- Email: [email protected]
Short-term reproducibility of impaired fasting glycaemia, impaired glucose tolerance and diabetes
diabetes research and clinical practice 80 (2008) 146–152
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/diabres
Short-term reproducibility of impaired fasting glycaemia, impaired glucose tolerance and diabetes The ADDITION study, DK S.S. Rasmussen a,*, C. Glu¨mer b, A. Sandbaek c, T. Lauritzen c, B. Carstensen a, K. Borch-Johnsen a,c a
Steno Diabetes Center, Niels Steensensvej 2, DK-2820 Gentofte, Denmark Research Centre for Prevention and Health, Glostrup, Denmark c Institution of Public Health, Department of General Practice, University of Aarhus, Denmark b
article info
abstract
Article history:
We evaluated variations in glucose measurements and the reproducibility of glucose
Received 27 August 2007
tolerance classification in a high-risk screening setting in general practice.
Accepted 4 November 2007 Published on line 21 December 2007
Screening for diabetes was performed in persons aged 40–69 years. Based on capillary fasting (FBG) and 2-h blood glucose (2 hBG) individuals with impaired fasting glycaemia (IFG), impaired glucose tolerance (IGT) and diabetes had a second test done after 14 days.
Keywords:
Intra-individual coefficients of variation (CV) were estimated in each glucose tolerance class
Capillary blood glucose
using the approximation CV2(x) = var(ln(x)). Bland–Altman plots with limits of agreement
Type 2 diabetes
were made.
Impaired fasting glycaemia
In the total population, the CVintra was 7.9% and 13.8% for FBG and 2 hBG, respectively.
Impaired glucose tolerance
Limits of agreement ranged from 1.15 to 1.67 mmol/l for FBG and from – 2.62 to 3.27 mmol/l
Intra-individual variation
for 2 hBG. One individual with IFG and 22.5% with IGT had diabetes at the second test, 76.1% with diabetes had this diagnosis confirmed, and about 30% with IFG and IGT had normal glucose tolerance at the second test. The expected values of repeated capillary blood glucose measurements were about 1 and 3 mmol/l for FBG and 2 hBG, respectively. Yet, 70% of high-risk prediabetic individuals were persistently classified with abnormal glucose regulation; diabetes was confirmed in 76% of the cases. # 2007 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Since 1999, the glucose tolerance classification by the World Health Organization (WHO) includes impaired fasting glycaemia (IFG) as well as impaired glucose tolerance (IGT) defined by fasting glucose and 2-h glucose after an oral glucose tolerance test (OGTT) [1]. The diagnosis of diabetes in asymptomatic
individuals requires two diabetic test results on separate days. This is not required for IFG or IGT. It is known though, that there are considerable intraindividual variations in fasting glucose and after an OGTT; giving rise to misclassification in the abnormal glucose tolerance groups [2–12]. Most studies, though, have used long time intervals between tests and therefore do not represent
* Corresponding author. Tel.: +45 44 43 90 77; fax: +45 44 43 07 06. E-mail address: [email protected] (S.S. Rasmussen). 0168-8227/$ – see front matter # 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.11.003
diabetes research and clinical practice 80 (2008) 146–152
pure reproducibility. They also reflect metabolic progressions and interventions taking place in the time interval between the tests. Pure reproducibility has to be examined by tests within a short-time interval where progression in glucose intolerance is unlikely to occur. Few studies have evaluated the reproducibility of glucose measurements by two repeated standard OGTTs within a short-time frame (about 2 weeks) [7,9,10,12]. These studies took place in centralised centres or at hospitals and did not evaluate the glucose tolerance categories including IFG. The risk of misclassification of IFG and IGT is greater than of diabetes because IFG and IGT represent intervals of glucose levels while diabetes has an upper ‘‘open end’’. In particular, is IFG based on an interval covering a range of only 0.5 mmol/l (blood glucose) or 0.8 mmol/l (plasma glucose) which may be exceeded by the intra-individual variation. Our aim was to evaluate the Short-term intra-individual variation of fasting and 2-h capillary whole blood glucose by repeating the tests within 14 days for individuals with IFG, IGT and diabetes, and to evaluate the magnitude of misclassification in a high-risk screened population in general practice.
2.
Subjects and methods
2.1.
Study population and design
The study population is based on the ADDITION study, DK, which is a population-based, high-risk screening and inter-
147
vention study for type 2 diabetes in general practice [13,14]. In the present substudy, based on an opportunistic screening approach (Fig. 1), a risk score questionnaire was handed out to patients aged 40–69 years, having an appointment at the participating general practices. This screening step was followed by the diagnostic test, fasting capillary blood glucose (FBG), on another day in individuals who reported high scores. At this second visit, a blood sample for HbA1c measurement was taken and mailed to the central laboratory. An OGTT was performed and the 2-h capillary blood glucose (2 hBG) measured within the same consultation if FBG was elevated or in a subsequent consultation if the FBG was normal but HbA1c 5.8%. All individuals identified with IFG, IGT or diabetes had a confirmatory test done within approximately 14 days and 2 hBG was measured at each visit if FBG was not in the diabetic range at the respective visit. After the first diagnostic visit, the persons were informed of their glucose tolerance status as they were explained that a second test on another day was indicated. They were instructed to be fasting at both the diagnostic tests. There were no restrictions on diet or physical activity until fasting and no specific instruction was given on lifestyle changes before the second test. Because of the high-risk screening approach, individuals considered at low risk at the initial steps did not proceed for the diagnostic tests. Individuals with normal glucose tolerance (NGT) were therefore to a large extend not identified. Only few had normal glucose tolerance at the diagnostic step and it was impractical to have these few persons recruited for a confirmatory test.
Fig. 1 – The modified screening algorithm in the ADDITION study, DK – the opportunistic screening substudy.
148
diabetes research and clinical practice 80 (2008) 146–152
All participants gave written informed consent. The study was in accordance with the Declaration of Helsinki and approved by the scientific ethics committee of Aarhus, Denmark.
2.2.
coefficient of variation (CVbio) was determined as the variation left after excluding the measurement variation. Estimates of linear dependence and CVs were obtained by an implementation of the model from Carstensen [18] in WinBUGS, using the R2WinBUGS interface [19].
Definitions and measurements 2.3.3.
