General practitioners’ opportunities for preventing ill health in healthy vs morbid obese adults: a general population study on consultations

  • Thomas von Lengerke
  • Jürgen John
  • KORA Study Group
Original Article

Abstract

To determine whether overweight and obese adults with and without physical morbidity show an excess utilization of general practitioners in terms of consultation and, among users, number of consultations. In a general adult population survey in the Augsburg region, Germany (KORA Survey S4 1999/2001), body mass index (BMI in kg/m2) was assessed anthropometrically, physical morbidity via computer-aided personal interview with an adapted version of the Functional Comorbidity Index (Groll et al., J Clin Epidemiol 58:595–602, 2005) and consultations with general practitioners in three computer-aided telephone interviews over half a year. Analysis was performed using multiple logistic and zero-truncated negative binomial regressions (two-part model). Data were adjusted for gender, age, socio-economic status, marital status, health insurance and place of residence. Among healthy respondents, i.e. those with no morbidity, neither moderately nor severely obese respondents had significantly higher odds for GP use, or higher numbers of consultations among users, than those in the normal weight range. In contrast, among respondents with any physical morbidity, obese respondents showed excess utilization of GP in that moderately obese adults had significantly higher odds of any GP contact (odds ratio = 2.09, p < 0.01), and, among users, the severely obese group showed an excess number of consultations [incident rate ratio = 1.73, p < 0.05 (adjusted: 1.59, p < 0.10)]. Physical morbidity did not predict any GP use, but tended to be associated with number of consultations among users (incident rate ratio = 1.84, p < 0.10). Under the present conditions of utilization of general practitioners by obese adults in Germany, this group of physicians seems to have the most opportunities for secondary and tertiary prevention in this group of patients. With regard to obese adults who are as yet by and large healthy (and usually of relatively young age), primary prevention efforts may be viable not predominantly by primary care, but community-oriented policies. How far general practice can be an integrative part of primary disease prevention by obesity management is an issue for further investigation.

Keywords

General practitioners Health care utilization Disease prevention Obesity Physical morbidity 

Introduction

According to relevant evidence-based guidelines in Germany, general practitioners (GP) are supposed to play a key role in the long-term management of overweight and obesity (German Obesity Association et al. 2006). In order to live up to these expectations, GP have to be utilized by the relevant population in the first place. In other words, consultations can be considered preconditions for disease prevention in overweight and obese groups to happen, at least as far as it can be achieved in and via general practice.

Thus, the question arises whether GP have sufficient opportunities to interact with overweight and obese patients. Of course, a straightforward contention is that if (co-)morbidities are present, excess health care use by overweight and obese groups will occur. Unsurprisingly, there is evidence for this (e.g. Sander and Bergemann 2003), which breaks the ground for measures of secondary and tertiary prevention. In contrast, primary prevention, i.e. measures to avoid the onset of morbidity in the first place, implies counselling those patients who are affected by the physiological risk factor “overweight/obesity” but do not (yet) suffer from (co-)morbidities. Conceptually, this use of terms implies, of course, that obesity represents “only” a risk factor vs “already” a disease—an issue which is controversially discussed in the literature (e.g. Bray 2004; Heusch 2006). As always in such cases, distinctions between primary vs secondary vs tertiary prevention are fuzzy. However, the 10th Revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10), by distinguishing morbid obesity from other forms, literally classifies the latter as related health problems. Thus, to refer to measures of disease prevention in obese but “otherwise” healthy people as forms of primary prevention seems to be—with the exception of morbid obesity—appropriate.

Against this background, the present study employs an excess utilization rationale to analyse GP consultations by healthy vs morbid obese adults. Specifically, it scrutinizes whether these groups—compared to their normal-weight peers and in a defined time frame—have higher odds of visiting a GP at all, and whether those who do, do so more often. By this approach, we aim to supplement earlier analyses of recognition, management and treatment effects in overweight and obese primary care patients (Bramlage et al. 2004a,b). In those contributions, primary care of obesity and selected comorbidities was studied among patients, rendering important information on what happens in the primary care practice (and, nota bene, what does not). In contrast, the present study uses population data to compare GP use by healthy vs morbid overweight or obese adults with that of normal-weight peers. This makes it possible to assess whether overweight and obesity “drive” the use of GP in the first place. In sum, opportunities for primary prevention (in obese but healthy individuals), and secondary and tertiary prevention (in obese and morbid individuals), for GP will be scrutinized.

