Optimal Threshold of the Finnish Diabetes Risk Score (Findrisc) for Screening at-Risk Adults in an African Population in Southern Benin

This study aims to determine the performance of the FINDRISC tool for screening adults at risk for type2 diabetes in an African population in southern Benin. This retrospective study included 536 subjects aged 25 to 65 years. The FINDRISC questionnaire score was calculated using seven components: age, body mass index, waist circumference, regular practice of 30 minutes of physical activity, regular consumption of five portions of fruits and vegetables, the family history of diabetes and the presence of high blood pressure. The performance of the tool was determined from the FINDRISC score and fasting blood glucose The "Receiver Operator Characteristic" (ROC) Curve of the FINDRISC score was used to define the optimal cut-off value of the tool for. The FINDRISC threshold-optimal value for detecting T2D risk in southern Benin was 8.5 with 77% sensitivity, 89% specificity, 45% positive predictive value and 71% negative predictive value. The area under the ROC curve was 0.86 (95% CI: 0.81-0.90). This study revealed that the optimal threshold value for detecting the medium or high risk of T2D observed in Benin is lower than Finland's one. Further studies including larger and more representative samples need to be carried out to confirm this finding.


Introduction
The prevalence of type2 diabetes (T2D) is increasing worldwide to reach nearly epidemic proportions. As populations age and urbanization intensifies, the threat of T2D will continue to grow [1]. Diabetes is expected to reach 642 million people worldwide by 2040 [2]. In 2015, the International Diabetes Federation (IDF) has estimated that 9.5 to 29.3 million people live with diabetes in the African Region of which, ¾ are undiagnosed [2]. According to IDF, 80% of people with diabetes live in lowand middle-income countries. T2D represents 90% of cases and is associated with several complications, leading to morbidity, disability and premature mortality [2,3]. Diabetes also carries a heavy economic burden for patients, households and health care systems [4,5]. Prevention remains the best cost effective intervention to counteract this disease. Many studies have shown that interventions at pre-diabetic stage based on lifestyle changes [6][7][8][9][10] or medications [11] can prevent, or at least delay the progression of the disease. This means that early identification of people at high risk of T2D is necessary to enable faster implementation of preventive measures in order to reduce risk [12,13]. The prevention of the disease is very cost-effective [14,15]. Fasting blood glucose, 2h oral glucose tolerance test (OGTT) and glycated hemoglobin levels (HbA1c) are the recommended methods for diabetes screening [16]. However, those methods are expensive for mass screening. Several risk scores are available and can help to identify subjects at risk for type2 diabetes. Before being used in practice, performance and validity of the tools should be evaluated in the population of interest [17]. The FINDRISC for the Screening of Subjects at Risk for Type2 Diabetes is a powerful and valid tool in several Northern countries [18][19][20][21][22].
However, this tool has not been much validated in black populations. Compared to other tools that use some metabolic syndrom components (triglycerides levels, OGTT) to predict T2D in adults [23,24], FINDRISC score is cost-effective because the questionnaire easy to completed without need of blood test. Also, score calculation and its interpretation is easy. In our study, the FINDRISC score was made by comparing this score to 64 Optimal Threshold of the Finnish Diabetes Risk Score (Findrisc) for Screening at-Risk Adults in an African Population in Southern Benin fasting blood glucose values (mmol/l). The objective of the study was to examine the performance of the FINDRISC tool in selecting adults at risk for T2D in an African population.

Setting
The study took place in the economic capital of Benin Republic that is Cotonou, and in urban and rural areas of Ouidah city. These areas are located in the South of the country bordering the Atlantic Ocean. Cotonou city has 13 districts that include 140 neighborhood covering an area of 79 km 2 [25]. The population of Cotonou is estimated at 679012 inhabitants in 2012 [26]. The township of Ouidah is located at 50 km from Cotonou with an estimated population of 162034 inhabitants in 2007, situed on an area of 363 km 2 [26]. The urban area are divided into 70 neighborhood, while the rural area are subdivided into 38 villages.