Glucose tolerance was classified by the WHO1999 definition using capillary FBG and 2 hBG before and after a standard OGTT: impaired fasting glycaemia (5.6 mmol/l FBG < 6.1 mmol/l and 2 hBG < 7.8 mmol/l); impaired glucose tolerance (FBG < 6.1 mmol/l and 7.8 mmol/l 2 hBG < 11.1 mmol/l) and diabetes (FBG 6.1 mmol/l or 2 hBG 11.1 mmol/l) [1]. In the case of a diabetic value of the FBG, an oral glucose load was not performed, which is in full accordance with the WHO recommended diagnostic procedure. Capillary whole blood was analysed with a HemoCue Bglucose analyser based on glucose dehydrogenase reaction ¨ ngelholm, Sweden). Analyses of plasma (HemoCue AB, A glucose would have required immediate spinning and cooling of the blood samples and transport to the laboratory, procedures which are seldom available in general practice. Two replicate capillary blood samples with 1 min intervals were taken and the average of the two results was used for glucose tolerance classification [15]. All practices were trained by the HemoCue Company in analysing the blood samples and the daily calibration of the machines [13]. Blood glucose was measured after fasting from the evening before; persons instructed not to eat, drink or smoke. The persons stayed in the consultation during the 75-g OGTT and after 2 h the blood glucose was measured. HbA1c was analysed using liquid chromatography on a Tosoh machine (TOSOH A1c 2.2, TOSOH/ Eurogenetics, Germany; normal range 4.2–6.3%) on venous plasma. Information on age, sex and BMI categories were drawn from the self-administered risk score which is described elsewhere [13].
2.3.
Statistical analyses
Analyses were based on two replicate measurements at each test-day (visit) for individuals classified with IFG, IGT and diabetes.
2.3.1.
Differences in blood glucose
The mean difference and the 95% CI in FBG and 2 hBG between the two visits were calculated. Limits of agreement were calculated to represent prediction limits for differences between individual measurements at the two visits. Bland– Altman plots [16,17] were made to visually assess whether difference and variance was constant across the range of measurements. The relationship of the difference to the mean of the two visits was estimated in a model linking the second visit to the first by a linear relation [18].
2.3.2.
CVintra
The intra-individual coefficient of variation (CVintra) was determined as the standard deviation of the natural logarithm of the measurements after controlling for the individual level of measurement and visit: CV2(x) = var(ln(x)). The biological
Reclassification
Each individual was classified based on the mean of the two replicate measurements at each visit according to the WHO1999 definition [1]. The proportions of individuals with IFG and IGT on the first test and diabetes on the second test and the 95% CI were calculated. The proportion of individuals with diabetes on both tests (confirmed diabetes) was also calculated.
3.
Results
From October 2005 to October 2006, 923 individuals with highrisk scores had the first diagnostic tests done. Of those, we excluded four with incomplete tests and one with previously known diabetes. We identified 48 individuals with IFG, 43 with IGT, 74 with diabetes and 753 with NGT. Based on this first classification the number of individuals who attended the confirmatory tests was 42 for IFG, 40 for IGT, 71 for diabetes. These individuals were included in the present analyses where 75% were in the age group of 55–69 years, 39% had a BMI between 24 and 31 kg/m2 and 43% had a BMI above 31 kg/m2. There were 49% women. The median time interval between first and second diagnostic visit was 14.0 days (inter quartile range: 9–24). The mean differences in glucose results are presented in Table 1 showing a significantly lower FBG in all categories at the second visit. Regarding the 2 hBG, this was the case in the total population and the IGT group. There was no statistically significant difference in 2 hBG between the visits for IFG and diabetes, where 2 hBG as a consequence of the definition only was measured at both visits in a low number of cases. The CVintra for the total population was 7.9% and 13.8% for FBG and 2 hBG, respectively, with insignificant differences between the IFG and IGT categories (Table 1). In the more extreme category, the diabetes group, where 80% had only a FBG measured, the CVintra for FBG was significantly greater than in IFG and IGT (9.6% versus 5.8% and 6.6%). Generally, the CVintra was only marginally larger than the CVbio. The Bland–Altman plots (Fig. 2) show that the glucose values at the first visit were generally higher than those at the second visit, but there was an insignificant tendency that this difference was smaller for higher values of FBG. The estimated linear relationships between the first and second visits were: FBG2 ¼ 0:67 þ 1:07 FBG1 ð95% CI for slope : 0:74 1:56Þðmmol=lÞ
2hBG2 ¼ 0:19 þ 0:97 2hBG1 ð95% CI for slope : 0:81 1:14Þðmmol=lÞ Data were too few to provide reliable results of the slopes for the individual glucose tolerance groups.
149
diabetes research and clinical practice 80 (2008) 146–152
Table 1 – Short-term variation [mean or percent (95% CI)] of fasting and 2-h capillary whole blood glucose in the IFG, IGT and diabetes categories (WHO1999 [1]) Glucose tolerance category at first visit
N
Mean (visit1–visit2) (mmol/l)
Fasting blood glucose Total population IFG IGT Diabetes
153 42 40 71
0.26 0.22 0.15 0.34
2-h blood glucose Total population IFG IGT Diabetes
153 42 40 13
0.33 0.01 0.97 0.50
Large variations of the differences were found with limits of agreement ranging from 1.15 to1.67 mmol/l for FBG and from 2.62 to 3.27 mmol/l for 2 hBG in the total population. There was a tendency to wider limits of agreement from IFG over IGT to diabetes (most pronounced for the FBG values).
(0.15–0.37) (0.11–0.34) (0.01–0.29) (0.13–0.56)
(0.02–1.47) (0.35–0.37) (0.47–1.47) (1.51–0.51)
CVintra (%)
CVbio (%)
7.9 5.8 6.6 9.6
(7.2–8.8) (4.9–7.1) (5.4–8.1) (8.3–11.5)
7.0 4.2 5.2 9.0
(6.2–8.0) (3.0–5.7) (3.7–6.9) (7.6–10.9)
13.8 13.6 13.9 10.6
(12.0–16.1) (11.3–17.0) (11.1–17.7) (7.4–17.5)
13.4 12.9 13.6 10.4
(11.6–15.8) (10.4–16.4) (10.8–17.5) (7.1–17.4)
Table 2 summarises the glucose tolerance classification at the second visit by the initial classification. About a third with initially IFG had NGT on the second visit. This proportion was reduced over IGT to diabetes, where it was less than 10%. One individual with initially IFG had diabetes at the second visit
Fig. 2 – For each glucose tolerance group – IFG, IGT and diabetes (DM) – are Bland–Altman plots shown. Fasting blood glucose (FBG) at first visit – FBG at second visit (mmol/l) is plotted versus the average of FBG at first and second visit, and likewise for 2-h blood glucose (2 hBG). The middle solid line represents the mean of (FBG at first visit – FBG at second visit) and the mean of (2 hBG at first visit – 2 hBG at second visit), respectively. The other solid lines represent limits of agreements. Dashed lines are 95% confidence intervals.