Methods

Population and sampling

Data come from the KORA Survey S4 1999/2000, a representative, cross-sectional health survey in the Augsburg region (city of Augsburg plus two adjacent administrative districts), Germany. Approval of the responsible Ethics Committee (Bavarian Medical Association, Munich) was secured. The target population consisted of all German residents of the region born between 1 July 1925 and 30 June 1975. A sample of N = 6,640 was drawn in a two-stage sampling procedure. In the first stage, in addition to Augsburg city, 16 of 70 communities from the adjacent counties were chosen by cluster sampling with probability proportional to size. Using public registry office listings, stratified random sampling was performed within each community, yielding ten strata of equal size according to gender and age. Selection within each stratum used the RANUNI function in SAS 8.1 for Windows. Fieldwork lasted from October 1999 to April 2001. The response rate was 67%, comparing well to other surveys. A non-responder survey via telephone, in which 49% participated, revealed that non-responders more often had lower education (maximally “Hauptschule”: 65 vs 54%) and fair or poor self-rated health (28 vs 21%), were more often unmarried (34 vs 29%) and smokers (29 vs 26%) and more frequently reported physician contact in the last 4 weeks (46 vs 38%), myocardial infarction (6 vs 3%) and diabetes (7 vs 4%) (Hoffmann et al. 2004). Ultimately, N = 4,261 participated in this “main part” of the survey.

Of these 4,261 respondents, a random sample of N = 1,186 with 30 nearly balanced strata by gender, age and body mass index (BMI indicating normal weight, overweight or obesity; see “Measures”) was drawn for a three-wave computer-aided telephone interview (CATI) part of the survey after 2, 4 and 6 months. Ultimately, N = 947 participated in all three waves (response rate: 80%). Fieldwork lasted from October 1999 to August 2001 and averaged over 7.5 months for any participant. Non-responders were more often men (23 vs 17%), from the lowest socio-economic stratum (23 vs 19%), unmarried (27 vs 17%) and smokers (29 vs 17%), but did not differ in health care use or morbidity. N = 5 with a BMI < 18.5 were excluded from analysis for reasons of cell count and possible underweight-specific health problems.

Measures

GP consultations were assessed via self-report in each of the three CATI waves for each of the preceding 8-week periods. The items read “How often did you visit a physician in the last 8 weeks?”, and (for each visit) “Which medical field did that physician belong to: ‘general practitioner’, ‘internal specialist’, (for women) ‘gynaecologist’, ‘otorhinolaryngologist’, ‘dermatologist’, ‘dentist’, or ‘other’?”. For the present purposes, consultations with GP were summed up for those with at least one GP visit to indicate number of consultations among users, and a dichotomous variable was created for the entire sample to indicate “any consultation with GP”.

Obesity

Body weight and height were assessed in anthropometric examinations in the survey’s main part. Calibration of instruments was ensured by weekly or daily inspections using standard weights or resistors, as appropriate. Body mass was indexed by dividing weight in kg by height in m2. Groups were defined following WHO definitions (WHO 2000): normal weight (18.5 ≤ BMI < 25), overweight (25 ≤ BMI < 30), moderately obese (30 ≤ BMI < 35) and severely obese (BMI ≥ 35).

