Diabetes prevalence and risk factors, underestimated without oral glucose tolerance test, in rural Gombe-Matadi Adults, Democratic Republic of Congo, 2019

Study design

This was a population-based cross-sectional survey targeting households.

Settings

The study was conducted in May 2019 in the Gombe-Matadi Rural HZ which has 15 health areas (HA) (Fig. 1). The study was carried in Gombe-Matadi (26 villages), Yanda (26 villages), and Ntimansi (25 villages) HA. Ntimansi is a strictly rural environment, Gombe Matadi a commercial place serving as transit for travelers from Kinshasa, and Yanda (strictly rural) a religious environment due to the presence of the city of Nkamba (the international headquarters of the Kimbanguiste church) and also influenced by international visitors.

The map of Gombe Matadi Health Zone (Source: The Health Zone of Gombe Matadi, Software QGIS 3.4 2018, https://www.qgis.org).

Study population

The eligibility criteria included individuals with 19 years of age or above¹⁹ (the risk of diabetes in the youngest age is minimal), both sexes, and have signed the consent form. Pregnant women (defined by amenorrhea of more than 2 months for women younger than 46 years, outside exclusive breastfeeding), individuals with severe psychiatric conditions, non-Congolese and non-resident individuals were excluded.

Sampling

The methodology followed is the one proposed by the World Health Organization (WHO)¹⁹ and the American Diabetes Association (ADA)²⁰ which allows for standardization of methods and comparison with other countries.

A sample of 1600 individuals was obtained by the following formula: n = (Z²pq/d²)  * g * 1/1 − f. Z_95% coefficient was equal to 1.96. The prevalence of diabetes, p = 0.05 (q = 1 − p), corresponded to that found in Kisantu, in DRC¹⁸. The desired degree of precision d was equivalent to 0.02, and g represented the correction coefficient for the cluster effect estimated at 2²¹. We used a proportional sample (Table 1). The fraction of non-responses f was estimated at 10% .

To select the survey participants, we proceeded in five stages^22,23. In the first stage, three HA selected by simple random sampling were considered as clusters. In the second stage, villages within clusters were selected according to demographic weight (> 200 inhabitants) and distance (< 4 km from the study site). In the third stage, a systematic drawing of the inhabited plots in each selected village after having listed and numbered all the inhabited plots was performed. The sampling step for each HA village was the number of inhabited plots (N) divided by the proportional sample of the village (n). All the plots identified were chosen. In the fourth degree, after having listed all households, a simple random selection of households was carried out in each plot. In the fifth degree, an eligible subject was drawn by simple random selection from each household.

Operational definitions

Diabetes was defined according to the ADA 2018²⁰ and the WHO/International Diabetes Federation 2006²⁴, on the basis of, for a new case at least two altered glycaemia (fasting plasma glucose (FPG) ≥ 126 mg/dLL) and/or oral glucose tolerance test (OGTT) ≥ 200 mg/dL. For a known case, a diabetic notebook or an anti-diabetic treatment was considered. The impaired fasting glycaemia (IFG) was defined by an FPG between 100–125 mg/dL on day one or two and an OGTT < 200 mg/dL. An impaired glucose tolerance (IGT) was defined by an FPG < 126 mg/dL on day one and an OGTT between 140–199 mg/dL.

Risk factors were defined as this. General obesity was defined as a body mass index (BMI) (weight (Kg)/ height (m²)) ≥ 30 kg/m²²⁵ and abdominal obesity by a waist circumference measurement ≥ 94 cm and ≥ 80 cm, respectively for men and women²⁶. Hypertension was confirmed on the basis of two consecutive measurements of systolic blood pressure (SBP) value ≥ 140 mmHg and diastolic blood pressure (DBP) ≥ 90 mmHg^21,27, or self-reported. Age was categorized into < 40 years old and ≥ 40 years old, education in low school level (none and primary) and high school level (the other levels), and the usual mode of mobility in a group with physical exercises (foot and bike) and non-physical exercises (motorbike, car). General obesity as well as abdominal obesity in obese and non-obese and the profession in unemployed (student, unemployed, housewife, without profession, disabled, retired) and employees (employed in the state, employed in the private sector, self-employed, farmer, poultry farmer, volunteer, domestic) were also considered. Other binary variables were categorized in yes/no, family history of diabetes, hypertension, macrosomia, alcohol consumption, and smoking.