150
diabetes research and clinical practice 80 (2008) 146–152
Table 2 – Glucose tolerance classification at first and second visit; [n (row percent)] Visit1
Visit2 NGT
IFG
IGT
Diabetes Total
IFG IGT Diabetes
15 (35.7) 9 (22.5) 6 (8.5)
20 (47.6) 6 (15.0) 4 (5.6)
6 (14.3) 16 (40.0) 7 (9.9)
1 (2.4) 9 (22.5) 54 (76.1)
42 40 71
Total
30 (19.6)
30 (19.6)
29 (19.0)
64 (41.8)
153
(2.4% (95% CI: 0.1–12.6)). Of individuals with IGT 9 (22.5% (10.8– 38.5)) had diabetes at the second visit. Diabetes at first visit was confirmed in 54 cases (76.1% (64.5–85.4)).
blood specimens [20]. However, in clinical and public health context, the notions are used irrespective of the measurement method. We evaluated the reproducibility of the glucose tolerance classification including IFG as well as IGT and diabetes. It would have been nice to evaluate the NGT category, too, which we (due to the study design) were unable to do. To our knowledge, no evaluation of the IFG group has yet been published. We would like to stress that the statistically correct method to calculate CVintra is to take the individual means into account. In the literature, it is often calculated as (S.D.dif/H2) or S.D. divided by the population mean or median of the individual average.
4.2.
4.
In the present study, the CVintras were 7.9% and 13.8% for capillary FBG and capillary 2 hBG, respectively, in the total population; calculated correctly to reflect the true CVintra. These variations in FBG and especially in 2 hBG can be understood by the limits of agreement. For a given FBG measured in an individual, a new measurement on a second day will be within 1 mmol/l, and for a given 2 hBG within 3 mmol/l. Nevertheless, we found that 76.1% of the individuals initially classified with diabetes had this diagnosis confirmed, and 64.3% and 77.5% of those with IFG and IGT, respectively, were still classified with abnormal glucose tolerance at the second visit.
4.1.
Comparison with existing literature
Discussion
Strength and limitations of this study
A strength of our study is that we evaluate and provide estimates on reproducibility of glucose measurements taking place in general practice under standardised, yet not too cumbersome, every-day life conditions using an instant measurement method. Previous studies indicate that intraindividual variations of glucose measurements in general practice or in outpatients are greater than in laboratory workers or inpatients; maybe due to lesser control of the fasting state [4,8]. Patients in our study were fasting from the day before testing. There were no restrictions on diet or physical activity until fasting. After the first visit, the persons were informed of their glucose tolerance status as they were explained that a second test on another day was indicated. It may have affected their diet and physical activity level between the two visits and thus affected their fasting state at the second visit. We were unable to control for this, but a previous study indicates that changes in lifestyle habits do not explain the difference between first and second test; at least in a short timeframe [12]. The estimated linear relationship for FBG and 2 hBG between the two tests in our study indicate insignificant trends for the difference between the measurements to be smaller at higher glucose values. Weaknesses of the glucose tolerance definition itself such as rounding errors and poor data basis may cause concerns as to whether the same groups of individuals are being compared when glucose tolerance categories are based on different
A Tanzanian study showed lower blood glucose levels at the repeated test, and interpreted it as a consequence of anxiety due to the unknown test situation at the first test, because the phenomenon occurred similar to an acclimatising reflex observed in blood pressure [9]. This was not confirmed in an European population [10]. We also found lower mean glucose values at the second visit. As the tests were performed under familiar conditions at the general practitioners, it may be due to regression to the mean. Possibly because the IFG group was not selected on high 2 hBG its mean 2 hBG did not differ significantly between tests. Too few 2 hBG were measured in the diabetes group to identify an effect. Calculating CVintra without taking the individual means into account provides results which may not be correct and the direction or magnitude of the error cannot be predicted. Had we calculated the CVintras as it was calculated in the Hoorn study [10] the results for the total population would have been 8.3% and 13.5% for FBG and 2 hBG, respectively. It is by chance that using the more correct calculation we find values of same order as already acknowledged: Two Shortterm studies in mainly normal glycaemic persons based on blood or plasma glucose found CVintras for FBG to be 5.5–6% and for 2 hBG to be 13.6–15% [7,11]. The Hoorn study, which made separate analyses for NGT, IGT and diabetes, found CVintras of IGT to be 6.1% and 15.0%, of diabetes to be 7.1% and 12.7% [10]. Studies covering a broader time span showed CVintra for fasting glucose of 4.7–12.0% and for 2-h glucose of 11.0–17.7% [5,8]. Intra-individual variation in measurements of blood constituents is the sum of measurement error and ‘‘true’’ biological variation. Glucose values measured in capillary blood vary about twice as much as when measured in plasma or venous whole blood [20]. However, measurement errors (CVintra CVbio) in the present study were negligible in relation to the intra-individual variation (Table 1). Of clinical relevance in screening for type 2 diabetes is the proportion of confirmed diabetes, which in this study was very satisfactory, and few (8.5%) had NGT on the confirmatory tests. Only one person with IFG had diabetes at the second visit but 64.3% were still classified with abnormal glucose tolerance. For the IGT group, 22.5% had diabetes on the confirmatory tests, which is more than found in the Tanzanian and Hoorn studies (4.5–12.6%) and only 22.5% reverted to NGT in contrast
diabetes research and clinical practice 80 (2008) 146–152
151
to 39.0–65.9% [9,10]. These findings reflect the high-risk profile of individuals in our study.
references
4.3.
