Strength and endurance training lead to different post exercise glucose profiles in diabetic participants using a continuous subcutaneous glucose monitoring system

European Journal of Clinical Investigation (2005) 35, 745–751
Strength and endurance training lead to different post
exercise glucose profiles in diabetic participants using
a continuous subcutaneous glucose monitoring system
E. Cauza*, U. Hanusch-Enserer*, B. Strasser†, K. Kostner‡, A. Dunky* and P. Haber†
*Wilhelminenspital, †Medical University, Vienna, Austria, ‡Princess Alexandra Hospital, Brisbane, Australia
Abstract
Background Although both strength training (ST) and endurance training (ET) seem to
be beneficial in type 2 diabetes mellitus (T2D), little is known about post-exercise glucose
profiles. The objective of the study was to report changes in blood glucose (BG) values after
a 4-month ET and ST programme now that a device for continuous glucose monitoring
has become available.
Materials and methods Fifteen participants, comprising four men age 56·5 ± 0·9 years
and 11
± 0·9 years with T2D, were monitored with the MiniMed (Northridge, CA, USA) continuous glucose monitoring system (CGMS) for 48 h beforeand after 4 months of ET or ST. The ST consisted of three sets at the beginning, increasingto six sets per week at the end of the training period, including all major muscle groups andET performed with an intensity of maximal oxygen uptake of 60% and a volume beginningat 15 min and advancing to a maximum of 30 min three times a week.
Results A total of 17 549 single BG measurements pretraining (619·7 ± 39·8) and post-
training (550·3 ± 30·1) were recorded, correlating to an average of 585 ± 25·3 potential
measurements per participant at the beginning and at the end of the study. The change in
BG-value between the beginning (132 mg dL−1) and the end (118 mg dL−1) for all
participants was significant (P = 0·028). The improvement in BG-value for the ST
programme was significant (P = 0·02) but for the ET no significant change was measured
(P = 0·48). Glycaemic control improved in the ST group and the mean BG was reduced
by 15·6% (Cl 3–25%).
Conclusion In conclusion, the CGMS may be a useful tool in monitoring improvements
in glycaemic control after different exercise programmes. Additionally, the CGMS may help
to identify asymptomatic hypoglycaemia or hyperglycaemia after training programmes.
Keywords Continuous subcutaneous glucose monitoring system, strength and endurance
training, type 2 diabetes mellitus.
Eur J Clin Invest 2005; 35 ( 12): 745–751
Introduction
Type 2 diabetes mellitus (T2D) is a serious chronic diseaseassociated with hyperglycaemia, obesity and the metabolic Department of Internal Medicine V, Department of Diabetes and Rheumatology, Wilhelminenspital, (E. Cauza, syndrome [1,2]. The effectiveness of physical exercise has U. Hanusch-Enserer, A. Dunky); Department of Internal long been recognized for the treatment of T2D [3,4]. Endur- Medicine IV, Division of Sports Medicine, Medical University ance training (ET) has been advocated as the most suitable (B. Strasser, P. Haber), Vienna, Austria; Department of Medicine, form of exercise mode [5,6] with many positive metabolic University of Queensland, Princess Alexandra Hospital, Brisbane, effects, such as decreased blood glucose (BG) levels and reduced body fat (BF), but also undetected hypoglycaemic Correspondence to: Edmund Cauza, MD, Wilhelminenspital, episodes during and after training. In comparison, only Department of Internal Medicine V, Montleartstr. 37, A-1160 limited information on the effect of strength training (ST) Vienna, Austria. Tel.: 01-49150-2508; fax: 01-49150-2509; on T2D is available [7–10]. Measurement of HbA1C is the standard test for assessment of glycaemic control in diabetic Received 18 May 2005; accepted 26 September 2005 subjects. As glycosylated haemoglobin (HbA1C) reflects an average of all blood glucose values for a duration of 3 months recommendations to maintain their energy intake unchanged and HbA1C is negatively influenced by hypoglycaemic episodes, it therefore does not give good information on The study was approved by a local Ethics Committee.