Data collection

Authorizations for the investigation were obtained from the political and administrative authorities and a census of inhabited compounds was performed. The investigation lasted 39 days. Ten field workers recruited among nurses followed a 3-day training course on diabetes, hypertension, the questionnaire, and the census of inhabited compounds. After evaluation, six field workers were selected. A pre-test was organized for all team in an HA of Gombe-Matadi not selected for the survey, on 25 individuals.

The census of inhabited compounds identified during 3 days all the plots by proceeding area by area. In each area the whole team worked together before going to another area. The coordinator and three supervisors came from the Kinshasa School of Public Health.

Fixed study sites were defined to avoid mobilizing too much human and material resources and to improve data quality^20,23. We chose 3 sites (health center, school or church) by HA. Registration of participants, and first FPG were performed at home after obtaining their consent. An appointment card was provided to the participant for the OGTT, to answer the questionnaire, to take measurements and do FPG in case of not fasting. In the absence of the chosen individual, the field worker returned the same or the next day. No refusal was noted. In the case the elected individual does not meet the criteria, another participant will be drawn randomly. One plot without eligible individuals was skipped and the next was taken. The coding of names was carried out in the evening by the team, by assigning each name a specific code and copying it into their questionnaire.

At the same time, the population was sensitized through the churches, district leaders and community health workers. The team was divided into 2 groups of 3 field workers and 1 supervisor to work in the first 2 sites before ending up all together in the last site. The investigation did not disturb usual work within the site. A snack was provided to avoid the impatience of the hungry participants.

On the first day an FPG was carried out at home, the following day a second FPG was assessed as well as the OGTT, and the interview and measurements were carried out on the study site. Those who did not have their FPG at home (not fasting) had it on the first day at the site and the OGTT the next day. The community health workers recovered the absents from their homes.

After the calibration of devices every morning, the glycemia was assessed using a Godefree glucometer (witch references on plasma). The OGTT was performed every day except for 25 people, two individuals with a glycemia ≥ 200 mg/dL on day one, two individuals with a glycemia ≥ 200 mg/dL on day two, and 21 who were previously diagnosed cases of diabetes. A capillary glycemia was taken and the subject orally ingested 75 g of anhydrous glucose in 250 mL of water for 5 min. After 2 h another capillary glycemia was taken. In total, each participant underwent 3 glycemia tests.

Statistical analyses

Data quality checks were performed daily by the supervisors’ team. Data management and analyses were performed with IBM’s SPSS 23.0 statistical software. Descriptive analyses provided the measures of frequency, central tendency and dispersion. The chi-square has compared the proportions. The differences between the non-respondents were compared to the respondents (age, sex, first-day glycemia). Age-standardized prevalence was calculated using the standard population of Doll et al.²⁸. The inferential statistics were based on 95% CI and Mann Whitney’s test which compared the medians of glycaemia between respondents and non-respondents. Multivariable logistic regression was used to explore the association between diabetes and its risk factors (age < 40 and ≥ 40 years, education, usual physical activity, alcohol consumption, smoking, general obesity, family history of diabetes, occupation, hypertension, and sex). A test was considered statistically significant when p was < 0.05.

Ethic statement

The study was approved by the National Committee for Ethics of DRC, under the number 104/CNES/BN/PMMF/2018 of 23/01/2019. All participants gave their informed consent and the data were confidentially kept. Patients were treated and referred to health centers for follow up.

We confirm that all methods were carried out in accordance with relevant guidelines and regulations.