[1] World Health Organization, Definition, Diagnosis and Classification of Diabetes Mellitus and its Complications. Part 1: Diagnosis and Classification of Diabetes Mellitus. Report of a WHO Consultation, Geneva, 1999. [2] K.M. West, J.A. Wulff, D.G. Reigel, D.T. Fitzgerald, Oral carbohydrate tolerance tests, Arch. Intern. Med. 113 (1964) 641–648. [3] O.P. Ganda, J.L. Day, J.S. Soeldner, J.J. Connon, R.E. Gleason, Reproducibility and comparative analysis of repeated intravenous and oral glucose tolerance tests, Diabetes 27 (1978) 715–725. [4] I.F. Godsland, Intra-individual variation: significant changes in parameters of lipid and carbohydrate metabolism in the individual and intra-individual variation in different test populations, Ann. Clin. Biochem. 22 (1985) 618–624. [5] S.T. Cummings, C.G. Fraser, Variability of capillary plasma glucose in healthy individuals in repeated 75 g oral glucose tolerance tests, Ann. Clin. Biochem. 25 (1988) 634–637. [6] P. Home, The OGTT: gold that does not shine, Diab. Med. 5 (1988) 313–314. [7] K. Schousboe, J.E. Henriksen, K.O. Kyvik, T.I. Sorensen, P.P. Hyltoft, Reproducibility of S-insulin and B-glucose responses in two identical oral glucose tolerance tests, Scand. J. Clin. Lab. Invest. 62 (2002) 623–630. [8] E.J. Feskens, C.H. Bowles, D. Kromhout, Intra- and interindividual variability of glucose tolerance in an elderly population, J. Clin. Epidemiol. 44 (1991) 947–953. [9] A.B. Swai, D.G. McLarty, H.M. Kitange, P.M. Kilima, G. Masuki, B.I. Mtinangi, et al., Study in Tanzania of impaired glucose tolerance. Methodological myth? Diabetes 40 (1991) 516–520. [10] J.M. Mooy, P.A. Grootenhuis, H. de Vries, P.J. Kostense, C. Popp-Snijders, L.M. Bouter, et al., Intra-individual variation of glucose, specific insulin and proinsulin concentrations measured by two oral glucose tolerance tests in a general Caucasian population: the Hoorn Study, Diabetologia 39 (1996) 298–305. [11] J.L. Sievenpiper, L.A. Leiter, V. Vuksan, Intrasubject coefficient of variation corresponds to diagnostic reproducibility in diabetes screening, Can. J. Diab. 26 (2002) 105–112. [12] G. Brohall, C.J. Behre, J. Hulthe, J. Wikstrand, B. Fagerberg, Prevalence of diabetes and impaired glucose tolerance in 64-year-old Swedish women: experiences of using repeated oral glucose tolerance tests, Diab. Care 29 (2006) 363–367. [13] J.O. Christensen, A. Sandbaek, T. Lauritzen, K. BorchJohnsen, Population-based stepwise screening for unrecognised Type 2 diabetes is ineffective in general practice despite reliable algorithms, Diabetologia 47 (2004) 1566–1573. [14] T. Lauritzen, S. Griffin, K. Borch-Johnsen, N.J. Wareham, B.H. Wolffenbuttel, G. Rutten, The ADDITION study: proposed trial of the cost-effectiveness of an intensive multifactorial intervention on morbidity and mortality among people with Type 2 diabetes detected by screening, Int. J. Obestet. Relat. Metab. Disord. 24 (Suppl. 3) (2000) S6–S11. [15] A. Sandbaek, T. Lauritzen, K. Borch-Johnsen, K. Mai, J.S. Christiansen, The comparison of venous plasma glucose and whole blood capillary glucose in diagnoses of Type 2 diabetes: a population-based screening study, Diab. Med. 22 (2005) 1173–1177. [16] J.M. Bland, D.G. Altman, Statistical methods for assessing agreement between two methods of clinical measurement, Lancet 1 (1986) 307–310.
Implications for clinical practice
Most screening for diabetes take place in general practice, where health care providers meet high-risk individuals. Evidence on reproducibility of glucose measurements and hence glucose tolerance classification, upon which treatment decisions are based, is important in this setting. The present study indicates that using capillary whole blood is an attractive option compared to venous whole blood or plasma in screening strategies for diabetes. Despite quantified variations from day to day in blood glucose, we found the measurement results robust with respect to classification of abnormal glucose tolerance in high-risk individuals. About 70% with IFG or IGT were still classified with abnormal glucose regulation in a repeated test and 76% had two results in the diabetic range on separate days. The criterion of two results in the diabetic range on separate days for the diagnosis of diabetes reduces the risk of disease labelling persons who by chance had a test result in the diabetic range. The disease labelling possesses numerous implications to both the individual and the society (insurances, intensive care and treatment). We would not suggest to abandon this criterion of confirmation of the diagnosis of diabetes in asymptomatic persons. Those with one test result in the diabetic range, regardless on first or second test day, have abnormal glucose regulation and not the diagnosis of diabetes. Annual glucose control in individuals with abnormal glucose regulation identified by high-risk strategies is recommendable as the risk of developing diabetes in these individuals is high [21]. It is known that lifestyle intervention at least in individuals with IGT reduces the risk of developing diabetes by up to 58% [22–25]. It seems adequate to initiate lifestyle intervention, which has no side effects, in individuals with IFG or IGT based on one test result when identified by high-risk strategies because of the robustness of the classification shown. We suggest that this also applies to individuals in whom a diabetic glucose value is not confirmed by the second test
Acknowledgements This fully investigator initiated, designed and controlled study received funding from: the Danish Centre for Evaluation and Health Technology Assessment; the Danish Research Foundation for General Practice; the counties of Aarhus, Copenhagen, Ringkoebing, Ribe, and South Jutland; the National Board of Health; the Danish Medical Research Council no. 22-04-0390; the Danish Diabetes Association; the A.P. Møller Foundation; the Bernhard and Marie Kleins Trust; Novo Nordisk A/S and Novo Nordisk Scandinavia AB; Astra-Zeneca; Phizer; Servier; GlaxoSmithKline; and HemoCue.
Conflict of interest The authors state that they have no conflict of interest.
152
diabetes research and clinical practice 80 (2008) 146–152
[17] J.M. Bland, D.G. Altman, Measuring agreement in method comparison studies, Stat. Methods Med. Res. 8 (1999) 135– 160. [18] B. Carstensen, Comparing and predicting between several methods of measurement, Biostatistics 5 (2004) 399–413. [19] S. Sturtz, U. Ligges, A. Gelman, R2WinBUGS: Running WinBUGS and OpenBUGS from R. 2006 http:// mirrors.dotsrc.org/cran/doc/vignettes/R2WinBUGS/ R2WinBUGS.pdf accessed 1st February 2007. [20] B. Carstensen, J. Lindstrom, J. Sundvall, K. Borch-Johnsen, J. Tuomilehto, and the DPS study group. Measurement of blood glucose: comparison between different types of specimens, Ann. Clin. Biochem., in press. [21] A.G. Rassam, J. Mcleod, M.R. Burge, D.S. Schade, Use of the HemoCue blood glucose analyzer in research studies, Diab. Care 21 (1998) 1369–1370. [22] J.L. Chiasson, R.G. Josse, R. Gomis, M. Hanefeld, A. Karasik, M. Laakso, STOP-NIDDM Trial REsearch Group, Acarbose
for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial, Lancet 359 (2002) 2072–2077. [23] W.C. Knowler, E. Barrett-Connor, E.S.E. Fowler, R.F. Hamman, J.M. Lachin, E.A. Walker, et al., Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin, N. Engl. J. Med. 346 (2002) 393–403. [24] X.R. Pan, G.W. Li, Y.H. Hu, J.X. Wang, W.Y. Yang, Z.X. An, et al., Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study, Diab. Care 20 (1997) 537–544. [25] J. Tuomilehto, J. Lindstrom, J.G. Eriksson, T.T. Valle, H. Hamalainen, P. Ilanne-Parikka, et al.,Finnish Diabetes Prevention Study Group, Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance, N. Engl. J. Med. 344 (2001) 1343–1350.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/diabres
Short-term reproducibility of impaired fasting glycaemia, impaired glucose tolerance and diabetes The ADDITION study, DK S.S. Rasmussen a,*, C. Glu¨mer b, A. Sandbaek c, T. Lauritzen c, B. Carstensen a, K. Borch-Johnsen a,c a
Steno Diabetes Center, Niels Steensensvej 2, DK-2820 Gentofte, Denmark Research Centre for Prevention and Health, Glostrup, Denmark c Institution of Public Health, Department of General Practice, University of Aarhus, Denmark b
article info
abstract
Article history:
We evaluated variations in glucose measurements and the reproducibility of glucose
Received 27 August 2007
tolerance classification in a high-risk screening setting in general practice.
