Study design and participating physicians

The design of the present study has been described earlier [11]. It was a 2-week cross-sectional prevalence study at General practitioners' and internists' offices using a stratified, randomized sampling method for physician selection. Ethical approval was granted from the local ethics committee of the University Hospital of the University Duisburg-Essen, Germany.

Study population

The study population comprised consecutive patients with an age of 18 years and above with either gender who visited their GP at the participating sites on the day of the survey. The only reasons for exclusion from the study were conditions that made it impossible for the patient to participate (serious disabilities or diseases), acute emergencies, or pregnancies and breast-feeding within the previous 3 months. Participating GPs and their staff assessed the patients with a standardized questionnaire with a focus on cardiovascular risk factors and basic data on sociodemographic as well as anamnestic information and life style. In total, 1,511 general practices (8.75% of initially contacted random sample of 17,271 physicians) from 397 out of 438 German cities and administrative districts enrolled 35,869 patients (age range: 18–99 years, women 61.1%).

Diagnostic procedures

Body weight and height were provided by the treating physician (or nurse) indicating whether measured (1/3^rd of cases) or anamnestic data (about 2/3^rd) were used. Overweight was defined as a Body Mass Index (BMI) of ≥ 25 and < 30 kg/m²; obesity as a BMI of ≥ 30 kg/m². Waist circumference (WC) was measured with a common tape provided to all physicians and measured midway between the last rib and the highest part of the iliac crest in a standardized manner. A high WC was diagnosed > 102 cm in men and > 88 cm in women. Blood glucose: Initial capillary blood glucose (BG) quick test was performed independent of the fasting status to identify all patients with a BG concentration of < 5.6 mmol/L or those with a BG concentration of ≥ 11.1 mmol/L. Patients with a non-fasting BG level of 5.6 mmol/L and < 11.1 mmol/L were scheduled for a follow-up visit within the following 2 weeks for a second fasting blood sample. Additionally, venous blood samples were collected and analyzed for levels of glucose, LDL-cholesterol, HDL-cholesterol, total cholesterol and triglycerides using enzymatic assays (Roche Hitachi MODULAR Systems) in a central laboratory.

Calculation of the PROCAM and SCORE cardiovascular risk scores

The PROCAM Score was used to calculate the 10-year risk for cardiovascular morbidity [12]. The score is computed using the following 8 independent risk variables, ranked in order of importance: age, LDL-cholesterol, smoking, HDL-cholesterol, systolic blood pressure, family history of premature myocardial infarction, diabetes mellitus, and triglycerides. Analyses were also done using the SCORE classification [13]. The following parameters were used: sex, age, total cholesterol, systolic BP, and smoking status.

Statistical analyses

For the main variables of the study, basic descriptive statistics, number of observations, mean, standard deviation, median, complemented by 95% confidence intervals for prevalences were calculated. We calculated prevalence rate ratios (PRRs) stratified for age and gender. All statistical analyses were conducted using the statistical software package SAS 9.1 (SAS Institute, Cary, NC, USA) [14].

Sample characteristics

35,869 patients from the GEMCAS study were the basis for the present analysis. Baseline characteristics have been reported previously [11, 15]. In short, 61.1% of patients were female; mean age was 51.7 ± 16.1 years with an average BMI of 27.0 ± 5.2 kg/m². 54.2% were smokers either presently (25.1%) or in the past (29.1%). Any cardiovascular disease was present in 16.3% of patients (n = 5,535) with a history of myocardial infarction or acute coronary syndromes being the most frequent diagnoses (n = 1,968, 12.3%).

Prevalence of overweight using BMI and WC thresholds

In men, the prevalence of overweight was 45.7% and of obesity 24.7% (Figure 1 and Table 1). There was an age-dependent increase in BMI from the age group 18–34 until 65–74 with a slight decline in the prevalence of overweight (-0.5%) and a substantial decline of obesity (-11.1%) thereafter. 26.2% of patients had moderately increased waist circumference (WC > 94 & ≤ 102 cm), 36.4% had a high WC (> 102 cm). Again, there was an age-related increase until the age group 65–74; the proportion of patients with an increased WC was comparable in the elderly, while the prevalence of a high WC dropped by 5.3%.