Morbidity

Given the aim of the present study to examine excess utilization of GP attributable to overweight and obesity in healthy and morbid adults, it was crucial to assess the individual health status of each respondent on both a broad and valid basis. To this end, respondents’ medical histories were drawn upon, which had been assessed by self-report for the last 12 months via computer-aided personal interviewing (CAPI). Methodologically, these anamneses build on the MONICA Augsburg protocol (Löwel et al. 2005; Holle et al. 2005). From these, an index for physical morbidity was constructed, based on the Functional Comorbidity Index (FCI) by Groll et al. (2005). Specifically, the number of the following conditions was calculated: myocardial infarction, cerebral infarction, diabetes, angina pectoris, arterial obstructive disease, cardiac insufficiency, arthrosis, attrition of vertebral column and/or intervertebral discs, osteoporosis, bronchial asthma, gastrointestinal diseases and chronic obstructive pulmonary disease. In weighting the index for disease severity, the standardised beta coefficients provided by Groll et al. (2005, Table 3) were used, except for degenerative disc disease, for which a weight was extrapolated from that study reported by Groll et al. (2005) that included this condition (0.33). Also, the present index did not include obesity (as our focal explanatory variable) nor neurological disease (such as multiple sclerosis or Parkinson’s), depression, anxiety or panic disorders, visual impairment (such as cataracts, glaucoma, macular degeneration) and hearing impairment (very hard of hearing, even with hearing aids), since they were not available to us. This reduction from originally 18 to 12 conditions has been judged by the originator of the FCI as nonetheless rendering an index with acceptable validity (Dianne Groll, personal communication, 1 April 2006). However, as our adaptation excludes all mental diseases, we will, for conceptual precision, use the designation “physical FCI” throughout this paper. Besides, while authors debate the usefulness of considering disease severity in indexing morbidities (Groll et al. 2005), we employed the weighted version due to more evidentiary associations with complementary health care use in former analyses (von Lengerke et al. 2006). Finally, in general terms, the selection of the FCI as the basis for indexing morbidity basically rested on two rationales. First, an alternative index employed in a previous KORA study (Werner et al. 2005) has only been attested as a dichotomous index (i.e. any vs no morbidities). Second, when compared to other indices allowing analysis of severity-weighted multimorbidity [e.g. the Charlson Index (Charlson et al. 1987) or others (see e.g. de Groot et al. 2003)], the FCI is the one most strongly relatable to medical histories as assessed in the MONICA/KORA Augsburg studies and also has been shown to be a good option for the ease in its administration and scoring (Fortin et al. 2005).

Socio-demographic and socio-economic factors

Gender, age and place of residence of the respondents were known via the sampling procedure. Marital status and socio-economic status (SES) were assessed in the survey’s CAPI part. SES was operationalised by the revised version of the Helmert Index, which follows national recommendations (Jöckel et al. 1998) and allows one to distinguish SES strata based on education, occupational status and equivalised income (Mielck 2000). Health insurance was assessed via CAPI by asking participants in which fund they were insured and coding these as statutory (German ‘GKV’) or private (‘PKV’).

Statistical analysis

Following descriptive analyses, two-part models (Diehr et al. 1999) of GP consultations were performed both for the healthy and morbid group of respondents. This classification of health status was operationalised by dichotomising the morbidity index (“physical FCI”, see “Measures”) in that respondents with zero values were distinguished from those with values greater than zero. Subsequently, within each subsample, the first part of the two-part model was implemented using multiple logistic regression analysis (via LOGISTIC in STATA/SE 8.1 for Windows), which scrutinized the probability that respondents reported any use of GP at all. In contrast, the second part of each model used zero-truncated negative binomial regression (procedure TRNBIN0 in STATA/SE 8.1 for Windows) in order to model the degree of GP use among users. This procedure is appropriate here since counts are examined with no possibility of zero values. Correspondingly, incident rate ratios (IRR) are reported as coefficients, which describe changes in the outcome associated with a one-unit increase in regressors. In each model, the focal regressor was BMI group, with normal weight serving as the reference group and adjusting for gender, age, socio-economic status, kind of health insurance (statutory vs private), place of residence (rural vs urban) and marital status (single or widowed or divorced vs married). Furthermore, the model for the morbid group included a hierarchical procedure. In addition to the regressors described just now, which formed the first hierarchical step, in a second one severity-adjusted morbidity was included (physical FCI) in order to identify the impact of excess body weight net of morbidity. In all models, no outlier trimming was applied.