The purpose, nature and potential risks of the study were Continuous glucose monitoring system (CGMS) has explained in detail to the participants before obtaining their now been added to the repertoire of technological devices useful in the management of patients with diabetes. Specif-ically, such monitoring enables clinicians to detect occulthypoglycaemia or hyperglycaemia not otherwise discernable Training programme and study design
with intermittent testing of blood glucose. Therefore, CGMSmay provide us with new and important information on glucose profiles over a longer period of time which will givea more accurate picture of daily blood glucose excursions The study tried to define comparable training units for both than can be determined by HbA1C or finger-stick methods groups. A unit was defined as an organizational unit for both and allow identification of the glycaemic effect after training.
training groups where training occurred. This was facilitated The main aim of the present study was to investigate by taking comparable training units of top athletes of each continuous blood glucose profiles over 48 h at the beginning training group. A top weight-lift body-builder for example and at the end of a 4-month ET and ST programme.
does 30 units per muscle group per week, whereas a top Secondary aims were to evaluate whether it is possible to endurance athlete trains for 10–12 h per week. This study detect hypoglycaemic and hyperglycaemic episodes before decided that 15–20% of these training units for each group reflected an achievable workload for the study participants.
Materials and methods
Systematic ET was performed on a cycle ergometer on threenonconsecutive days of the week. In the first 4 weeks, seven Study population
ET participants trained for a duration of 15 min per sessionthree times per week. The duration of exercise was increased The study randomized 15 patients, four men (mean age ± SE: every 4 weeks by 5 min per session. The total exercise time 56·5 ± 0·9, range: 51–69 years) and 11 women (mean age ± per week, excluding warm-up and cool-down, was 90 min SE: 57·4 ± 0·9, range: 50–70 years), attending the Endo- crinology and Metabolism Department between September Heart rate (HR) was monitored continuously throughout 2000 and May 2002. The participants were divided into two the training period using a Polar® continuous heart rate groups (ST vs. ET) and none of the participants from either monitor (Polar Electro Oy, Kempere, Finland). Based on group was involved in organized exercise training pro- the linear correlation between oxygen consumption (VO ) grammes. All participants fulfilled the diagnosis of T2D, and HR the training was controlled by a HR reflecting 60% according the WHO criteria, with a fasting glucose concen- of VO max, which was derived from ergometry using the tration of 7·0 mmol L−1 (126 mg dL−1). Eight participants (mean age ± SE: 55·1 ± 1·7) undertook ST for 4 months and HR = HRrest + (HRmax − HRrest) × 0·6 ± 5 beats min−1, seven participants (mean age ± SE: 60·3 ± 3·1) undertookET. A physician performed a medical history and physical where HRrest is the HR after a break of 5 min in supine examination on each subject. The participants were excluded if they had rapidly progressive or terminal illness, myocardialinfarction, uncontrolled arrhythmias, third-degree heart block, elevated blood pressure (> 200/100 under therapy),valvular heart disease, nephropathy (microalbuminuria > Eight subjects participated in a 4-month systematic ST 50 µg min−1 albumin excretion), severe peripheral or auto- programme on three nonconsecutive days of the week. A nomic neuropathy or diabetic proliferative retinopathy. All brief warm-up period, which involved 10 min of moderate participants were on antidiabetic drug treatment [Sulphony- cycling with very low intensity, was performed before each lureas (SU) in one case, Metformin in three cases and training session. Instructions in correct exercise techniques combination SU/Metformin in eight cases] and three and supervision of the participants throughout the entire participants were on Metformin in combination with training period were performed by a professional instructor insulin therapy which was commenced within 6 months and an experienced physician. During the first 2 weeks, the weight was kept to a minimal level in order for the participants All participants were advised to maintain their current to learn the exercise techniques, adapt their muscles to medications unaltered during the entire study period and training and prevent muscle soreness. From the third week, no new medication was started during the exercise period.
the training aimed for hypertrophy and started with three sets Medications (especially Sulphonylureas) were modified to per muscle group per week. One set consisted of 10–15 avoid hypoglycaemia only. The participants received specific repetitions, without interruption, until severe fatigue occurred 2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751
Continuous glucose monitoring in diabetic participants capillary blood glucose measurements into the CGMSfor calibration.
Routine HbA1C levels were measured using standard tech-niques. Commercially available standard kits were used for allmeasurements and were performed in quality certified labs.
Figure 1 Study flow chart.
