Evaluation of Indian Diabetes Risk Score and Random Blood Sugar Testing for Opportunistic Screening of Type 2 Diabetes Patients at a District Hospital of Gujarat

Abstract

Background:

India is home to 69.2 million diabetics. For opportunistic screening of type 2 diabetes mellitus (DM), random capillary blood sugar (RBS) testing is used. Another method is Indian Diabetes Risk Score (IDRS), which is a simple and cost effective method for opportunistic screening of type 2 DM patients. The aim is to evaluate the screening test parameters of RBS testing and IDRS for opportunistic screening of undiagnosed type 2 DM patients.

Materials and Methods:

A cross-sectional study was done during February 2017 to August 2017 at a district hospital of Western Gujarat. A sample size of 317 patients was calculated using Buderer’s formula. Systematic random sampling was used and every third patient was selected from the general Outpatient Department(OPD) attendees of 30 years or more. MS Excel and Epi Info v7.2 was used for statistical analysis. Screening parameters and accuracy of IDRS and RBS were calculated taking result of the oral glucose tolerance test as clinical reference.

Results:

The mean age of study participants was 50.9 (SD 12.17) years with 44.2% males and 55.8% females. Sensitivity and specificity of RBS was 72.4% and 69.1%. Sensitivity and specificity of IDRS was 93.1% and 29.0%. On simultaneous (parallel) screening by IDRS and RBS, sensitivity was 98.3% and specificity was 23.2%. In sequential screening, where IDRS was used followed by RBS, sensitivity was 67.2% and specificity was 74.9%.

Conclusions:

This study has found that sequential screening using a simple diabetes risk score like IDRS followed by RBS is having higher accuracy and reduced cost of opportunistic screening of type 2 diabetes. Adopting sequential screening using IDRS as first step of screening followed by RBS in those found as high risk by IDRS is recommended.

INTRODUCTION

India is experiencing raising trend of diabetes with estimated 8.7% diabetic population in the age group of 20 and 70 years. India is home to second largest number of adult diabetics worldwide. Indians succumb to diabetes, high blood pressure, and heart attacks 5 to 10 years earlier than their western counterparts, during their most productive years.[1,2,3]

Under the National Program for Prevention and Control of Cancer, Diabetes, Cardiovascular Diseases and Stroke (NPCDCS), opportunistic screening is done in patients of 30 years or more using Random Capillary Blood Glucose testing (RBS) by Glucostrip method for early diagnosis of type 2 diabetes at the point of primary contact with any health care facility. Those having RBS ≥140 mg/dl are further tested for FBS (fasting blood sugar levels) and PP2BS (testing of blood glucose 2 hours after the lunch) by testing of venous plasma.[4,5]

Various studies have estimated screening test paraments of RBS. These studies suggest that RBS is having low sensitivity of around 70% and low positive predictive value (PPV), although specificity of RBS is good at around 90% for screening of diabetes mellitus.[6-9]

The International Diabetes Federation has recommended the use of ethnic-specific brief questionnaires to help health-care professionals to quickly identify people who may be at a higher risk. In limited resource setting, diabetes risk scores can be a simple and cost-effective method of identifying people with undiagnosed type 2 diabetes or at risk of developing diabetes.[10] Indian Diabetes Risk Score (IDRS) is a simple and easy-to-use risk score developed using large-scale population-based study by Mohan et al.[11] The performance of IDRS has been tested and validated in various studies and settings. It has been seen in these studies that IDRS has greater sensitivity of 70% to 100% but has lesser specificity ranging from 17% to 62%.[12-14]

This study was conducted with the aim to estimate screening test parameters of RBS testing and IDRS for opportunistic screening of undiagnosed type 2 diabetes patients. Evaluation of RBS and IDRS in a single study has also provided screening test parameters of RBS and IDRS if these two screening tests were used sequentially or simultaneously. These screening test parameters would be useful to decide appropriate screening test algorithm.