Results

Descriptive analysis

Table 1 shows the resulting analysis sample. It distinguishes moderate (class 1) vs severe obesity (classes 2–3) to provide appropriate sample sizes for subsequent analyses of GP use. In this context, three features of this sample are most notable. First, while respondents with any morbidity (i.e. those with a value greater than zero on the “physical FCI”) represent some more than half of the sample (55.2%) overall, this proportion is over 60% among both obese groups (62.5 and 60.5%, respectively). Second, while distributions of gender, SES, health insurance and place of residence across the BMI groups are roughly comparable between the healthy and the morbid subsample, there are clear differences in age and marital status. That is, as expectable, morbid individuals are generally older, and the probability of being 65 or older given severe obesity is particularly low (e.g. compared to being 25–34 years of age) in the “no morbidity” group (3.1 vs 34.4%). Also, morbid obese individuals are much more likely to be divorced or widowed (23.4% in the moderately and 24.5% in the severely obese group) than their healthy peers (5.7 and 6.3%, respectively), reflecting the age differences. Third, within the subsample with any morbidity, there in an increase with BMI in the likelihood to be severely morbid: 26.5% of those in the normal weight range have severe scores on the physical FCI, a figure rising to 40.8% among the severely obese group. In contrast, mild morbidity is about evenly distributed across BMI groups.
Table 1

Sample description and the distribution of potential confounders, by BMI groupsa, b

 

Normal weightc

Overweight

Moderately obese

Severely obese

Total

No physical morbidity

N = 153

(50.3%)

N = 150

(46.3%)

N = 87

(37.5%)

N = 32

(39.5%)

N = 422

(44.8%)

Gender

Female

85

(55.6%)

70

(46.7%)

41

(47.1%)

21

(65.6%)

217

(51.4%)

Male

68

(44.4%)

80

(53.3%)

46

(52.9%)

11

(34.4%)

205

(48.6%)

Age (years)

25–34

52

(34.0%)

48

(32.0%)

32

(36.8%)

11

(34.4%)

143

(33.9%)

35–44

34

(22.2%)

38

(25.3%)

23

(26.4%)

8

(25.0%)

103

(24.4%)

45–54

28

(18.3%)

33

(22.0%)

15

(17.2%)

9

(28.1%)

85

(20.1%)

55–64

21

(13.7%)

18

(12.0%)

8

(09.2%)

3

(09.4%)

50

(11.8%)

65–74

18

(11.8%)

13

(08.7%)

9

(10.3%)

1

(03.1%)

41

(09.7%)

SES (Helmert Index)

Upper

36

(23.5%)

27

(18.0%)

12

(13.8%)

5

(15.6%)

80

(19.0%)

Middle

98

(65.3%)

101

(56.0%)

48

(67.8%)

20

(62.5%)

111

(62.3%)

Lower

19

(12.4%)

19

(12.7%)

27

(31.0%)

7

(21.9%)

72

(17.1%)

Marital status

Married

99

(64.7%)

108

(72.0%)

69

(79.3%)

22

(68.8%)

298

(70.6%)

Single

41

(26.8%)

29

(19.3%)

213

(14.9%)

28

(25.0%)

91

(21.6%)

Divorced/ widowed

13

(08.5%)

13

(08.7%)

5

(05.7%)

2

(06.3%)

33

(07.8%)

Health insurance

Private

22

(14.5%)

24

(16.9%)

10

(11.6%)

4

(12.9%)

60

(14.6%)

Statutory

130

(85.5%)

118

(83.1%)

76

(88.4%)

27

(87.1%)

351

(85.4%)

Place of residence

Urban

72

(47.1%)

60

(40.1%)

34

(39.1%)

15

(46.9%)

181

(42.9%)

Rural

81

(52.9%)

90

(60.0%)

53

(60.9%)

17

(53.1%)

241

(57.1%)

Any physical morbidity

N = 151

(49.7%)

N = 174

(53.7%)

N = 145

(62.5%)

N = 49

(60.5%)

N = 519

(55.2%)

Gender

Female

78

(51.7%)

94

(54.0%)

70

(48.3%)

33

(67.3%)