	BMI				WC
	25,0 – 29,9 kg/m2		≥ 30,0 kg/m2		> 94 & ≤ 102 cm (♂) > 80 & ≤ 88 cm (♀)		> 102 cm (♂) > 88 cm (♀)
	%	95%CI	%	95%CI	%	95%CI	%	95%CI
18–34
male	31.2	[29.1;33.3]	13.1	[11.6;14.7]	14.0	[12.5;15.7]	12.8	[11.4;14.4]
female	18.0	[16.8;19.3]	12.8	[11.7;13.9]	14.2	[13.1;15.4]	19.2	[18.0;20.6]
35–44
male	42.9	[40.9;45.0]	22.0	[20.2;23.7]	23.3	[21.5;25.1]	25.7	[24.0;27.6]
female	24.5	[23.2;25.8]	16.7	[15.6;17.8]	20.1	[18.9;21.4]	27.9	[26.6;29.3]
45–54
male	45.2	[43.4;47.0]	27.7	[26.1;29.3]	28.6	[27.0;30.3]	36.6	[34.8;38.3]
female	30.6	[29.3;31.9]	24.5	[23.3;25.7]	21.8	[20.7;23.0]	40.5	[39.1;41.9]
55–64
male	50.3	[48.5;52.1]	28.6	[27.0;30.3]	28.4	[26.8;30.0]	44.3	[42.5;46.1]
female	36.0	[34.5;37.5]	30.3	[28.9;31.8]	23.0	[21.7;24.3]	52.6	[51.0;54.2]
65–74
male	51.6	[49.7;53.5]	30.0	[28.2;31.8]	30.2	[28.5;32.0]	49.7	[47.8;51.6]
female	40.3	[38.6;42.0]	33.2	[31.6;34.9]	21.7	[20.2;23.1]	63.2	[61.5;64.9]
75+
male	51.1	[48.2;54.0]	19.5	[17.3;21.9]	30.7	[28.0;33.4]	44.4	[41.5;47.3]
female	41.2	[39.0;43.5]	24.5	[22.5;26.5]	24.1	[22.1;26.1]	58.3	[56.0;60.6]
Total
male	45.7	[44.9;46.5]	24.7	[24.0;25.4]	26.2	[25.5;26.9]	36.4	[35.6;37.2]
female	30.6	[30.0;31.2]	23.3	[22.8;23.9]	20.6	[20.1;21.1]	41.5	[40.8;42.1]

Figure 2 and Table 1 show the respective prevalence rates of anthropometric variables in women. The prevalence of obesity was 23.3% and comparable to male while the rate of overweight was substantially lower than in male (30.6% in women vs. 45.7%). Again, there was an age-related increase in prevalence with a decline in the prevalence of obesity beyond an age of 74 years. A high WC (> 88 cm) was substantially more frequent (41.5%, total) than obesity with a peak in the age group 65–74 years (63.2%). The prevalence rates of an increased WC (> 80 and ≤ 88 cm) were comparable between age groups (between 20.1 and 24.1%) except in the young (14.2%).

Proportion of patients with an increased WC stratified by BMI subgroups

Most obese patients had a high waist circumference (87.9% of male and 94.7% of female patients) with lower rates in the young (71.1% of male, 87.8% of female). Most normal-weight patients had a normal waist circumference (83.5% of male and 73.0% of female patients) with a decline in the elderly (60.2% male, 42.5% female). However, only 39.8% of normal-weight men and 57.4% of normal-weight women in the age group beyond 74 had a normal or only elevated WC. Patients in the overweight BMI group Figure 3 had the greatest heterogeneity in WC. 27.1% of overweight men had a normal WC, 42.3% an increased and 30.6% a high WC (women 13.4, 33.1, 53.5%). The proportion of overweight patients with a normal WC was lowest in the highest age group (13.9% male, 6.5% female). On the other hand, young overweight male patients had a normal waist circumference in 59.4% (female 30.4%).

Cardiovascular risk factors in overweight patients depending on WC

The relation of WC to high LDL-cholesterol, low physical activity and low HDL-cholesterol was not uniform throughout age groups and genders. An elevated LDL-cholesterol (> 115 mg/dl) was slightly more frequent in patients with a high WC, but not in all age groups. The same was true for low physical activity. A low HDL-cholesterol was more prevalent in patients with a high WC except in the elderly (≥ 65 years).

PROCAM and SCORE Score per BMI and WC

Prevalence of (abdominal) obesity in adults seeking primary care

Although difficult to compare with other studies due to considerable a difference in methods and sampling the present data indicate a high prevalence of overweight and obesity in primary health care in Germany. According to the "Shape of the Nations survey" 39% of all people visiting a primary care physicians worldwide were overweight or obese; in North America, this proportion was 49% [17]. In the Netherlands, a population closely resembling the German population, obesity was observed in 8.9% of men and 12.4% of women; for overweight these percentages were 42.2% and 30.4%, respectively [18]. On the other hand, in primary care of the southeastern United States (1999–2002) substantially higher prevalence rates for overweight (white women 26%, men 24%) and obesity (white women 36%, men 32%) have been reported [19]. Previous data from German primary care are available for 2001 from the HYDRA- and for 2003 from the DETECT-study [4, 20]. Prevalence rates in 2001 were 37.9% for overweight and 19.4% for obesity, prevalence rates in 2003 were 37.3% for overweight and 22.5% for obesity.