Study Design
This was a retrospective study that used data from a previous study performed on cardio-metabolic risk factors, lifestyle and diet in Benin Republic from August 2005 to February 2006. A national survey [27] of the same populations as in this study (sample size = 5126) yielded an average fasting glucose level of 5.17 mmol / l ± 3.27. This average blood glucose level in 2015 compared to the average blood glucose level in this study in 2006, does not show a significant difference between the two means with p-value = 0, 9974. We assume that there is probably no significant difference between the other components of the Findrisc score between 2006 and 2015.

Inclusion and Exclusion Criteria
The study was set for adults, aged 25 to 65 years at the time of the study. Participant were included in condition there are in good health i.e. no cardiometabolic disorders (hypertension, heart disease, diabetes or other metabolic disease), born of black Beninese parents and grandparents and live in Ouidah or Cotonou cities permanently for at less 6 months at the moment of the study. We excluded pregnant and lactating women, people with a critical health condition (bedridden) and those with physical or mental disability.

Sample Size
Minimum sample size By fixing the error of the first species α at 5% and that of the second species β at 20%, the minimum area under the ROC1 curve at 0.60, the negative case ratio (low FINDRISC) on positive cases (high FINDRISC) at 4 and the area under the ROC curve under the null hypothesis at 0.5, a population of 480 subjects was sufficient to use the area under the ROC curve to assess the performance of the FINDRISC score using the MedCalc software.
For the study 541subjects were randomly selected from Cotonou, the largest city (n = 200), Ouidah, a small town (n = 171) and surrounding rural areas of Ouidah (n = 170).
Out of the 541 subjects, five (05) had high blood glucose levels and were excluded, then 536 subjects remain for the study.

Selection of Subjects during the Study in 2006
Random sampling was performed as described below. In Cotonou, the main urban town, ten out of the 140 neighbourhood were chosen randomly. Per neighbourhood, 20 households were chosen randomly based on the list of households. In each household, an eligible adult was randomly chosen, male and female alternatively. A total of 200 subjects were selected.
In a smaller town, Ouidah as sub-urban area, five out of the 22 neighbourhoods were selected by random. In each neighbourhoods, 34 compounds were randomly selected from the list of compound after assigning number to them. In the household of the compound, an eligible adult is chosen to participate in the study. A total sample of 171 subjects was obtained.
In the rural area of Ouidah, five villages were randomly selected from the 38 existing. A census of the number of hamlets of each selected village was made. The list of families per village has been drawn up. 34 families were drawn randomly from the total list of families in each selected village. The selection of a household per family and then of an adult among the eligible adults of the selected household was done. A sample of 165 subjects was obtained.

Data Collection
Anthropometric measurements were taken by trained investigators. Venous blood samples were collected in the morning after 12 hours fasting. Those blood samples were kept in ice and centrifuged within two hours. The obtained serum samples were stored at -30 °C until analysis in the biochemistry laboratory of the Institute of Applied Biomedical Sciences (ISBA) of Cotonou.

Age
Age was obtained by interview and has been classified into four groups according to FINDRISC criteria: less than 45 years old=0 point; 45 to 55 years=2 points old=; 55 to 64 years old=3points; 64 years old and more=4 points.

Waist Circumference (WC)
WC was measured using a flexible, unstretched graduated measuring tape with an accuracy of 0.1 cm. The tape was placed midway between the last rib and the iliac crest [14]. The IDF thresholds: 94 cm for men and 80 cm for women were used to detect with abdominal obesity. WC has been classified into three groups as followed:  WC < 94 cm for men and WC <80 cm for women: Findrisk score = 0 points  WC = 94-102 cm for men and WC =80-88 cm for women: Findrisk score = 3 points  WC > 102cm for men and WC > 88 for women: Findrisk score = 4 points.