275

(53.0%)

Male

73

(48.3%)

80

(46.0%)

75

(51.7%)

16

(32.7%)

244

(47.0%)

Age (years)

25–34

13

(08.6%)

11

(06.3%)

7

(04.8%)

4

(08.2%)

35

(06.7%)

35–44

25

(16.6%)

25

(14.4%)

28

(19.3%)

6

(12.2%)

84

(16.2%)

45–54

28

(18.5%)

35

(20.1%)

28

(19.3%)

16

(32.7%)

107

(20.6%)

55–64

40

(26.5%)

51

(29.3%)

42

(29.0%)

11

(22.4%)

144

(27.7%)

65–74

45

(29.8%)

52

(29.9%)

40

(27.6%)

12

(24.5%)

149

(28.7%)

SES (Helmert Index)

Upper

24

(15.9%)

30

(17.2%)

16

(11.0%)

5

(10.2%)

130

(25.0%)

Middle

98

(54.9%)

104

(63.8%)

86

(59.3%)

26

(67.4%)

314

(60.3%)

Lower

29

(19.2%)

40

(23.0%)

43

(29.7%)

18

(36.7%)

130

(25.0%)

Marital status

Married

107

(70.9%)

149

(85.6%)

100

(69.0%)

31

(63.3%)

387

(74.6%)

Single

17

(11.3%)

10

(05.7%)

211

(07.6%)

26

(12.2%)

44

(08.5%)

Divorced/ widowed

27

(17.9%)

15

(08.6%)

34

(23.4%)

12

(24.5%)

88

(17.0%)

Health insurance

Private

30

(19.9%)

21

(12.1%)

19

(13.1%)

5

(10.2%)

75

(14.5%)

Statutory

121

(80.1%)

153

(87.9%)

126

(86.9%)

44

(89.8%)

444

(85.5%)

Place of residence

Urban

72

(47.7%)

78

(44.8%)

69

(47.6%)

18

(36.7%)

237

(45.7%)

Rural

79

(52.3%)

96

(55.2%)

76

(52.4%)

31

(63.3%)

282

(54.3%)

Physical FCId

Mild

66

(43.7%)

70

(40.2%)

54

(37.2%)

20

(40.8%)

210

(40.5%)

Moderate

45

(29.8%)

56

(32.2%)

46

(31.7%)

9

(18.4%)

156

(30.1%)

Severe

40

(26.5%)

48

(27.6%)

45

(31.0%)

20

(40.8%)

153

(29.5%)

aGender, age and place of residence were used as stratification dimensions in sampling (besides BMI; for details, see text); thus, their cross-tabulations with BMI groups may not be viewed as reflecting the situation in the population.

bColumn percent are shown

cNormal weight: 18.5 ≤ BMI < 25; overweight: 25 ≤ BMI < 30; moderately obese: 30 ≤ BMI < 35; severely obese: BMI ≥ 35.

dAmong respondents with physical morbidity on the “physical FCI”, the 50th and 75th percentiles defined cut-points for “mild” vs “moderate” vs “severe” [FCI: Functional Comorbidity Index (Groll et al. 2005)].

Coming to the outcomes of the present study, Table 2 shows the proportions of respondents with any GP use, and number of consultations among users, for different subgroups among the healthy and morbid subsamples. While in general, no large differences are found between women and men, utilization tends to increase with age, especially among the group with physical morbidity. Specifically, from the age group 25–44 upwards, both the probability to have any GP consultation during the last half a year, and the mean number of consultations among users, increase from 45.2% and 2.6 to 72.5% and 4.4, respectively. No indications are available as to why the lowest age group (25–34) tends to report slightly higher use, at least in comparison to the closest group (35–44).
Table 2

GP consultations over half a year in different subgroups: descriptive, bivariate resultsa, b

 

No physical morbidity

Any physical morbidity

Any consultation with GP

p

Number of consultations among users [mean (median; min; max)]

p

Any consultation with GP

p

Number of consultations among users [mean (median; min; max)]

p

n

%

n

%

Gender

Female

88

40.6

0.272c

2.7

(2; 1; 8)