Accepted 4 November 2007 Published on line 21 December 2007
Screening for diabetes was performed in persons aged 40–69 years. Based on capillary fasting (FBG) and 2-h blood glucose (2 hBG) individuals with impaired fasting glycaemia (IFG), impaired glucose tolerance (IGT) and diabetes had a second test done after 14 days.
Keywords:
Intra-individual coefficients of variation (CV) were estimated in each glucose tolerance class
Capillary blood glucose
using the approximation CV2(x) = var(ln(x)). Bland–Altman plots with limits of agreement
Type 2 diabetes
were made.
Impaired fasting glycaemia
In the total population, the CVintra was 7.9% and 13.8% for FBG and 2 hBG, respectively.
Impaired glucose tolerance
Limits of agreement ranged from 1.15 to 1.67 mmol/l for FBG and from – 2.62 to 3.27 mmol/l
Intra-individual variation
for 2 hBG. One individual with IFG and 22.5% with IGT had diabetes at the second test, 76.1% with diabetes had this diagnosis confirmed, and about 30% with IFG and IGT had normal glucose tolerance at the second test. The expected values of repeated capillary blood glucose measurements were about 1 and 3 mmol/l for FBG and 2 hBG, respectively. Yet, 70% of high-risk prediabetic individuals were persistently classified with abnormal glucose regulation; diabetes was confirmed in 76% of the cases. # 2007 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Since 1999, the glucose tolerance classification by the World Health Organization (WHO) includes impaired fasting glycaemia (IFG) as well as impaired glucose tolerance (IGT) defined by fasting glucose and 2-h glucose after an oral glucose tolerance test (OGTT) [1]. The diagnosis of diabetes in asymptomatic
individuals requires two diabetic test results on separate days. This is not required for IFG or IGT. It is known though, that there are considerable intraindividual variations in fasting glucose and after an OGTT; giving rise to misclassification in the abnormal glucose tolerance groups [2–12]. Most studies, though, have used long time intervals between tests and therefore do not represent
* Corresponding author. Tel.: +45 44 43 90 77; fax: +45 44 43 07 06. E-mail address: [email protected] (S.S. Rasmussen). 0168-8227/$ – see front matter # 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.11.003
diabetes research and clinical practice 80 (2008) 146–152
pure reproducibility. They also reflect metabolic progressions and interventions taking place in the time interval between the tests. Pure reproducibility has to be examined by tests within a short-time interval where progression in glucose intolerance is unlikely to occur. Few studies have evaluated the reproducibility of glucose measurements by two repeated standard OGTTs within a short-time frame (about 2 weeks) [7,9,10,12]. These studies took place in centralised centres or at hospitals and did not evaluate the glucose tolerance categories including IFG. The risk of misclassification of IFG and IGT is greater than of diabetes because IFG and IGT represent intervals of glucose levels while diabetes has an upper ‘‘open end’’. In particular, is IFG based on an interval covering a range of only 0.5 mmol/l (blood glucose) or 0.8 mmol/l (plasma glucose) which may be exceeded by the intra-individual variation. Our aim was to evaluate the Short-term intra-individual variation of fasting and 2-h capillary whole blood glucose by repeating the tests within 14 days for individuals with IFG, IGT and diabetes, and to evaluate the magnitude of misclassification in a high-risk screened population in general practice.
2.
Subjects and methods
2.1.
Study population and design
The study population is based on the ADDITION study, DK, which is a population-based, high-risk screening and inter-
147
vention study for type 2 diabetes in general practice [13,14]. In the present substudy, based on an opportunistic screening approach (Fig. 1), a risk score questionnaire was handed out to patients aged 40–69 years, having an appointment at the participating general practices. This screening step was followed by the diagnostic test, fasting capillary blood glucose (FBG), on another day in individuals who reported high scores. At this second visit, a blood sample for HbA1c measurement was taken and mailed to the central laboratory. An OGTT was performed and the 2-h capillary blood glucose (2 hBG) measured within the same consultation if FBG was elevated or in a subsequent consultation if the FBG was normal but HbA1c 5.8%. All individuals identified with IFG, IGT or diabetes had a confirmatory test done within approximately 14 days and 2 hBG was measured at each visit if FBG was not in the diabetic range at the respective visit. After the first diagnostic visit, the persons were informed of their glucose tolerance status as they were explained that a second test on another day was indicated. They were instructed to be fasting at both the diagnostic tests. There were no restrictions on diet or physical activity until fasting and no specific instruction was given on lifestyle changes before the second test. Because of the high-risk screening approach, individuals considered at low risk at the initial steps did not proceed for the diagnostic tests. Individuals with normal glucose tolerance (NGT) were therefore to a large extend not identified. Only few had normal glucose tolerance at the diagnostic step and it was impractical to have these few persons recruited for a confirmatory test.
Fig. 1 – The modified screening algorithm in the ADDITION study, DK – the opportunistic screening substudy.
148
diabetes research and clinical practice 80 (2008) 146–152
All participants gave written informed consent. The study was in accordance with the Declaration of Helsinki and approved by the scientific ethics committee of Aarhus, Denmark.
2.2.
coefficient of variation (CVbio) was determined as the variation left after excluding the measurement variation. Estimates of linear dependence and CVs were obtained by an implementation of the model from Carstensen [18] in WinBUGS, using the R2WinBUGS interface [19].
Definitions and measurements 2.3.3.