Cardiorespiratory fitness was measured by an exercise stress and completion of further repetitions was impossible. The test. All subjects performed a cycling test to exhaustion on training load was systematically adapted to keep the maximal an electrically braked cycle ergometer (Ergo-metrics 900, possible repetition per set between 10–15 repetitions. When Ergoline, Germany). Heart rate was continuously monitored more than 15 repetitions were successfully performed at a via an electrocardiogram and blood pressure measured in given weight, the weight was increased by an amount that the final minute of each work level. The exercise was started allowed approximately 10 repetitions to be performed. The with a work load of 50 W and increased stepwise by 25 W number of sets for each muscle group were systematically increased from three, at commencement of the programme,to four, five and finally six sets per week at the end of the programme. The ST programme consisted of exercises for allmajor muscle groups. Exercises to strengthen the upper body Maximal strength of a muscle was determined by one re- included bench press (pectoralis), chest cross (horizontal petition maximum (1 RM in kgf ) using the Concept 2 Dyno® flexion of the shoulder joint), shoulder press (trapezius and (Concept 2 Ltd, Wilford, UK), where 1 RM is the maximal latissimus dorsi), pull downs (back muscles), bicep curls, strength that a muscle group is able to generate with a single triceps extensions and exercises for abdominal muscles (sit- contraction. A maximum of three tests were allowed to avoid ups). Lower body exercises included leg press (quadriceps muscle fatigue. The representative exercise for determina- femoris), calf raises and leg extensions (biceps femoris).
tion of 1 RM, as measured by the Concept 2 Dyno,included a bench press performed in a seated position.
The CGMS was inserted at baseline and after the 4-monthperiod of ET, or ST, in the abdominal tissue and calibrated Data analysis was performed using the Statistical Package for over a 60-min period and operated for 48 h (Fig. 1). The Social Sciences (SPSS 10·0). All parameters were described by CGMS was inserted in all participants within 1 week of mean values ± standard error of the mean (SE). A Student’s completing their training programmes.
paired t-test was used to assess significant differences of thesame variables within the participants before and after thetraining period. Any values of P ≤ 0·05 were considered statistically significant and 95% confidence limits (CL) werecalculated. To estimate change in BG-values between the begin- Continuous glucose monitoring system measurements ning and the end of the training programme, multiple regressionmodels and t-tests were applied. As dependent variable log (bz- This is a Holter-style sensor system designed to continu- value) was used as this resulted in approximately normally ously monitor interstitial fluid glucose levels within a range distributed residuals. Owing to different intervals being meas- of 40 – 400 mg dL−1. The glucose sensor is a microelectrode ured per participant and period (pre/post) the analysis included that is inserted into the subcutaneous tissue and generates ‘time of day’ and ‘day-off ’ period effects into the model. The an electronic signal proportional to the amount of glucose final regression model included sine and cosine components present in the surrounding interstitial fluid. Each participant of ‘time of day’ as well as a linear ‘day-off ’ period effect to arrived at the office before the first training unit started and account for systematic variation in observations. Mean values the catheter of the CGMS device was inserted horizontally of the resulting residuals were then analyzed by t-tests.
into the abdominal subcutaneous tissue. The signal was sentto a portable monitor that recorded sensor signals every5 min and converted them into blood glucose readings.
After 2 days the data were downloaded via the Com-Station using the MiniMed Solutions Software version 2·0b(MiniMed).
A summary of the participants demographic and clinical All participants were instructed in the use of the CGMS characteristics at baseline and after 4 months’ strength or device and asked to enter at least four daily self-obtained 2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751
Table 1 Summary of participants demographic and clinical characteristics at baseline and after 4 months of strength or endurance training
ST, strength training; ET, endurance training; HbA1C, glycosylated haemoglobin; mean BG; average of blood glucose value measured PeakVo2 max. maximal oxygen uptake; mean ± SE; NS = not significant; BMI, body mass index; P1 value, difference in each group before and after 4 months of strength or endurance training.
At study entry, participants who undertook ST had Changes in medications
higher average BG baseline levels [(138 ± 13·1 vs. 124·4 ±13·5) mg% ± SE], lower body mass index (BMI) (29·9 ± 0·8 After 4 months of training, the antidiabetic medication in vs. 36·3 ± 4·7, P = 0·03) and lower fat mass (FM) (38·9 ± 2·3 ST participants was reduced by 3·3% for SU and was vs. 46·9 ± 4·0, P = 0·04) kg than the ET group. The HbA1C unchanged for Metformin, and for the two ST participants was not significantly different in either group.
receiving insulin therapy the insulin dose was unchanged.
