RSS Feeds
-
Integer-points based scoring systems for diabetes prediction: The German Diabetes Risk Score
Submit responseKahn et al. (1) report two new scoring systems to predict future type 2 diabetes based on data from the ARIC study. Interestingly, one of the prediction models does not require any data from blood samples. The rationale for this model is the often ignored lack of such data in populations. The low availability of fasting blood glucose measurements in the US (2) is also observed in other Westernized countries:In Germany, participation in routine health check-ups including measurement of fasting glucose is generally low (21% of those eligible in 2001), although they are available free of charge for adults older than 34 years (3). Thus, scoring systems depending on metabolic markers will have low population coverage, and reported characteristics of such models are probably overestimating their performance in the general populations under “real world” conditions. A two- or multi-step screening process including a basic prediction model, and subsequent measurement of metabolic markers in high-risk individuals identified in the initial screen, could largely reduce the number of blood tests. Such a process could minimize false-positive tests, and allow a higher overall sensitivity than the suggested 60% (2) at acceptable costs. It would be quite informative to evaluate the ARIC study data accordingly.
Kahn et al. (1) also summarize other diabetes prediction models which allocate integer points to categorized risk factors. Unfortunately, this summary does not include the German Diabetes Risk Score (4) for which a more simplified score computation using integer points has been developed previously (Table 1) (5). The integer-point based score correlates well with the original score (r=0.97), and provides similar performance characteristics (ROC-AUC: 0.83). Addition of fasting blood glucose, HbA1c, triglycerides, HDL-cholesterol, and liver enzymes markedly enhances performance (Schulze et al., submitted for publication).Although a direct comparison of these results with those reported from the ARIC study (1) may not be possible, it would be interesting to evaluate whether a more refined categorization of risk factors, particularly of age and anthropometric characteristics (as has been done for the German Diabetes Risk Score), would enhance performance of the ARIC scoring systems.
The rationale to concentrate on simple and parsimonious models (1, 2) is understandable. However, current technology (e.g. in computerized calculation or in WebTools) is capable to easily calculate more complicated risk scores. Thus, we believe that more emphasis should in future be put on the accuracy of risk prediction than on the simplicity of calculation.
References
1. Kahn HS, Cheng YJ, Thompson TJ, Imperatore G, Gregg EW. Two risk-scoring systems for predicting incident diabetes mellitus in U.S. adults age 45 to 64 years. Ann Intern Med 2009;150:741-51
2. Herman WH. Predicting risk for diabetes: choosing (or building) the right model. Ann Intern Med 2009;150:812-4
3. Robert Koch-Institut. Gesundheitsberichterstattung des Bundes. Heft 24. 2005
4. Schulze MB, Hoffmann K, Boeing H, Linseisen J, Rohrmann S, Möhlig M, Pfeiffer AF, Spranger J, Thamer C, Häring HU, Fritsche A, Joost HG. An accurate risk score based on anthropometric, dietary, and lifestyle factors to predict the development of type 2 diabetes. Diabetes Care 2007;30:510-5
5. Schulze MB, Holmberg C, Hoffmann K, Boeing H, Joost HG. Brief questionnaire to determine the risk of diabetes according to the German diabetes risk score. Ernährungs Umschau 2007;54:698–703
Table 1. Risk factors and allocated points, German Diabetes Risk Score
Risk factor
Points
Risk factor
Points
Age (years)
Smoking status
<35
0
Never
0
35–39
1
Formerly, <20 cigarettes/d
0
40–44
3
Formerly, >=20 cigarettes/d
3
45–49
5
Current, <20 cigarettes/d
0
50–54
7
Current, >=20 cigarettes/d
6
55–59
9
60–64
11
Alcohol consumption
65–70
13
Never or rarely
2
1-4 glasses/d
0
Waist circumference (cm)
<4 glasses/d
2
<75
0
75–79
4
Physical activity
80–84
8
<5 h/wk
1
85–89
12
>=5 h/wk
0
90–94
16
95–99
20
Coffee consumption
100–104
24
0-1 cup/d
2
105–109
28
2-5 cups/d
1
110–114
32
>5 cups/d
0
115–119
36
>=120
40
Whole-grain cereal consumption (1 portion ≈ 50g)
0 portions/d
5
Height (cm)
1 portion/d
4
<152
11
2 portions/d
3
152–159
9
3 portions/d
2
160–167
7
4 portions/d
1
168–175
5
>4 portions/d
0
176–183
3
184–191
1
Red meat consumtion
≥192
0
Never or rarely
0
1-2 times /wk
1
History of hypertension
3-4 times /wk
2
No
0
5-6 times /wk
4
yes
5
Once Daily
5
> once daily
8
To calculate the probability to develop type 2 diabetes the following equation can be used:
5-year diabetes probability = 1 – 0.999854 exp (points/10 + 1.1)
Conflict of Interest:
None declared
-
Pulse Mass Index for Predicting Diabetes and Cardiovascular Risk
Submit responsePulse Mass Index for Predicting Diabetes, Cardiovascular Risk and effects of anti diabetic drugs
Henry S. Kahn et al developed an scoring system for predicting Diabetes mellitus which includes, among other parameters, stature or height, increased weight and rapid pulse. These are the component of both, the Body Mass Index and the Pulse Mass Index.
I reported ten years ago in Lancet March 13, 1999, that the Pulse Mass Index (Resting Heart Rate multiplied by the Body Mass Index and divided by 1730) has a very high and direct correlation with the global cardiovascular risk as calculated by the Framingham Risk Score, and that persons with a Pulse Mass Index of 1.3 or over, have a high probability to be at high global cardiovascular risk.
Now, in this issue of Annals (June 2, 2009), the study by Kahn et al demonstrates that the elements of the Pulse Mass Index are also useful for predicting Diabetes mellitus.
In this issue I also comment that cardiovascular or metabolic drugs that reduce or not increase the Pulse Mass Index, like Beta blockers or metformin, tend to improve the long term cardiovascular prognosis, contrary to drugs like rapid acting vasodilators that increase pulse rate, retain water and do not reduce mortality, or like glitazones that increase weight or intensive glycemic control with glitazones and insulin, that do not improve the long term cardiovascular prognosis when associated with weight gain or hypoglycaemia (and related tachycardia), so that the Pulse Mass Index can also be of help to predict the potential benefits or adverse reactions of cardiovascular and anti diabetic drugs.
Prof. Enrique Sánchez-Delgado MD
Internist-Clinical Pharmacologist
Director of Medical Education
Hospital Metropolitano Vivian Pellas
Managua, Nicaragua
Conflict of Interest:
None declared