Glucose tolerance was classified by the WHO1999 definition using capillary FBG and 2 hBG before and after a standard OGTT: impaired fasting glycaemia (5.6 mmol/l FBG < 6.1 mmol/l and 2 hBG < 7.8 mmol/l); impaired glucose tolerance (FBG < 6.1 mmol/l and 7.8 mmol/l 2 hBG < 11.1 mmol/l) and diabetes (FBG 6.1 mmol/l or 2 hBG 11.1 mmol/l) [1]. In the case of a diabetic value of the FBG, an oral glucose load was not performed, which is in full accordance with the WHO recommended diagnostic procedure. Capillary whole blood was analysed with a HemoCue Bglucose analyser based on glucose dehydrogenase reaction ¨ ngelholm, Sweden). Analyses of plasma (HemoCue AB, A glucose would have required immediate spinning and cooling of the blood samples and transport to the laboratory, procedures which are seldom available in general practice. Two replicate capillary blood samples with 1 min intervals were taken and the average of the two results was used for glucose tolerance classification [15]. All practices were trained by the HemoCue Company in analysing the blood samples and the daily calibration of the machines [13]. Blood glucose was measured after fasting from the evening before; persons instructed not to eat, drink or smoke. The persons stayed in the consultation during the 75-g OGTT and after 2 h the blood glucose was measured. HbA1c was analysed using liquid chromatography on a Tosoh machine (TOSOH A1c 2.2, TOSOH/ Eurogenetics, Germany; normal range 4.2–6.3%) on venous plasma. Information on age, sex and BMI categories were drawn from the self-administered risk score which is described elsewhere [13].
2.3.
Statistical analyses
Analyses were based on two replicate measurements at each test-day (visit) for individuals classified with IFG, IGT and diabetes.
2.3.1.
Differences in blood glucose
The mean difference and the 95% CI in FBG and 2 hBG between the two visits were calculated. Limits of agreement were calculated to represent prediction limits for differences between individual measurements at the two visits. Bland– Altman plots [16,17] were made to visually assess whether difference and variance was constant across the range of measurements. The relationship of the difference to the mean of the two visits was estimated in a model linking the second visit to the first by a linear relation [18].
2.3.2.
CVintra
The intra-individual coefficient of variation (CVintra) was determined as the standard deviation of the natural logarithm of the measurements after controlling for the individual level of measurement and visit: CV2(x) = var(ln(x)). The biological
Reclassification
Each individual was classified based on the mean of the two replicate measurements at each visit according to the WHO1999 definition [1]. The proportions of individuals with IFG and IGT on the first test and diabetes on the second test and the 95% CI were calculated. The proportion of individuals with diabetes on both tests (confirmed diabetes) was also calculated.
3.
Results
From October 2005 to October 2006, 923 individuals with highrisk scores had the first diagnostic tests done. Of those, we excluded four with incomplete tests and one with previously known diabetes. We identified 48 individuals with IFG, 43 with IGT, 74 with diabetes and 753 with NGT. Based on this first classification the number of individuals who attended the confirmatory tests was 42 for IFG, 40 for IGT, 71 for diabetes. These individuals were included in the present analyses where 75% were in the age group of 55–69 years, 39% had a BMI between 24 and 31 kg/m2 and 43% had a BMI above 31 kg/m2. There were 49% women. The median time interval between first and second diagnostic visit was 14.0 days (inter quartile range: 9–24). The mean differences in glucose results are presented in Table 1 showing a significantly lower FBG in all categories at the second visit. Regarding the 2 hBG, this was the case in the total population and the IGT group. There was no statistically significant difference in 2 hBG between the visits for IFG and diabetes, where 2 hBG as a consequence of the definition only was measured at both visits in a low number of cases. The CVintra for the total population was 7.9% and 13.8% for FBG and 2 hBG, respectively, with insignificant differences between the IFG and IGT categories (Table 1). In the more extreme category, the diabetes group, where 80% had only a FBG measured, the CVintra for FBG was significantly greater than in IFG and IGT (9.6% versus 5.8% and 6.6%). Generally, the CVintra was only marginally larger than the CVbio. The Bland–Altman plots (Fig. 2) show that the glucose values at the first visit were generally higher than those at the second visit, but there was an insignificant tendency that this difference was smaller for higher values of FBG. The estimated linear relationships between the first and second visits were: FBG2 ¼ 0:67 þ 1:07 FBG1 ð95% CI for slope : 0:74 1:56Þðmmol=lÞ
2hBG2 ¼ 0:19 þ 0:97 2hBG1 ð95% CI for slope : 0:81 1:14Þðmmol=lÞ Data were too few to provide reliable results of the slopes for the individual glucose tolerance groups.
149
diabetes research and clinical practice 80 (2008) 146–152
Table 1 – Short-term variation [mean or percent (95% CI)] of fasting and 2-h capillary whole blood glucose in the IFG, IGT and diabetes categories (WHO1999 [1]) Glucose tolerance category at first visit
N
Mean (visit1–visit2) (mmol/l)
Fasting blood glucose Total population IFG IGT Diabetes
153 42 40 71
0.26 0.22 0.15 0.34
2-h blood glucose Total population IFG IGT Diabetes
153 42 40 13
0.33 0.01 0.97 0.50
Large variations of the differences were found with limits of agreement ranging from 1.15 to1.67 mmol/l for FBG and from 2.62 to 3.27 mmol/l for 2 hBG in the total population. There was a tendency to wider limits of agreement from IFG over IGT to diabetes (most pronounced for the FBG values).
(0.15–0.37) (0.11–0.34) (0.01–0.29) (0.13–0.56)
(0.02–1.47) (0.35–0.37) (0.47–1.47) (1.51–0.51)
CVintra (%)
CVbio (%)
7.9 5.8 6.6 9.6
(7.2–8.8) (4.9–7.1) (5.4–8.1) (8.3–11.5)
7.0 4.2 5.2 9.0
(6.2–8.0) (3.0–5.7) (3.7–6.9) (7.6–10.9)
13.8 13.6 13.9 10.6
(12.0–16.1) (11.3–17.0) (11.1–17.7) (7.4–17.5)
13.4 12.9 13.6 10.4
(11.6–15.8) (10.4–16.4) (10.8–17.5) (7.1–17.4)
Table 2 summarises the glucose tolerance classification at the second visit by the initial classification. About a third with initially IFG had NGT on the second visit. This proportion was reduced over IGT to diabetes, where it was less than 10%. One individual with initially IFG had diabetes at the second visit
Fig. 2 – For each glucose tolerance group – IFG, IGT and diabetes (DM) – are Bland–Altman plots shown. Fasting blood glucose (FBG) at first visit – FBG at second visit (mmol/l) is plotted versus the average of FBG at first and second visit, and likewise for 2-h blood glucose (2 hBG). The middle solid line represents the mean of (FBG at first visit – FBG at second visit) and the mean of (2 hBG at first visit – 2 hBG at second visit), respectively. The other solid lines represent limits of agreements. Dashed lines are 95% confidence intervals.