After the ST/ ET period there were significant differences For ET participants, SU therapy was reduced by 1·6% from for BMI (29·9 ± 1 vs. 36·3 ± 2·6, P = 0·03), for FM (33·5 ± baseline, and Metformin therapy and the insulin dose was 2·9 vs. 44·4 ± , P = 0·04) kg and a significant increase in unchanged. None of these changes was statistically significant.
bench press weights (59·65 ± 6·5 vs. 31·7 ± 4·0, P ≤ 0·01)between groups.
A total of 17 549 single BG measurements comprising pretraining (619·7 ± 39·8) and post training (550·3 ± 30·1), Discussion
corresponding to an average (585 ± 25·3) potential measure-ment per participant at the beginning and end, were evaluated.
This study observed significant improvements in glycaemic The change in BG-value between the beginning (132 mg control, as shown by reduced BG-profiles, in participants dL−1) and the end (118 mg dL−1) for all participants was with diabetes on ST, while the effects of ET on the respective significant (P = 0·028). Also, the change in BG-value was parameters were only moderate. As expected, maximum significant for the ST programme (P = 0·02), but for the strength (1 RM) of bench press (24·2%) increased after 4 ET programme no significant change was measured (P = months of ST in contrast to no improvements after 4 months 0·48). Glycaemic control improved significantly in the ST of ET. Not surprisingly, improvements were observed in peak group where the mean BG was reduced significantly by VO after ET (7·5%), while no such changes were seen in 15·6% (Cl 3–25%), while in ET group only a trend in mean ST. The later findings were predictable from the specificity BG decrease of 5·5% was measured. Two summary figures of the training stimulus and demonstrated that the training containing the mean values for the respective time points was adequate in both groups, which showed that the specific for ET and ST are shown in Fig. 2 and Fig. 3.
training stimulus was sufficient for both training groups.
Asymptomatic nocturnal hypoglycaemic episodes with Medications, especially Sulphonylureas, were reduced glucose values 40 mg dL−1 were recorded in 1/0 participants only to avoid hypoglycaemia. Reduction of Sulphonylureas before and 1/1 participants after ET/ST.
was greater during ST (−3·3%) than during ET (−1·6%) even Prolonged hyperglycaemic periods with values ≥ 300 mg though this did not reach statistical significance. Insulin dose dL−1 were found in 2/1 participants before and no particip- and Metformin therapy were unchanged during ST and ET.
More importantly, the observations were made in the pre- In one participant (NG) an interesting glucose sensor sumed absence of dietary changes during the training period.
profile with frequent hypoglycaemic and hyperglycaemic The positive change in the glycaemic profiles of the episodes before and after ET was measured. In another participants after training are therefore presumably owing participant, female, (SG) an example of a prolonged un- to the effectiveness of the training programme.
recognised hypoglycaemic period during the night after ST The use of CGMS has opened a new window through which it is possible to observe directly in vivo what happens to 2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751
Continuous glucose monitoring in diabetic participants Figure 2 Mean glucose values of all
participant before and after ST.
Figure 3 Mean glucose values of all
participant before and after ET.
participants with diabetes. The first noticeable observation cells are consuming glucose rapidly. In general, differences in our participants after strength or ET was the improvement between plasma glucose and interstitial fluid glucose have of the glycaemic profile (estimated from 17 549 single BG been reported to be relatively minor with the lag time measurements), even in the absence of significant changes between them usually being < 10 min [11]. As the study in the usual daily schedule (oral antidiabetic medication and determined whole daily glucose profiles before and after the dose of insulin and diet). This new device showed that it is training periods and not only single glucose values this possible to obtain exact information on glycaemic profiles limitation of the device was compensated.
after ET or ST programmes while HbA1C values only showed Extensive large studies have validated the agreement trends in improvements during the same time period. It was and accuracy of sensor readings and shown a significant also understandable that numerous glucose elevations during correlation between capillary glucose determinations and the day and not a single glucose value are an important simultaneous sensor readings [12–15].
determinant of the overall glycaemic control.
