Abstract
T
In addition to the increasing epidemic, the Asian diabetes scenario is characterized by onset at a relatively young age and low body mass index. 3 A prospective study in the United Kingdom showed that Asians had equivalent diabetes incidence rates at substantially lower obesity levels (25.2 kg/m2) than the conventional European cut point (30 kg/m2). 4 A large survey conducted in the Eastern province of Saudi Arabia showed that the risk for hypertension and diabetes progressively increased beginning with a body mass index as low as 21 kg/m2. 5 Another study conducted among Omani Arabs showed that the optimal cutoffs for body mass index among men and women are <22.6 and <22.9 kg/m2, respectively, for the 10-year risk of cardiovascular disease using the Framingham score. 6
It is well known that chronic complications are the major outcome of T2DM progression, which reduces the quality of life of patients, incurs a heavy burden on healthcare systems, and increases mortality. 7 –9 Epidemiologic evidence also suggests that the complications of T2DM begin early in the progression from normal glucose tolerance to frank diabetes.
To compound to these problems, the International Diabetes Federation has estimated that as many as 175 million people worldwide, or close to half of all people with diabetes, are unaware of their disease; the proportion of undiagnosed diabetes in Asia and Middle East ranges from 40.7% to 63%. 2 Although undiagnosed diabetes is a substantial problem, population-wide screening for diabetes is not appropriate. Opportunistic identification of people with risk factors for undiagnosed T2DM is feasible and cost-effective. The Hoorn Screening Study in The Netherlands showed that the prevalence of microvascular complications was similar in diabetes patients who were detected by a targeted screening procedure and in newly diagnosed diabetes patients in general practice, but that the glucose and hemoglobin A1c levels were lower in those identified by targeted screening. 10 This again highlights the importance of early targeted screening.
Diabetes risk scores (DRS) have been developed in many countries based on epidemiological surveys. These scores provide an easy way to assess an individual's risk through noninvasive methods. Numerous diabetes risk models have been developed in various countries. Notable risk models developed among whites are FINDRISC, AUSDRISK, the Danish Risk Score, the American Diabetes Association questionnaire, and DESIR. Some of the risk scores developed among Asian population are IDRS, the Chaturvedi score, and the Thai risk score. There is conflicting evidence on the performance of these scores in different populations. A study conducted in India showed that the area under the curve was overlapping for white and Asian DRS, indicating similar diagnostic efficiency. 11 Another study in the United States assessing the performance of FINDRISC showed favorable results. 12 However, in a study conducted among Omani Arabs, the Thai, Dutch, Finnish, and Danish DRS showed poor performance. 13 The existing DRS for Middle East countries such as the Oman risk score are developed from a large representative sample and should work well in the Saudi population, but the Kuwait risk score was developed from a small sample size of government employees, and its performance would be questionable. 13,14
The study by Memish et al. 15 with a focus of developing a risk score for Saudi Arabians has an adequately large representative sample. Appropriate sampling techniques, data cleaning, quality check, and statistical analysis make this tool reliable. The decision to fix a cutoff with higher sensitivity will also reduce the number of true-positives not getting confirmatory testing, and this is essential as DM complications begin early. The downside of this is it will increase the number of false-positive cases, thereby increasing the burden on healthcare systems and which also will make the people skeptical about the usefulness of the tool. A study by Basu et al. 16 on the Indian risk score has shown that large-scale community-based screening will lead to a large number of false-positive cases, and the number needed to screen value to get one newly diagnosed case of diabetes will range from 15.2 to 33.6, depending on the scale used. The number needed to screen is affected by the sensitivity and specificity of the tool used and the prevalence of the disease in the community. Because the risk score of Memish et al. 15 has a similar diagnostic efficiency as that of the Indian DRS and there is a slightly higher prevalence of diabetes in the Middle East, the number needed to screen would be similar or slightly decreased for Middle East countries.
In order to improve the diagnostic efficiency, multiple screening tests can be used either simultaneously or sequentially. Table 1 shows the net sensitivity and specificity for the score developed by Memish et al. 15 with casual blood sugar and urine sugar testing. Simultaneous testing would be the best approach for reducing the number of false-positives qualified for glucose tolerance testing but will increase the cost. On the other hand, sequential testing with urine or blood sugar for those classified as high risk will be cost-effective, but it will lead to large numbers of true-positives undetected, thereby not receiving a confirmatory test. The choice should be made after taking into account the morbidity and cost incurred by diabetes, the capacity of the health system to cope with increasing number of newly identified patients, and the percentage of GDP allocated to health care.
Median sensitivity for blood and urine of 81% and 43%, respectively, and specificity of 92% and 99%, respectively, were used to compute net sensitivity and specificity (using World Health Organization criteria for type 2 diabetes mellitus).
Large numbers of interventional studies have been done on those with prediabetes and other high-risk groups, targeting the prevention or delaying the progress toward T2DM, and there is substantial favorable evidence on such intervention. 17 Although case finding is important for the prediabetes state, there is no need to develop a separate risk model predicting the prediabetes state, as the risk factors are common. A separate cutoff can be tried in the DRS to predict prediabetes. It will also be interesting to study the changes in the risk score and development of T2DM. As all the noncommunicable diseases share common risk factors, these scales can also be useful in identifying the risk for other diseases, but more prospective research is required to validate the cutoffs. Smoking as a variable among diabetes risk factors is common among the white DRS; this may be due to the high prevalence of smoking in this group's makeup, and this is also a relevant variable for the prediction of cardiovascular diseases.
Finally, the use of such a scale in community-based screening should lead to decreases in the incidence of DM, complications of DM, and mortality and an increase in the quality of life. The impact of only a few DRS (FINDRISC and AUSD-RISK) has been reported, and there is a favorable shift in the risk factor level. 18 Impact studies will be necessary to determine the success in risk score development measured in terms of patient-relevant intermediate outcomes and final outcomes rather than in terms of the statistical properties of the tool; a qualitative component explores both facilitators and barriers of using the score in a real-world setting. 18,19
Footnotes
Author Disclosure Statement
No competing financial interests exist.