0.093d

175

63.6

0.374c

3.6

(2; 1; 18)

0.269d

Male

94

45.9

 

2.3

(2; 1; 7)

146

59.8

4.0

(2; 1; 20)

Age (years)

25–34

55

38.5

0.119

2.6

(2; 1; 8)

0.663

21

60.0

<0.001

2.9

(2; 1; 18)

0.004

35–44

47

45.6

2.5

(2; 1; 8)

38

45.2

2.6

(2; 1; 09)

45–54

31

36.5

2.3

(2; 1; 8)

59

55.1

3.6

(2; 1; 17)

55–64

27

54.0

2.9

(3; 1; 7)

95

66.0

3.8

(3; 1; 20)

65–74

22

53.7

2.0

(2; 1; 5)

108

72.5

4.4

(3; 1; 17)

SES

Upper

30

37.5

0.145

2.6

(2; 1; 7)

0.681

38

50.7

0.009

4.6

(3; 1; 18)

0.406

Middle

114

42.2

2.3

(2; 1; 8)

190

60.5

3.6

(2; 1; 20)

Lower

38

52.8

2.8

(2; 1; 8)

93

71.5

3.8

(3; 1; 14)

Marital status

Married

135

45.3

0.135

2.5

(2; 1; 8)

0.587

233

60.2

0.188

3.8

(3; 1; 20)

0.998

Single

31

34.1

2.4

(2; 1; 7)

26

59.1

3.7

(2; 1; 18)

Divorced/widowed

16

48.5

2.1

(2; 1; 8)

62

70.5

3.8

(3; 1; 13)

Health insurance

Private

16

26.7

0.008

2.7

(2; 1; 7)

0.602

31

41.3

<0.001

3.5

(3; 1; 20)

0.718

Statutory

158

45.0

2.5

(2; 1; 8)

290

65.3

3.8

(2; 1; 17)

Place of residence

Urban

61

33.7

<0.001

2.3

(2; 1; 8)

0.224

135

57.0

0.036

3.6

(2; 1; 17)

0.373

Rural

121

50.2

2.6

(2; 1; 8)

186

66.0

3.9

(3; 1; 20)

Body masse

Normald

62

40.5

0.281

2.5

(2; 1; 8)

0.760

82

54.3

0.009

3.3

(2; 1; 18)

0.083

Overweight

60

40.0

2.5

(2; 1; 8)

102

58.6

4.0

(3; 1; 17)

Moderately obese

45

51.7

2.4

(2; 1; 6)

105

72.4

3.6

(2; 1; 20)

Severely obese

15

46.9

2.9

(2; 1; 8)

32

65.3

4.8

(3; 1; 14)

Physical FCIf

None

182

43.1

2.5

(2; 1; 8)

 

 

Mild

128

61.0

0.207

3.3

(2; 1; 18)

0.007

Moderate

90

57.7

3.6

(2; 1; 18)

Severe

103

67.3

4.5

(3; 1; 20)

aUnadjusted data

bRow percent are shown

cp values refer to chi-square statistics

dp values refer to regression estimates

eNormal weight: 18.5 ≤ BMI < 25; overweight: 25 ≤ BMI < 30; moderately obese: 30 ≤ BMI < 35; severely obese: BMI ≥ 35.

fAmong respondents with physical morbidity on the “physical FCI”, the 50th and 75th percentiles defined cut-points for “mild” vs “moderate” vs “severe” [FCI: Functional Comorbidity Index (Groll et al. 2005)].

Turning to socio-structural features, participants significantly differ primarily in the probability to have any GP consultation at all. On the one hand, among the group with any physical morbidity, this proportion is largest in those from the lowest SES stratum (71.5%), while only just reaching 50% in those with highest SES (50.7%). Among those with no physical morbidity, this pattern holds as well, however with a gradient not as steep as in the morbid subsample. On the other hand, in both the healthy and the morbid subsample, those who are statutorily insured and live in a rural rather than urban area are more likely to have had GP contact within the 6 months of the observational period. Most notably, almost two-thirds (65.3%) of those in the statutory system (“GKV”) who were affected by some kind of physical morbidity had seen a GP, while only 41.3% of their peers with private health insurance (“PKV”) had done so. In contrast, among GP users, the average number of consultations only slightly differs between GKV and PKV groups (3.8 vs 3.5). Largely similar patterns pertain to the healthy group and for urban vs rural area of residence.