Reports on the sensitivity of detecting hypoglycemia have The CGMS sensor measured the amount of glucose in been controversial. In a recently published paper [16] the the interstitial fluid and then calculated the expected reliability and sensitivity of a CGMS in detecting hypo- corresponding blood glucose level. During physical glycaemia was weak and in other papers the CGMS often exercise, the interstitial fluid glucose levels may decrease failed in detecting hypoglycaemia [17–19]. In contrast to more quickly than the plasma glucose level because the these results are the outcomes of other studies [20,21].
2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751
In the recently published paper of the Diabetes Research CGMS may help to identify asymptomatic hypoglycaemia in Children Network study group [16], the authors speculated or hyperglycaemia after physical training programmes.
that the greatest value of a CGMS may be for detectingtrends and not for serving as a sentinel for single hypo-glycaemia. In this study asymptomatic nocturnal hypoglycae-mic episodes with glucose values 40 mg dL−1 were recorded References
in one participant before training and two participants afterthe training programme. In these two cases after the training 1 DeFronzo RA, Simonson D, Ferrannini E. Hepatic and programme, the episodes of hypoglycemia were foreseeable peripheral insulin resistance. A common feature of type 2 because of slowly lowering BG over a longer time period (non-insulin-dependent) and type 1 (insulin dependent) before hypoglycemia appeared, and the accuracy of sensor diabetes mellitus. Diabetologia 1982;23:313–9.
readings showed significant correlations with the capillary 2 Unger RH, Grundy S. Hyperglycemia as an inducer as well as glucose determinations of the participants.
a consequence of impaired islet cell function and insulin resistance. Implications for the management of diabetes. Prolonged hyperglycaemic periods with values Diabetologia 1985;28:119–21.
> 300 mg dL−1 were found in three participants before 3 American Diabetes Association. Clinical practise training and none after training. CGMS is an accurate recommendations 1998. Diabetes Care 1999;22 (Suppl.):49–53.
device in sensing higher glucose levels, which has also been 4 Young JC. Exercise prescription for individuals with metabolic disorders: Practical consideratioNS Sports Med 1995;19:43–54.
A limitation of this study was that participants randomized 5 Devlin JT, Horton ES. Effects of a prior high-intensity exercise for the ST group had higher baseline levels for BG than on glucose metabolism in normal and insulin resistant men. participants randomized for the ET group. It should be Diabetes 1985;34:973–9.
mentioned that values that are high and outside a normal 6 Fritz T, Rosenquist U. Walking for exercise – immediate effect physiological range can be reduced more easily than values on blood glucose levels in type 2 diabetes. Scand J Prim Health
Care 2001;19:31–3.
that lie closer to the normal range. Although BGs in the ST 7 Castaneda C, Layne JE, Munoz-Orians L, Gordon PL, group were higher and outside the physiological range at Walsmith J, Foldvari M et al. A randomized controlled trial of study entry, after the 4-month training period the values resistance exercise training to improve glycemic control in older were closer, or nearly equal, to normal values than those in adults with type 2 diabetes. Diabetes Care 2002;25:2335–41.
8 Dunstan DW, Daly RM, Owen N, Jolley D, De Courten M, Another limitation of this study was that it is a com- Shaw J et al. High-intensity resistance training improves parative type study without a control group (as there was glycemic control in older patients with type 2 diabetes. Diabetes no nonexercise control group) and it is uncertain as to what Care 2002;25:1729–36.
extent the changes following ST, or ET, are on top of the 9 Dunstan DW, Puddey IB, Beilin LJ, Burke V, Morton AR, changes that could possibly be expected if no exercise training Stanton KG. Effects of a short-term circuit weight training programme on glycaemic control in NIDDM. Diabetes Res Clin was undertaken. However, as the improvement in maximum Pract 1998;40:53–61.
strength (1 RM) of the ST subjects was highly significant (25% 10 Erikson J, Taimela S, Eriksson K, Parviaianen S, Peltonen J, of initial levels) and this finding was predictable from the Kujala U. Resistance training in the treatment of specificity of the training stimulus, the authors believe that non-insulin-dependent diabetes mellitus. Int J Sports Med the specific training stimulus was sufficient and probably 1997;18:242–6.
responsible for the changes in BG-values after 4 months.
