Diagnostic accuracy of the Finnish Diabetes Risk Score (FINDRISC) for the prediction of undiagnosed type 2 diabetes, prediabetes, and metabolic syndrome in the working community of the Lebanese University (LU)

Risk scores were mainly proved to predict undiagnosed type 2 diabetes mellitus (UT2DM) in a non-invasive manner and to guide earlier clinical treatment. The objective of the present study was to assess the performance of the Finnish Diabetes Risk Score (FINDRISC) for detecting three outcomes: UT2DM, prediabetes, and the metabolic syndrome (MS). This was a prospective, cross-sectional study during which employees aged between 30 and 64, with no known diabetes and working within the faculties of the Lebanese University (LU) were conveniently recruited. Participants completed the FINDRISC questionnaire and their glucose levels were examined using both fasting blood glucose (FBG) and oral glucose tolerance tests (OGTT). Furthermore, they underwent lipid prole tests with anthropometry.

Blood Pressure measurement was taken after the participant had been seated and relaxed for 5 minutes without any distractions, using an automatic monitor (Ross max monitoring, Swiss design) with appropriate cuff's size. Furthermore, the person's upper arm was put into the cuff loops, 1 or 2 cm above the elbow, then letting it comfortably rest on the table (22). Those who had a blood pressure level higher than 120/80 mmHg upon measurement, were noti ed to seek medical advice.

Laboratory measurements
Participants were instructed to fast for at least 12 hours and abstain from vigorous exercise in the evening and the morning of the investigation. They were also asked to abstain from caffeine and smoking on the morning of the visit.
After ensuring that the fasting period was accomplished completely, the blood sampling procedure was explained by a trained laboratory technician. A total of 3-5 ml of fasting venous blood sample was collected in a serum clot activator tube and centrifuged at 4000 rpm for 10 min on the same day, and then serum was transferred to another tube and stored at -22°C for biochemical examination. Following this, a load of 75 g of anhydrous glucose in a volume of 200 ml was administered to each individual for the OGTT test (14). After two hours, a second blood sample was drawn to assess the glucose levels. Fasting blood glucose (FBG), triglycerides (TG), total cholesterol (TC), and high-density lipoprotein-cholesterol (HDL-C) levels were detected using a biochemical analyzer (Unicel DxC 600, Synchron Clinical System, BECKMAN COULTER, Cobas C111 ). However, low-density lipoprotein-cholesterol (LDL-C) level was calculated by the Freidwald formula, (23) only if the total TG level did not exceed 300 mg/dl. All these blood analysis procedures were conducted in certi ed laboratories located in each region (Lebanese University Medical Centre, Mount Lebanon; Hammoud hospital, South Lebanon; Libano Français Hospital, Bekaa and Tripoli Medical Center, North Lebanon). The results of the blood tests were provided to the participants.

Outcomes
We had three outcome variables of interest: UT2DM, Prediabetes, and MS.
Both UT2DM and Prediabetes were de ned according to the latest American Diabetes Association (ADA) criteria (14).
Whereas, MS was de ned according to the latest National Cholesterol Education Program Adult treatment panel III (NCEP ATP III) diagnostic criteria (2005 revision) (24). At least 3 of the following criteria were present Abdominal obesity (WC ≥ 102 cm in men and ≥ 88 cm in women) Hypertriglyceridemia (TG ≥ 150 mg/dl (1.695 mmol/l) Low HDL-C (HDL < 40 mg/dl in men and < 50 mg/dl in women) Elevated blood pressure (Systolic blood pressure (SBP) >130 mmHg or Diastolic blood pressure (DBP) > 85 mmHg) or the use of antihypertensive medication.