150
diabetes research and clinical practice 80 (2008) 146–152
Table 2 – Glucose tolerance classification at first and second visit; [n (row percent)] Visit1
Visit2 NGT
IFG
IGT
Diabetes Total
IFG IGT Diabetes
15 (35.7) 9 (22.5) 6 (8.5)
20 (47.6) 6 (15.0) 4 (5.6)
6 (14.3) 16 (40.0) 7 (9.9)
1 (2.4) 9 (22.5) 54 (76.1)
42 40 71
Total
30 (19.6)
30 (19.6)
29 (19.0)
64 (41.8)
153
(2.4% (95% CI: 0.1–12.6)). Of individuals with IGT 9 (22.5% (10.8– 38.5)) had diabetes at the second visit. Diabetes at first visit was confirmed in 54 cases (76.1% (64.5–85.4)).
blood specimens [20]. However, in clinical and public health context, the notions are used irrespective of the measurement method. We evaluated the reproducibility of the glucose tolerance classification including IFG as well as IGT and diabetes. It would have been nice to evaluate the NGT category, too, which we (due to the study design) were unable to do. To our knowledge, no evaluation of the IFG group has yet been published. We would like to stress that the statistically correct method to calculate CVintra is to take the individual means into account. In the literature, it is often calculated as (S.D.dif/H2) or S.D. divided by the population mean or median of the individual average.
4.2.
4.
In the present study, the CVintras were 7.9% and 13.8% for capillary FBG and capillary 2 hBG, respectively, in the total population; calculated correctly to reflect the true CVintra. These variations in FBG and especially in 2 hBG can be understood by the limits of agreement. For a given FBG measured in an individual, a new measurement on a second day will be within 1 mmol/l, and for a given 2 hBG within 3 mmol/l. Nevertheless, we found that 76.1% of the individuals initially classified with diabetes had this diagnosis confirmed, and 64.3% and 77.5% of those with IFG and IGT, respectively, were still classified with abnormal glucose tolerance at the second visit.
4.1.
Comparison with existing literature
Discussion
Strength and limitations of this study
A strength of our study is that we evaluate and provide estimates on reproducibility of glucose measurements taking place in general practice under standardised, yet not too cumbersome, every-day life conditions using an instant measurement method. Previous studies indicate that intraindividual variations of glucose measurements in general practice or in outpatients are greater than in laboratory workers or inpatients; maybe due to lesser control of the fasting state [4,8]. Patients in our study were fasting from the day before testing. There were no restrictions on diet or physical activity until fasting. After the first visit, the persons were informed of their glucose tolerance status as they were explained that a second test on another day was indicated. It may have affected their diet and physical activity level between the two visits and thus affected their fasting state at the second visit. We were unable to control for this, but a previous study indicates that changes in lifestyle habits do not explain the difference between first and second test; at least in a short timeframe [12]. The estimated linear relationship for FBG and 2 hBG between the two tests in our study indicate insignificant trends for the difference between the measurements to be smaller at higher glucose values. Weaknesses of the glucose tolerance definition itself such as rounding errors and poor data basis may cause concerns as to whether the same groups of individuals are being compared when glucose tolerance categories are based on different
A Tanzanian study showed lower blood glucose levels at the repeated test, and interpreted it as a consequence of anxiety due to the unknown test situation at the first test, because the phenomenon occurred similar to an acclimatising reflex observed in blood pressure [9]. This was not confirmed in an European population [10]. We also found lower mean glucose values at the second visit. As the tests were performed under familiar conditions at the general practitioners, it may be due to regression to the mean. Possibly because the IFG group was not selected on high 2 hBG its mean 2 hBG did not differ significantly between tests. Too few 2 hBG were measured in the diabetes group to identify an effect. Calculating CVintra without taking the individual means into account provides results which may not be correct and the direction or magnitude of the error cannot be predicted. Had we calculated the CVintras as it was calculated in the Hoorn study [10] the results for the total population would have been 8.3% and 13.5% for FBG and 2 hBG, respectively. It is by chance that using the more correct calculation we find values of same order as already acknowledged: Two Shortterm studies in mainly normal glycaemic persons based on blood or plasma glucose found CVintras for FBG to be 5.5–6% and for 2 hBG to be 13.6–15% [7,11]. The Hoorn study, which made separate analyses for NGT, IGT and diabetes, found CVintras of IGT to be 6.1% and 15.0%, of diabetes to be 7.1% and 12.7% [10]. Studies covering a broader time span showed CVintra for fasting glucose of 4.7–12.0% and for 2-h glucose of 11.0–17.7% [5,8]. Intra-individual variation in measurements of blood constituents is the sum of measurement error and ‘‘true’’ biological variation. Glucose values measured in capillary blood vary about twice as much as when measured in plasma or venous whole blood [20]. However, measurement errors (CVintra CVbio) in the present study were negligible in relation to the intra-individual variation (Table 1). Of clinical relevance in screening for type 2 diabetes is the proportion of confirmed diabetes, which in this study was very satisfactory, and few (8.5%) had NGT on the confirmatory tests. Only one person with IFG had diabetes at the second visit but 64.3% were still classified with abnormal glucose tolerance. For the IGT group, 22.5% had diabetes on the confirmatory tests, which is more than found in the Tanzanian and Hoorn studies (4.5–12.6%) and only 22.5% reverted to NGT in contrast
diabetes research and clinical practice 80 (2008) 146–152
151
to 39.0–65.9% [9,10]. These findings reflect the high-risk profile of individuals in our study.
references
4.3.