11 Rebrin K, Steil GM, van Antwerp WP, Mastrototaro JJ. The authors were aware that the improvements from training Subcutaneous glucose predicts plasma glucose independent of would be more impressive when compared with participants insulin: implications for continuous monitoring. Am J Physiol without training, but the Ethics Committee had reserva- 1999;277:E561–71.
tions about a group where exercise was withheld from the 12 Hay LC, Wilmshurst EG, Fulcher G. Unrecognized hypo- and hyperglycemia in well-controlled patients with type 2 diabetes mellitus: the results of continuous glucose monitoring. Diabetes Other points to be discussed are training intensity and Technol Ther 2003;5:19–26.
volume in ST and ET programmes. The ET with an intensity 13 Kaufman FR, Gibson LC, Halvorson M, Carpenter S, of VO2 max 60% and a volume (starting at 15 min and Fisher LK, Pitukcheewanont P. A pilot study of the continuous increasing to a maximum of 30 min/three times a week) is glucose monitoring system: clinical decisions and glycemic at the lower end of an endurance training programme; this control after its use in pediatric type 1 diabetic subjects. was selected because of the poor training status of the Diabetes Care 2001;24:2030–4.
participants at the start of the study. As all the participants 14 Zavalkoff SR, Polychronakos C. Evaluation of conventional were new to physical training the study had to start with blood glucose monitoring as an indicator of integrated glucose values using a continuous subcutaneous sensor. Diabetes Care
2002;25:1603–6.
In conclusion, the CGMS is considered a useful tool in 15 Gross TM, Bode BW, Einhorn D, Kayne DM, Reed JH, monitoring improvements in glycaemic control after a ST White NH et al. Performance evaluation of the MiniMed or ET programme. Additionally, the CGMS is of clinical continuous glucose monitoring system during patient home utility in routine clinical practice because the glucose profile use. Diabetes Technol Ther 2000;2:49–56.
is representative of the overall control of the participants, 16 The Diabetes Research in Children Network (DirecNet) Study similar to a self-control of blood glucose. Furthermore, the Group. Accuracy of the GlucoWatch G2 Biographer and the 2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751
Continuous glucose monitoring in diabetic participants Continuous Glucose Monitoring System during hypoglycemia. continuous glucose monitoring system in pediatric patients with Diabetes Care 2004;27:722–6.
type 1 diabetes. J Pediatr 2002;141:625–30.
17 Metzger M, Leibowitz G, Wainstein J, Glaser B, Raz I. 21 Amin R, Ross K, Acerini CL, Edge JA, Warner J, Dunger DB. Reproducibility of glucose measurements using the glucose Hypoglycemia prevalence in prepubertal children with type 1 sensor. Diabetes Care 2002;25:1185–91.
diabetes on standard insulin regimen: use of a continuous 18 Guerci B, Floriot M, Bohme P, Durain D, Benichou M, glucose monitoring system. Diabetes Care 2003;26:662–7.
Jellimann S et al. Clinical performance of CGMS in type 1 22 The Diabetes Research in Children Network (DirecNet) Study diabetic patients treated by continuous subcutaneous Group. The accuracy of the GlucoWatch Biographer in children insulin infusion using insulin analogs. Diabetes Care with type 1 diabetes: results of the diabetes research in children 2003;26:582–9.
network (DirecNet) accuracy study. Diabetes Technol Ther 19 McGowan K, Thomas W, Moran A. Spurious reporting of 2003;5:791–800.
nocturnal hypoglycemia by CGMS in patients with tightly 23 The Diabetes Research in Children Network (DirecNet) Study controlled Type 1 diabetes. Diabetes Care 2002;25:1499–503.
Group. The accuracy of the CGMS in children with type 1 20 Kaufman FR, Austin J, Neinstein A, Jeng L, Halvorson M, diabetes: results of the diabetes research in children network Devoe DJ et al. Nocturnal hypoglycemia detected with the (DirecNet) accuracy study. Diabetes Technol Ther 2003;5:781–9.
2005 Blackwell Publishing Ltd, European Journal of Clinical Investigation, 35, 745–751

Source: http://www.cauza.at/cauza.pdf/Literaturliste/2.pdf