Blood Pressure (BP)
Systolic and diastolic measurements were made using a mercury sphygmomanometer placed on the left arm of each seated subject [28]. Two measurements were taken in 10-minutes intervals with the subject at rest for at least five minutes. For the same individual, the measurements were taken by the same person. The arithmetic mean of the obtained values during the two successive measurements have been used [29]. A harmonized definition was used to define high BP (systolic BP ≥ 130 mmHg or diastolic BP ≥ 85 mmHg). BP has been classified into two groups: 0 points for normal BP and 2 points for a high BP.

Fruit and Vegetable Consumption
The level of fruit and vegetable consumption has been assessed from three non-consecutive 24h reminders according to WHO standards [30]. For the FINDRISC questionnaire, this variable was classified into two groups: 0 points for daily consumption of fruits and vegetables > 400g and 1 point for daily consumption of fruits and vegetables <400g.

Regular Physical Exercises
Physical exercises data were collected using a questionnaire inspired by the WHO Chronic Disease Monitoring Questionnaire. The intensity of physical exercises is evaluated according to the energy consumption induced. Thus, low, medium and high intensity exercises are distinguished [31]. According to the WHO recommendations witch required 30mn of physical exercise per day [32], subjects are divided into two groups: the least active subjects and the active subjects. A physical exercise score of 0 points and 2 points was assigned to each of those groups.

Family History of Diabetes
This variable was documented by completing the questionnaire. The subjects are classified into three groups: 0 points for subjects with no family history of diabetes; 3 points for subjects with a family history of diabetes related to 2nd rank parents (grandparents, uncle, aunt, cousin); 5 points for subjects with a family history of diabetes related to first rank parents (father, mother, brother, sister) 2.6.8. Determination of the FINDRISC Score The FINDRISC score is calculated by summing the scores attributed to the variables: age, BMI, waist circumference, blood pressure, regular consumption of fruit and vegetables, regular practice of 30 minutes of physical exercise, family history of diabetes and history of treatment of High blood pression. History of diabetes treatment was removed from the score calculation. The total score of each subject (maximum of 21) provides the level of T2D risk. This total score were rank as followed: very low (<7), low (7)(8)(9)(10)(11), moderately high (12)(13)(14), high (15)(16)(17)(18)(19)(20) and very high.

Measurement of Blood Glucose
Blood glucose test was done in the laboratory of biochemistry of the Institute of Applied Biomedical Sciences (ISBA) of Cotonou. Venous blood samples (10 ml) were drawn after an over-night fast of 12 hours and were centrifuged within two hours. Fasting plasma glucose was determined using the glucose oxidase method. Serum concentrations of high-density lipoprotein Cholesterol (HDL) and Triglycerides were analyzed by enzymatic methods with certified laboratory standards from the USA. The blood glucose was classified according to the IDF/WHO standards: normal (<6.1 mmol/l), High blood glucose for fasting glucose (6.1-6,9 mmol/l) and intolerance glucose (6,9-7 mmol/l [31,33].

Data Analysis
The data were analysed with IBM SPSS Statistics 21 (IBM United States 2012) and Stata/SE 11.0 (StataCorp 2009). The quantitative variables were expressed as the mean followed by the standard deviation. Qualitative variables were expressed as a percentage. The Receiver Operator Characteristic (ROC) curve was used to determine the performance of the FINDRISC tool in selecting subjects at risk for T2D in this study compared to the fasting blood glucose test. FINDRISC threshold-optimal was determined by the point corresponding to the maximum value of the sensitivity. Using the Youden index calculated by the formula (sensitivity+ specificity -1) [34], we determined the risk score thresholds for both maximum sensitivity and specificity predicting the onset of type 2 diabetes. The statistical significance level for the tests was set at p <0.05.

66
Optimal Threshold of the Finnish Diabetes Risk Score (Findrisc) for Screening at-Risk Adults in an African Population in Southern Benin

Ethical Considerations
In the previous study, the data used was approved by the ethics committees of the Faculty of Medicine of University of Montreal and the Benin Ministry of Health. Written informed consent was obtained from each participant prior to his/her admission into the study.