MATERIALS AND METHODS

We conducted a cross-sectional study during February 2017 to August 2017 at a district hospital of Western Gujarat. Buderer’s formula was used to calculate the sample size.[15,16] For this, anticipated value of sensitivity or specificity of the screening test and estimated prevalence of disease is needed. To calculate prevalence, a pilot study in a sample of 24 patient was carried out, of which 4 patients were diagnosed having diabetes mellitus giving prevalence of 16.67%. Anticipated sensitivity of RBS based on various studies was taken as 73%, whereas anticipated sensitivity of IDRS was taken as 90%. Sample size based on anticipated sensitivity of RBS was found to be more than the sample size based on anticipated sensitivity of IDRS or anticipated specificity of RBS or IDRS. So, the final sample size of 317 was calculated taking values of Z²_1−α/2 = 3.84 (for 95% Confidence Interval), Sensitivity of RBS = 73%, Prevalence = 16.67%, and absolute precision L = 12%.

We used systematic random sampling to select every third patient from the general OPD attendees of 30 years or more coming to the district hospital. All the selected patients underwent RBS measurement using SD CodeFree™ Blood Glucose Meter and SD CodeFree™ Blood Glucose Test Strip, manufactured by SD BIOSENSOR (REPUBLIC OF KOREA).[17] Oral Glucose Tolerance Test (OGTT) was done after RBS using 75 gm glucose dissolved in 200 ml of water. The time of completion of 75 gm glucose was noted for each patient and venous blood was collected in fluoride bulb 2 h after the ingestion of 75 gm glucose. The venous blood sugar was measured at a National Accreditation Board for Laboratory (NABL) certified biochemistry laboratory. Those having RBS of ≥140 mg/dl were considered “positive” while those having RBS <140 mg/dl were considered “negative.

Inclusion criteria for this study was patients of age 30 years or more coming to the general OPD for any illness except treatment or follow-up of diabetes mellitus. The exclusion criterion was kept as patients who were known case of diabetes mellitus or those requiring urgent medical attention. Patients who could not complete 75 gm glucose or who vomited after ingestion of 75 gm glucose were also excluded from the study.

The data were entered in MS Excel and the data analysis was done using Epi Info v7.2. Sensitivity, specificity, PPV, negative predictive value (NPV), and accuracy of IDRS and RBS were calculated taking result of OGTT as clinical reference. The same parameters were calculated for simultaneous (parallel) and sequential testing by IDRS and RBS.

The study was approved by the Institutional Ethics Committee. Informed written consent was taken from all patients prior to enrolment in the study and confidentiality was assured.

RESULTS

In this study, data of 317 study participants were analyzed. The mean age of study participants was 50.9 years with SD of 12.17. Out of the total 317 study participants, 140 were male (44.2%) and 177 (55.8%) were female; 29% of the study participants were illiterate and similar number were able to read; and 23.3% had completed primary school, while 13.2% had completed secondary school. Work-profile wise, 26.2% were homemaker, 20% were doing job, while 18.3% were laborers; 16.1% had their own business whereas 14.2% were retired. 38.5% (n = 122) patients were screened positive, while 61.5% (n = 195) were screened negative for type 2 diabetes by RBS measurement.

It was found that sensitivity of RBS is 72.4%. RBS measurement missed 16 out of 58 diabetics. The NPV of RBS was found to be 91.8%. Specificity of screening by RBS measurement was found to be 69.1% and PPV 34.4%. Screening by RBS will require 122 out of 317 patients to undergo further diagnostic evaluation, that is, 38.5% patients will need to undergo FBS and PP2BS [Table 1].

Table 1.