Statistical analyses
Statistical analysis was conducted using SPSS (IBM Corp, SPSS Statistics version 23). Descriptive statistics of those who underwent blood tests were expressed as means (± standard deviation) for continuous variables and as proportions for categorical variables. Differences in the socio-demographic variables between genders were computed   using an independent samples t-test for continuous variables and a Chi-square test for categorical variables, while   between FINDRISC categories using a one-way ANOVA test for continuous variables and a Chi-square test for categorical variables. To evaluate the FINDRISC accuracy performance we calculated the area under the receiver operating curve (AUROC), sensitivity (the probability that the test is positive for subjects with type 2 diabetes), speci city (the probability that the test is negative for subjects without type 2 diabetes) with 95% CIs (95% con dence intervals).
To create the ROC curve, sensitivity was plotted on the y-axis, and the false-positive rate (1-speci city) was plotted on the x-axis. Then optimal cut-off points were determined by the point with the closest distance to (0; 1) in the ROC curve which maximizes the sensitivity and speci city of the test (Tradeoff between sensitivity and speci city).
These ndings indicate that the prevalence of both diabetes and prediabetes are high in different Lebanese settings. As for MS prevalence, it was estimated to be 36% among LU employees. Similarly, a recent cross-sectional study has been carried out in Notre Dame University (NDU) employees on the three campuses (Zouk Mosbeh, North, and Al Chouf) and found that 23.5% of the participants were suffering from MS (26). These ndings are alarming, suggesting that LU employees are, in general, unaware of their health status which is highlighted by a low percent of physical activity practice (85%) (27), high Waist circumference especially for men (102.8±12.1cm) (28) and an overweight population (29). These factors have been largely discussed and identi ed as risk factors for diabetes and 'metabolic syndrome' and associated health problems. Thus, the importance of the FINDRISC use among them is highlighted.
Performance of FINDRISC in detecting UT2DM, prediabetes and MS Originally, the FINDRISC questionnaire was developed longitudinally as a future predictor of diabetes in the Finnish population (30) and was validated from a multivariate logistic regression model 5 years later. It was subsequently crosssectionally validated using a maximum score of 26 (31). Later on, it has been assessed in a cross-sectional manner in several Asian (32)(33)(34), European (35)(36)(37)(38)(39)(40)(41)(42), and American countries (43)(44)(45). In these studies, the optimal cut-off points for detecting UT2DM varied widely from 8.5 to 17 with a sensitivity ranging from 48% to 84% and a speci city ranging from 30.9% to 95%. Also, the AUROC went from 0.569 to 0.88. This vast variability indicates the need for assessing the tool within its target population.
In this study, FINDRISC had a good discriminative ability for detecting UT2DM with an AUROC  (46). In other words, the AUROC for MS discrimination was 0.72 in men and 0.75 in women. However, the optimal cutoff values for detecting T2DM and prediabetes were both 11 with lower sensitivities and speci cities than the ones found in this study, and the optimum cutoff for MS was not established. Similarly, FINDRISC was also found to perform well in the detection of MS (AUC = 0.77) in Taiwanese (47), but the optimal cutoff point was not reported. One previous cross-sectional study in Greece (35) reported a threshold for MS of 15 which is higher than the one reported in our study. However, it is well known that prediabetes which is a combination of excess body fat and insulin resistance, is considered an underlying etiology of MS (48). In turns, MS is considered as a risk factor for T2DM (49) which may explain why 70% of people with prediabetes in this study had MS (P < 0.0001) and 76% of those with UT2DM had MS and that's why the threshold for MS is localized between the thresholds for prediabetes and UT2DM in our community.
To date, only one study assessed the predictive ability of FINDRISC in detecting incident cases of MS (AUC = 0.65) rather than prevalent cases at a cutoff of 12 (50).
It is also worth mentioning that men had always higher AUROC values as well as lower cutoff values than women, speci cally for UT2DM and MS in the current study. In other words, men tend to have more risk factors putting them at a higher risk for diabetes, prediabetes, and MS which improves the predictive ability of FINDRISC when compared with women and increases their scoring in FINDRISC and thus limiting their threshold to lower values. In this study, a synergistic interaction for the combined BMI (p < 0.0001), WC (p < 0.0001), smoking (p < 0.0001), could renders men more prone for diabetes with higher prevalence for UT2DM (p = 0.001) and MS (p < 0.0001) The usefulness of FINDRISC as a screening tool among LU workers The advantage of the FINDRISC relies on its self-report questions so that LU workers that reported to be extremely busy because of their work and daily life stressors can nd it easier to ll the FINDRISC quickly and rate their current health status. Being at higher risk based on FINDRIC score would be a su cient trigger for them to start applying lifestyle changes or to seek health professionals' help.
Strengths and limitations of the study Some limitations warrant considerations. First, a misclassi cation bias could be introduced because the diagnosis of diabetes of the respondents was self-reported. Further, the diagnosis of diabetes and prediabetes of the included participants was not con rmed by repeat testing on a separate day as recommended (14). However, these tests may pose additional costs on our limited budget. Second, a selection bias could be present as the participants were drawn only from LU campuses and, thus, the results may not be generalizable to the rest of the Lebanese citizens living in other settings. Third, we could not assess the ability of the FINDRISC to catch the future risk of having diabetes and MS as it was tested in some longitudinal studies (50)(51)(52) This study has also considerable strengths. To our knowledge, this is the second study that has been carried out in an Arabic country in the Middle East region which has investigated the validity of FINDRISC. A previous study was conducted in Kuwait and showed similar results (33). Additionally, a recent Jordanian study pointed out the usefulness of FINDRISC to screen for type 2 diabetes in a young student population but didn't have the opportunity to validate it (53). Second, a selection bias was avoided since our sample was fairly divided between men (44.8%) and women (50.2%) and thus the gender differences in the study outputs are not biased. Third, the diagnosis of diabetes was done based on a combination of two plasmatic tests as it is ideally recommended which are the FBG and OGTT. Thus, the misclassi cation bias would be lessened, and the estimation of the risk of T2DM as well as the performance of FINDRISC are optimized.

Conclusion and perspectives
This cross-sectional study has successfully demonstrated that FINDRISC could be useful as a rst-line screening tool that identi es employees with UT2DM, prediabetes, and MS that might bene t from lifestyle modi cation. FINDRISC model could be also bene cial for community-based interventions and screenings as well as in clinical practice by the health professionals. In future studies, FINDRISC should be validated on a larger and more representative sample of the Lebanese population so Lebanese citizens living in a resource-poor setting like rural areas would bene t the most. Also, FINDRISC should be assessed in a prospective manner (Longitudinal study) that allows the identi cation of incident cases of diabetes and MS rather than prevalent cases.

Declarations
Ethics approval and consent to participate This work was pertinent to the Declaration of Helsinki. The informed consent and questionnaire were approved by the   ROC curve for MS for the sample population and by gender