Sample Description
The general features of the participants were described in Table

Participant's FINDRISC Score in Different Study Areas
Subjects with high FINDRISC were almost twice as numerous in urban areas as in peripheral areas with p = 0.000 (Table 2).

FINDRISC Score According to the Subjects' Glycaemic Status
All subjects with a very low FINDRISC score had low glycaemic status, but all subjects with a very high FINRISC score all had glucose intolerance with p = 0.01 (Table 4).

Performance of FINDRISC Compare to Glycemia in Detection of at-Risk Subjects
Findrisc score have high performance to detect at-risk subjects as identified by the blood glucose test. The area under the ROC curve of the test was 0.86 (95% CI: [0.81-0.90]) as shown in Figure 1, with sensitivity of 77%, specificity of 89%, positive predictive value of 45% and a negative predictive value of 71%. The optimal risk score for identifying at-risk subjects was 8.5. The Youden index is 0.655. (Table 5).

Discussion
This study is one of the first in sub-Saharan Africa trying to evaluate the performance of the FINDRISC score in the detection of subjects at risk for type2 diabetes. Apart of the validation of the tool, the study identified the threshold-optimal point for the type2 diabetes risk score. The results showed that the optimal threshold of the FINDRISC score for detection of high blood glucose is 8.5 with high sensitivity of 77%, high specificity of 77%. Potential subjects at risk of autoimmune diabetes are reported to be very poorly represented in this study because age of study subjects is between 25 and 65 years of age and type 1 diabetes (T1D) is often diagnosed in children and adolescents [33,35]. In addition, it has been shown that type 2 diabetes covers 90% of diabetes [29]. In addition, it is increasingly recognized that type 1, autoimmune diabetes can occur at older ages [36,37]. With the increase in the prevalence of obesity in the world, more young people develop TD2, and many cases of TD1 become obese [38]. The emergence of adults with type 1 diabetes requires the adoption of new strategies for managing the disease [39]. Deficiencies of essential nutrients are thought to be involved in the onset of type 1 diabetes and cardiovascular disease [40]. One study suggested that adopting a westernized diet may increase the risk of diabetes especially in the high-risk population [41].In the case of pancreatitis, type 3c diabetes appears most often after the diagnosis of pancreatitis [42]. It is not always easy to diagnose and classify a patient with type 3c diabetes mellitus properly. A review of studies currently available on this topic suggests a prevalence of 5% to 10% for type 3c diabetes mellitus among all cases of diabetes mellitus in Western populations [43].
Nutrition education and lifestyle delay the onset of diabetes whatever the nature of diabetes, consideration of at-risk individuals of any type of diabetes is necessary to delay or prevent the appearance of the disease. Of course, these other types of diabetes are very poorly represented in the general population. For the present study, at-risk subjects considered are mainly for type 2 diabetes. Other types of diabetes have not been explored

The Performance of the FINDRISC Tool in Identifying Subjects at Risk for Type2 Diabetes
The research of an optimal threshold ideally requires to take into account epidemiological (disease prevalence) and medico-economic data (treatment cost, cost of adverse effects of treatment) [32]. In those conditions, when disease has an expensive treatment and serious side effects, the number of false positives should be reduced as possible, Universal Journal of Public Health 7(2): 63-72, 2019 69 so a high sensitivity should be chosen. In opposite, some diseases have serious complications that can be avoided if simple treatment is set up early, so the test would have high specificity. The principle was respected for the threshold value of type2 diabetes risk score obtained in our study. The threshold-value has a higher specificity than sensitivity. This can be explained by the fact that diabetes has serious complications that can be avoided with early treatment [33].
Olamoyegun and al. have shown that the FINDRISC tool provides an accurate estimate of the risk of type2 diabetes in a black population in Nigeria in 2017 [44]. On the other hand, the tool performs better when the prevalence of diabetes is not high [45].