Screening test results of RBS

Screening by RBS	OGTT result		Total (%)
	Diabetic	Non-diabetic
Positive (≥140 mg/dl)	42 (72.4)	80 (30.9)	122 (38.5)
Negative (<140 mg/dl)	16 (27.6)	179 (69.1)	195 (61.5)
Total	58 (100)	259 (100)	317 (100)

IDRS was applied to all the patients enrolled for the study. There are four variables in IDRS namely, age, family history of diabetes in parents, physical activity, and waist circumference. The maximum total score is 100. A score of ≥60 is considered as high risk for type 2 diabetes. In this study, those having IDRS ≥60 were considered as “Positive” and others as “Negative” to calculate the screening test parameters of IDRS [Table 2].

Table 2.

Screening test results of IDRS

Result of IDRS	OGTT result		Total (%)
	Diabetic	Non-diabetes
High Risk (Score≥60)	54 (93.1)	184 (71.0)	238 (75.1)
Non-High Risk (Score<60)	4 (6.9)	75 (29.0)	79 (24.9)
Total	58 (100)	259 (100)	317 (100)

It was found that IDRS is highly sensitive and detected 93.1% (n = 54) of true positive diabetics out of total 58 diabetic patients. The sensitivity of IDRS was found to be higher than that of RBS (72.4%). IDRS missed 6.9% of true positives. NPV of IDRS was also found to be very high at 94.9%.

Specificity of IDRS was found to be 29% which is quiet lower than that of RBS (69.1%). Similarly, PPV of IDRS was also low at 22.7%, which means there are large number of false positives when screened by IDRS. Screening by IDRS alone may lead to an important disadvantage of excessively burdening the diagnostics services because 238 out of 317 patients (75.1%) would require further diagnostics tests like FBS and PP2BS.

We examined simultaneous (parallel) and sequential screening approach using IDRS and RBS. In simultaneous screening, screening result was considered “positive” if either IDRS is ≥60 or RBS is ≥140 mg/dl or both. The screening test result was considered “negative” only if both tests were negative. Out of 317 study participant, 238 study participants were screened high risk by IDRS whereas 122 study participants were screened positive by RBS. For 104 study participants, both screening tests were positive. Total 256 study participants had either one or both tests positive [Figure 1, Table 3]. It was found that this approach is highly sensitive and lead to detection of 98.3% of true positives. NPV is also very high at 98.4%. So there was net gain in sensitivity and NPV as compared with individual testing by the two tests. But this happens at the cost of specificity and PPV which was found to be 23.1% and 22.3%, respectively.

Figure 1.

Simultaneous (parallel) screening by IDRS and RBS

Table 3.

Screening test result by simultaneous (parallel) testing

Screening test result by simultaneous testing	OGTT result		Total (%)
	Diabetic	Non-diabetic
Positive by either one or both tests	57 (98.3)	199 (76.8)	256 (80.8)
Negative by both tests	1 (1.7)	60 (23.2)	61 (19.2)
Total	58 (100)	259 (100)	317 (100)

This approach would result in large number of positively screened patients (256 out of 317, 80.8%) who will require further laboratory investigations which will lead to overburdening of laboratory services.

In sequential screening, all the participants were first screened by IDRS as it is non-invasive and has higher sensitivity. Those having IDRS ≥60 (i.e., identified as high risk by risk score) were further screened by RBS. If the study participant had RBS ≥140 mg/dl, he/she was considered as “positive.” If the study participant had IDRS <60 in the first step of screening, or the study participant had IDRS ≥60 but RBS <140 mg/dl in second step of screening, then he was considered as “negative.” As shown in Table 2, out of 317 patients, 238 had IDRS ≥60 and 79 had IDRS <60. If the second step of screening, that is, screening by RBS, was applied to the these 238 participants who had IDRS ≥60, than there would be 104 patients who had RBS ≥140 mg/dl, so they were considered “positive”, and 134 had RBS <140 mg/dl, so they were considered “negatives” [Figure 2, Table 4].

Figure 2.

Sequential screening by IDRS followed RBS

Table 4.