[1] World Health Organization, Definition, Diagnosis and Classification of Diabetes Mellitus and its Complications. Part 1: Diagnosis and Classification of Diabetes Mellitus. Report of a WHO Consultation, Geneva, 1999. [2] K.M. West, J.A. Wulff, D.G. Reigel, D.T. Fitzgerald, Oral carbohydrate tolerance tests, Arch. Intern. Med. 113 (1964) 641–648. [3] O.P. Ganda, J.L. Day, J.S. Soeldner, J.J. Connon, R.E. Gleason, Reproducibility and comparative analysis of repeated intravenous and oral glucose tolerance tests, Diabetes 27 (1978) 715–725. [4] I.F. Godsland, Intra-individual variation: significant changes in parameters of lipid and carbohydrate metabolism in the individual and intra-individual variation in different test populations, Ann. Clin. Biochem. 22 (1985) 618–624. [5] S.T. Cummings, C.G. Fraser, Variability of capillary plasma glucose in healthy individuals in repeated 75 g oral glucose tolerance tests, Ann. Clin. Biochem. 25 (1988) 634–637. [6] P. Home, The OGTT: gold that does not shine, Diab. Med. 5 (1988) 313–314. [7] K. Schousboe, J.E. Henriksen, K.O. Kyvik, T.I. Sorensen, P.P. Hyltoft, Reproducibility of S-insulin and B-glucose responses in two identical oral glucose tolerance tests, Scand. J. Clin. Lab. Invest. 62 (2002) 623–630. [8] E.J. Feskens, C.H. Bowles, D. Kromhout, Intra- and interindividual variability of glucose tolerance in an elderly population, J. Clin. Epidemiol. 44 (1991) 947–953. [9] A.B. Swai, D.G. McLarty, H.M. Kitange, P.M. Kilima, G. Masuki, B.I. Mtinangi, et al., Study in Tanzania of impaired glucose tolerance. Methodological myth? Diabetes 40 (1991) 516–520. [10] J.M. Mooy, P.A. Grootenhuis, H. de Vries, P.J. Kostense, C. Popp-Snijders, L.M. Bouter, et al., Intra-individual variation of glucose, specific insulin and proinsulin concentrations measured by two oral glucose tolerance tests in a general Caucasian population: the Hoorn Study, Diabetologia 39 (1996) 298–305. [11] J.L. Sievenpiper, L.A. Leiter, V. Vuksan, Intrasubject coefficient of variation corresponds to diagnostic reproducibility in diabetes screening, Can. J. Diab. 26 (2002) 105–112. [12] G. Brohall, C.J. Behre, J. Hulthe, J. Wikstrand, B. Fagerberg, Prevalence of diabetes and impaired glucose tolerance in 64-year-old Swedish women: experiences of using repeated oral glucose tolerance tests, Diab. Care 29 (2006) 363–367. [13] J.O. Christensen, A. Sandbaek, T. Lauritzen, K. BorchJohnsen, Population-based stepwise screening for unrecognised Type 2 diabetes is ineffective in general practice despite reliable algorithms, Diabetologia 47 (2004) 1566–1573. [14] T. Lauritzen, S. Griffin, K. Borch-Johnsen, N.J. Wareham, B.H. Wolffenbuttel, G. Rutten, The ADDITION study: proposed trial of the cost-effectiveness of an intensive multifactorial intervention on morbidity and mortality among people with Type 2 diabetes detected by screening, Int. J. Obestet. Relat. Metab. Disord. 24 (Suppl. 3) (2000) S6–S11. [15] A. Sandbaek, T. Lauritzen, K. Borch-Johnsen, K. Mai, J.S. Christiansen, The comparison of venous plasma glucose and whole blood capillary glucose in diagnoses of Type 2 diabetes: a population-based screening study, Diab. Med. 22 (2005) 1173–1177. [16] J.M. Bland, D.G. Altman, Statistical methods for assessing agreement between two methods of clinical measurement, Lancet 1 (1986) 307–310.
Implications for clinical practice
Most screening for diabetes take place in general practice, where health care providers meet high-risk individuals. Evidence on reproducibility of glucose measurements and hence glucose tolerance classification, upon which treatment decisions are based, is important in this setting. The present study indicates that using capillary whole blood is an attractive option compared to venous whole blood or plasma in screening strategies for diabetes. Despite quantified variations from day to day in blood glucose, we found the measurement results robust with respect to classification of abnormal glucose tolerance in high-risk individuals. About 70% with IFG or IGT were still classified with abnormal glucose regulation in a repeated test and 76% had two results in the diabetic range on separate days. The criterion of two results in the diabetic range on separate days for the diagnosis of diabetes reduces the risk of disease labelling persons who by chance had a test result in the diabetic range. The disease labelling possesses numerous implications to both the individual and the society (insurances, intensive care and treatment). We would not suggest to abandon this criterion of confirmation of the diagnosis of diabetes in asymptomatic persons. Those with one test result in the diabetic range, regardless on first or second test day, have abnormal glucose regulation and not the diagnosis of diabetes. Annual glucose control in individuals with abnormal glucose regulation identified by high-risk strategies is recommendable as the risk of developing diabetes in these individuals is high [21]. It is known that lifestyle intervention at least in individuals with IGT reduces the risk of developing diabetes by up to 58% [22–25]. It seems adequate to initiate lifestyle intervention, which has no side effects, in individuals with IFG or IGT based on one test result when identified by high-risk strategies because of the robustness of the classification shown. We suggest that this also applies to individuals in whom a diabetic glucose value is not confirmed by the second test
Acknowledgements This fully investigator initiated, designed and controlled study received funding from: the Danish Centre for Evaluation and Health Technology Assessment; the Danish Research Foundation for General Practice; the counties of Aarhus, Copenhagen, Ringkoebing, Ribe, and South Jutland; the National Board of Health; the Danish Medical Research Council no. 22-04-0390; the Danish Diabetes Association; the A.P. Møller Foundation; the Bernhard and Marie Kleins Trust; Novo Nordisk A/S and Novo Nordisk Scandinavia AB; Astra-Zeneca; Phizer; Servier; GlaxoSmithKline; and HemoCue.
Conflict of interest The authors state that they have no conflict of interest.
152
diabetes research and clinical practice 80 (2008) 146–152
[17] J.M. Bland, D.G. Altman, Measuring agreement in method comparison studies, Stat. Methods Med. Res. 8 (1999) 135– 160. [18] B. Carstensen, Comparing and predicting between several methods of measurement, Biostatistics 5 (2004) 399–413. [19] S. Sturtz, U. Ligges, A. Gelman, R2WinBUGS: Running WinBUGS and OpenBUGS from R. 2006 http:// mirrors.dotsrc.org/cran/doc/vignettes/R2WinBUGS/ R2WinBUGS.pdf accessed 1st February 2007. [20] B. Carstensen, J. Lindstrom, J. Sundvall, K. Borch-Johnsen, J. Tuomilehto, and the DPS study group. Measurement of blood glucose: comparison between different types of specimens, Ann. Clin. Biochem., in press. [21] A.G. Rassam, J. Mcleod, M.R. Burge, D.S. Schade, Use of the HemoCue blood glucose analyzer in research studies, Diab. Care 21 (1998) 1369–1370. [22] J.L. Chiasson, R.G. Josse, R. Gomis, M. Hanefeld, A. Karasik, M. Laakso, STOP-NIDDM Trial REsearch Group, Acarbose
for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial, Lancet 359 (2002) 2072–2077. [23] W.C. Knowler, E. Barrett-Connor, E.S.E. Fowler, R.F. Hamman, J.M. Lachin, E.A. Walker, et al., Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin, N. Engl. J. Med. 346 (2002) 393–403. [24] X.R. Pan, G.W. Li, Y.H. Hu, J.X. Wang, W.Y. Yang, Z.X. An, et al., Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study, Diab. Care 20 (1997) 537–544. [25] J. Tuomilehto, J. Lindstrom, J.G. Eriksson, T.T. Valle, H. Hamalainen, P. Ilanne-Parikka, et al.,Finnish Diabetes Prevention Study Group, Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance, N. Engl. J. Med. 344 (2001) 1343–1350.