Optimal Threshold for the Prediction of Type2 Diabetes with the FINDRISC Tool on Adults in Benin
We observed that the optimal threshold-value of risk score for Benin people is 8.5. Higher FINDRISC score thresholds were found in several studies.
In Algeria in 2014, the threshold-value obtained is 13 for women and 11 men and the FINDRISC score has been validated as a high performance tool with excellent ability to predict type2 diabetes mellitus in the Algerian community [46]. Those FINDRISC score thresholds are high compared with the threshold found in our study.
In addition, the threshold used in Caucasians studies, which ranges from 11 to 15 in most FINDRISC questionnaire validation studies is higher than that of our study. Optimal thresholds used were higher or equal to 11 [20,21,[47][48][49].
On the other hand, we have noted optimal thresholds of diabetes risk scores very similar to ours in a few studies aiming the validation of the tool. For example in Bergmann and al.'s study [12] where only six components were used instead of the eight components of FINDRISC, the optimal threshold-value obtained was 9. This finding is similar to the original study of diabetes risk score for the prediction of risk of type2 diabetes where the family history component of diabetes has not been taken into account [18]. In other words, the optimal risk thresholds for type 2 diabetes close to the threshold value observed in our study did not take into account all the current components of the FINDRISC questionnaire as for our study. The higher thresholds observed compared to our threshold could be due to the values of several components of the questionnaire that are different for race and ethnic. For example, for the age component of the tool, several Caucasian studies have reported a mean age greater than that of our study [12,19,21,36]. This increase in age could lead to an increase in the risk score in this population [12,21,47,49]. Several studies conducted in South Africa have suggested threshold values of waist circumference different from those currently used for African populations [50][51][52]. In addition, a study that looked at the optimal threshold-values of cardiometabolic risk for sub-Saharan Africans in 2015, revealed that the threshold-values of Benin people are different from those used for sub-Saharan Africans. The optimal WC cut-off points to predict the presence of at least one other component of the metabolic syndrome was 80 cm for men and 90 cm in women [53]. This difference in waistline threshold values between different population, especially for black and Caucasian subjects may explain the low diabetes risk score in sub-Saharan black subjects. However, in a study conducted in Nigeria in 2017 on the performance of the FINDRISC questionnaire to screen subjects with dysglycemia and undiagnosed subjects for type 2 diabetes, the threshold-value was 14 [44]. This study, therefore pointed out that the FINDRISC questionnaire provide a precise estimate of the absolute risk for screening for type2 diabetes and that the tool should be recommended in poor countries with insufficient resource for diabetes detection.
In North Africa, the FINDRISC score has been validated as a high performance tool with excellent ability to predict the development of type2 diabetes mellitus in the Algerian community with a threshold-value of 13 for women and 11 men in 2014 [46]. Those FINDRISC score thresholds are high compared with the threshold found in our study. This difference could be explained by the difference between the dates of the different studies. In fact, the data used in this study dates back to 2006 when the study conducted in Nigeria used 2012 data and that of Algeria used the 2014 data.
This study has some limitations. It was carried out only in the southern part of Benin and, therefore, the extrapolation of the results to other parts of the country requires some caution. In this study, we did not perform oral glucose tolerance test or glycated hemoglobin test (HbA1C), that are the best tests recommended to detect diabetes. The study was based only on fasting blood glucose levels. The risk factors for type2 diabetes that make up the questionnaire were calculated from a database that measured all those parameters. Larger studies of Africans living in different settings will be needed to advocate for specific risk thresholds for type2 diabetes in black populations.

Conclusions
This study conducted in southern Benin showed satisfactory results. The study have validated the Findrisc questionnaire and its score and identified the threshold-optimal value of the risk score for detection of type 2 diabetes in black populations like Beninese population. This threshold-optimal value of risk scores in Benin is lower than the threshold used in Caucasians. In view of the increasing prevalence of T2D and the economic costs of the disease, early identification of high-risk T2D 70 Optimal Threshold of the Finnish Diabetes Risk Score (Findrisc) for Screening at-Risk Adults in an African Population in Southern Benin subjects with simple and valid tool as Findrisc score is a public health priority that must be encouraged in developing countries like Benin Republic. Other studies including larger and more representative samples need to be done to reinforce our study findings.