Screening test result of sequential testing

Screening test result by RBS when done in patients having IDRS score ≥60	OGTT result		Total (%)
	Diabetic	Non-diabetic
Positive	39 (72.2)	65 (35.3)	104 (43.7)
Negative	15 (27.8)	119 (64.7)	134 (56.3)
Total	54 (100)	184 (100)	238 (100)

When screened sequentially by IDRS followed by RBS, 39 out of 58 diabetics were screened positive. Overall sensitivity and NPV were found to be 67.2% and 91.1%, respectively, which are slightly less than that of screening by RBS alone (72.4% and 91.8%).

The net specificity of sequential screening was found to be 74.9% which was an increase by 5.5% as compared with screening by RBS alone. As the specificity was increased, there were lesser number of false positives. The PPV of sequential testing was found to be 37.5% which was 3.1% higher than screening by RBS. This PPV is highest compared with other approaches like screening by RBS, IDRS, or simultaneous screening.

DISCUSSION

Various studies have evaluated screening test parameters of RBS at a cut off similar to this study. These studies found the sensitivity of 60% to 70%, specificity around 90%, PPV 30% to 70%, and NPV 90% to 98%.[6-9] In this study, it was found that the sensitivity, PPV, and NPV of RBS was almost similar to that reported in these studies but specificity was found to be lesser than these studies.

In the study done by Mohan V et al.[11] to develop IDRS, sensitivity, specificity, PPV, NPV, and accuracy were found to be 72.5%, 60.1%, 17.0%, 95.1%, and 61.3%, respectively. In this study, it was found that sensitivity was higher but specificity was lower than the original study. This may be because this study was conducted in a hospital, whereas the study done by Mohan V et al.[11] was a community-based study. Moreover, IDRS is applied to subjects of 30 years or more in this study which might have affected its screening test parameters.

Several other studies have used IDRS for screening and found that sensitivity of IDRS is in the range of 70% to 100% and specificity from 17% to 62%.[13,14,18,19] Most of these studies are community based or conducted in special groups in contrast to this study which is done in OPD attendees of more than 30 years at a hospital.

This study found that sequential screening by IDRS followed by RBS testing in those who have IDRS score ≥60 had sensitivity of 67.2% and specificity of 74.9%. PPV and NPV of this approach of screening were 37.5% and 91.1%, respectively. Accuracy of this approach was high at 73.5%. The requirement of RBS testing was reduced by almost 25% and need of confirmatory tests like FBS and PP2BS by 5.7%. This would lead to reduced cost of screening.

The findings of this study collaborates with results of the study done by Franciosi et al.[20] to evaluate an opportunistic screening strategy consisting screening by DRS as the initial instrument. This study concluded that use of DRS as initial screening instrument is inexpensive and is a simple approach having adequate sensitivity. Those having higher than cut-off score can be subjected to FBG. OGTT can be done for those having FBG higher than 5.6 mmol/l. Using this approach, two-thirds of the patients would need FBG. Further, 38% would need OGTT for confirmation of Diabetes instead of 56% individuals requiring an OGTT if FBG is used as initial screening test.

Martin et al.[21] concluded in his study that screening strategy applying the FINDRISK questionnaire followed by HbA1c testing and performing an OGTT on those individuals having HbA1c more than the cut-off could be a cost-saving approach for screening large populations.

Glumer et al.[22] has shown in their study that implementation of a stepwise screening strategy using a risk score as the initial step followed by biochemical test substantially reduced the workload of healthcare system, as the percentage of individuals who needed additional blood tests was reduced. This study concluded that targeted screening using a questionnaire followed by FPG was the strategy of choice.

CONCLUSION

This study has found that sequential screening using a simple diabetes risk score like IDRS followed by RBS will reduce the cost of screening. Adopting sequential screening by IDRS as first step of screening followed by RBS in those found as high risk by IDRS is recommended. Further studies may be carried out to develop such simple scoring systems to categorize populations at high risk of development of diabetes using various parameters. Use of such scoring systems should be explored to make the screening programs cost effective and convenient.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.