Determination of Prevalence of Type 2 Diabetes Mellitus by Screening tests using a mathematical formula in place of invasive blood tests (gold standard)
METHODS: From population of 18800 residing in two adjacent rural areas of Delhi, a systematic random sample of 1005 adult subjects was screened for diabetes by using urine benedicts test, Canrisk questionnaire, Madras Diabetes Research Foundation-Indian Diabetic Risk Score (MDRF-IDRS) and determined prevalence of diabetes (pA) gauzed by each of these screening tests. For determination of validities of these tests, simultaneously, each subject’s glycaemic status was confirmed by standard fasting (FBS) and postprandial blood sugar (PPBS) levels. The blood test was also used to determine true prevalence which was cross-checked with the each prevalence estimated (Pe) by the above stated screening tests using a mathematical formula Pe = [(P (A) + Sp-1)/ (Se+Sp-1)].
RESULTS: Fasting urine test had sensitivity and specificity of 65.2% and 99.9% when compared against the FBS levels (≥ 126mg/dl). Postprandial urine test’s sensitivity and specificity were 84.8% and 99.7% when compared with PPBS level (≥ 200 mg/dl). At cut off 7, the sensitivity and specificity of Canrisk were 89.1% and 53.2% with FBS. With PPBS the sensitivity and specificity of the Canrisk were 82.6% and 52.9%. For MDRF-IDRS at cut off with > 60 against FBS the sensitivity and specificity were 52.2% and 68.6%. Against the PPBS level the sensitivity and specificity of MDRF-IDRS were 52.1% and 68.6%. The true prevalence of T2DM in more than 18 years of population by FBS was 4.5% while that by using mathematical formulae that estimated by urine test, Canrisk test and MDRF-IDRS was 4.4%, 4.4 and 4.3% respectively. When more than 35 years age-group was selected, true prevalence was 7.4% and estimated prevalence by Canrisk test was 7.1% (Fasting) and 6.9% (PP). By fasting urine test it came out to be 7.2% and by PP urine test it was 7.4%. In population l8-35 years, the prevalence of diabetes was 1.1% by blood glucose test. By using Canrisk, it came out to be 1.04%.
CONCLUSIONS: Individual screening tests such as urine, Canrisk and MDRF-IDRS can be used to estimate prevalence rates of diabetes in rural areas by means of mathematical formula which would be close to true estimates.