtained.

All respondents underwent clinical examination to assess for periodontal disease, and an interviewer-administered questionnaire was used to obtain demographic data and assess for associated factors.

Each day of data collection, the research team checked the diabetes register to identify patients who were present that day and had never been recruited in this study. Eligible participants were given information regarding the nature of the study, and a written informed consent was obtained in a local language (Runyankore) from all enrolled participants by a trained research assistant.

Ethical consideration

Ethical clearance was sought from Mbarara University of Science and Technology (MUST) - Research Ethics Committee (MUST-2023-829), and MRRH administrative permission was sought from the Hospital Director. Informed consent was obtained from all participants by signing or inserting a thumbprint on the informed consent form before they were recruited into the study. Participants enrolled were free to voluntarily withdraw from the study at any time. Confidentiality was ensured through unique identifiers rather than names, and participants were examined and interviewed in a separate room with only the research team. Collected research data were kept under lock and key, only accessible by the study team with authorization from the study Principal Investigator.

Data collection tools

An interviewer-administered questionnaire, as shown in Appendix 1, was developed after a review of relevant literature from previous studies. The questionnaire consisted of three sections: A, B, and C. Closed-ended questions were used for all sections. Section A consisted of items about social demographic characteristics: age, sex, level of education, income level, and marital status. Section B consisted of items measuring the prevalence of periodontal disease using the World Health Organization (WHO) Oral Health Assessment form for adults [12]. Individual clinical assessments were carried out by trained dental surgeons who were this study's research assistants. Section C of the questionnaire consisted of items measuring factors associated with periodontal disease. These included diabetes type, presence of comorbidities, glycemic control, frequency of tooth brushing, frequency of dental visits, smoking, and alcohol use.

A two-day training for research assistants was conducted, followed by a pilot study on the research tools to establish their reliability in capturing the required information before the actual data collection. The pilot study was conducted among 20 volunteers who met this study's inclusion criteria at a different diabetic clinic in another region. Two dental surgeons (denoted with alphabets A and B) from the research team examined the volunteer diabetic patients in the sequence below, and the scores were recorded.

A examined the first batch of 10 diabetic patients, B examined the second batch of 10 diabetic patients, A re-examined the batch of diabetic patients that B had examined earlier, and, finally, B re-examined the diabetic patients that A had examined at first. The scores were recorded independently. Then, the results were cross-matched among the dental surgeons, and from all the batches, the reason for forming batches among the diabetic patients was not to overburden them.

The examiner consistency was assessed using the kappa statistic, and this resulted in 0.924 with p-value <0.001, indicating that there was a statistically significant and almost perfect level of agreement between examiners A and B in their caries assessments. This suggested a very high degree of consistency between the two examiners.

Throughout the whole procedure, clinical assessments were conducted using a mouth mirror and a calibrated periodontal probe to measure pocket depth, which enabled us to determine the Community Periodontal Index (CPI). The CPI values were recorded as follows: score 0 - normal, score 1 - bleeding on probing, score 2 - presence of calculus with plaques seen on probing, score 3 - pockets 4-5 mm (mild periodontitis), and score 4 - pockets ≥ 6 mm. Probing was done on the buccal, palatal, mesial, and distal surfaces of each tooth. We excluded the third molars in each quadrant.

Being diabetic was measured as fasting blood sugar of >7.1 mmol/L, random of >11.1 mmol/L, and hemoglobin A1c (HbA1c) of >6.5 mmol/L. Glycemic control was measured as controlled blood sugars (HbA1c: 4-6.5 mmol/L) and uncontrolled blood sugars (HbA1c: >6.5 mmol/L).

Data analysis

Data were entered into REDCap software (version 6.5; Vanderbilt University, Nashville, TN), cleaned, and exported to STATA (version 17; StataCorp LLC, College Station, TX) for analysis. The prevalence of periodontal disease was categorized according to participant CPI value scores. Participants with CPI values of 0 were recorded as normal, participants with CPI values of 1, 2, and 3 were recorded as having gingivitis, participants with CPI values of 4-5 were recorded to have mild periodontitis, and participants with CPI values of 6 and above were recorded to have severe periodontitis.

Bivariate and multivariate logistic regression were done to assess the association between the independent factors and periodontal disease. Variables that were significant p < 0.2 at 95% confidence interval at bivariate analysis were included in the multivariate model. Further variables that were not significant from the bivariate analysis but had biological and social plausibility in relation to the outcome were included in the model. Significance was set at p-value <= 0.05. Adjusted odds ratios and 95% confidence intervals were reported.