36.4% of male and 41.5% of female patients in our primary care cohort had a high WC (> 102 cm men, > 88 cm women). This appears high as compared to population-based samples across Europe [21–26], which one would expect based on the higher morbidity in a primary care sample. It is lower, however, as compared to the recent nationwide DETECT study of 55,518 consecutive German primary care attendees. In this survey, 43% of male and 53% of female patients met criteria for abdominal obesity (waist circumference ≥ 102 for men and ≥ 88 for women) [20]. Methodological differences between the GEMCAS and the DETECT cohort may account for differences in prevalence rates reported. For example, GEMCAS excluded physicians specialized in diabetology or cardiology in an attempt to document prevalence rates in a broad sample of primary care attendees not being confounded by a large group of multimorbid patients attending a specialized physician. Secondary to this, the difference may be related to the fact that more diabetic patients were included in DETECT (14.6% vs. 12.4%) and patients were slightly older 53.9 vs. 51.7 years. Data from another source became recently available from the global IDEA study in primary care in 63 countries [27, 28]. The data show a fairly consistent prevalence of overweight all over the world with substantial differences in obesity prevalence. Whether this observation could also be due to different BMI thresholds that may apply in particularly Asian countries is a matter of debate.

Waist circumference in the overweight

There is some confusion both from the physician and patient perspective as to being overweight measured by BMI indicates an increased cardiovascular risk. A modest elevation of BMI may be simply due to a higher muscle mass, at least in younger male patients. We were therefore interested to assess if and to what extent WC is suitable to discriminate between normal and elevated risk. The data from the present study clearly show that even in the subgroup of overweight as well as in the subgroup of normal weight patients there is a proportion of patients with an elevation of waist circumference and an unfavourable cardiovascular risk profile. This could guide both physicians and patients towards more aggressively lowering increased body fat in high risk patients with overweight also.

Earlier studies have focussed on whether BMI, abdominal obesity or measures like waist-to-height ratio (WHtR) or waist-to-hip ratio (WHR) may be best to describe an excess of fat and its relation to cardiovascular risk [1]. Several studies have described that WC is a better predictor of cardiovascular risk outcome than BMI [5–8]. While this debate is very valid, the mostly used measure of obesity in primary care is still BMI for which both components (weight, height) are easy and not offensive to determine. Given this standard, it is reassuring to know that normal-weight patients generally have a normal WC and a normal cardiovascular risk. On the other hand, obese persons have a high WC and a high cardiovascular risk. For the group of patients in between, however, it is necessary to measure WC to determine body fat distribution, since 31% of male and 54% of female patients have a high WC when in the overweight category by BMI. As shown in Table 2 &3 most risk factors are significantly elevated in overweight patients with a high WC, the high cut point showing the highest PRRs (stratified into age group and gender). Tables 4 and 5 nicely illustrate that it is particularly hazardous to have a low to moderate BMI but a high degree of abdominal obesity. A higher BMI with a low waist circumference on the other hand may point at individuals with either a more favourable fat distribution or a higher muscle mass (particularly in overweight male patients).

Strength and limitations

Despite the strengths of the study (sample size, representativeness for patients in primary care, simultaneous coverage of structural, doctors and patients perspective) two limitations need to be highlighted: While the data set is representative for the primary care population it may be less so for the general population. The degree of representativeness of the GEMCAS sample for primary care has been reported in more detail by Moebus et al. [11]. In short, 2,600 out of the 17,271 initially contacted primary health care practices responded to the invitation to participate in the study (15% response). The response rate varied slightly in the different postal code areas but no differences greater than ± 2% between contacted and participating sites for each region could be observed. The first 2,070 response faxes were collected and analyzed for eligibility. These revealed 1,835 eligible physicians out of which the first 1,700 were recruited for participation. Among these, 140 cancelled their participation before starting the study, mainly due to time-related issues, communication problems or illness. 33 of the 1,835 originally eligible physicians were recruited in a second run to replace cancellations among the 1,700. The second limitation was that the Body Mass Index was self-reported in 2/3^rd of the patients in GEMCAS. This may indicate, that the prevalences reported are likely lower bound estimates for the true prevalence in this population [29].