Finally, turning to the focal need factor scrutinized in the present study, differences in GP utilization parameters across the four BMI groups are larger in the morbid than in the healthy subsample. Compared to the normal-weight group, those with more or less elevated body mass had a higher proportion of any GP contact (58.6, 72.4 and 65.3% vs 54.3%) and—given any contact—a larger number of consultations (4.0, 3.6 and 4.8 vs 3.3). In contrast, among the healthy subsample, variations are smaller, and e.g. in the case of number of consultations elevated only in the severely obese group (2.9). Regarding variations of GP use along physical morbidity (“physical FCI”) within the morbid group, those with severe health problems most often have GP contact (67.3 vs 61 and 57.7% in those with mild and moderate morbidity). Also, given any use at all, they have the largest numbers of consultations by far (4.5 vs 3.3 and 3.6).

Regression modelling

In order to hedge these results against chance variations and elucidate whether any associations with GP use of being overweight or obese were mediated by morbidity, multiple regression modelling was run for each outcome (as described in the “Statistical analysis” section). Table 3 shows the results. Adjusted for gender, age, SES, marital status, kind of health insurance and place of residence, no significant differences are found for overweight or obese groups vs those in the normal weight range among the healthy respondents (i.e. those with no physical morbidity). Most notably, for the number of consultations among users, the IRR for the contrasts with the normal-weight group are close to unity, notably both for the overweight and both obese groups.
Table 3

GP consultations over half a year for healthy and morbid adults, by obesity (four BMI groups): regression results (logistic for ‘any visit to GP’ and zero-truncated negative binomial for ‘number of consultations with GP among users’)a, b

 

Any consultation with GP

Number of consultations among users

 

No physical morbidity

Model 1

 

Model 1

Normal weightc

Reference

1

 

1

Overweight

ORd

0.87

IRRe

1.04

95% CI

0.54–1.43

IRR

0.70–1.52

p

0.835

 

0.835

Moderately obese

OR

1.41

IRR

0.92

95% CI

0.80–2.47

IRR

0.61–1.39

p

0.001

 

0.001

Severely obese

OR

1.30

IRR

1.10

95% CI

0.58–2.93

IRR

0.62–1.97

 

Any physical morbidity

Model 1

Model 2

 

Model 1

Model 2

Normal weightc

Reference

1

1

1

1

Overweight

ORd

1.15

1.15

IRRe

1.24

1.19

95% CI

0.72–1.83

0.72–1.84

IRRc

0.86–1.81

0.82–1.73

p

0.835

0.835

 

0.835

 

Moderately obese

OR

2.09**

2.09**

IRRc

1.12

1.08

95% CI

1.26–3.47

1.26–3.47

IRRc

0.78–1.62

0.75–1.54

p

0.001

0.001

 

0.001

 

Severely obese

OR

1.34

1.35

IRRc

1.73*

1.59(*)

95% CI

0.66–2.72

0.66–2.74

IRRc

1.05–2.83

0.97–2.59

Physical FCI

 

0.88

IRRc

1.84(*)

95% CI

 

0.33–2.36

  

0.94–3.59

aMultiple regression models as follows: model 1 with gender, age, SES, marital status, kind of sickness fund and place of residence; model 2 (i.e. second hierarchical step for group with any physical morbidity): predictors from model 1 plus physical FCI.

b***p < 0.001; **p < 0.01; *p < 0.05; (*) p < 0.10

c Normal weight: 18.5 ≤ BMI < 25, overweight: 25 ≤ BMI < 30; moderately obese: 30 ≤ BMI < 35; severely obese: BMI ≥ 35

dOR odds ratio

eIRR incidence rate ratio

In contrast, significant excess utilization attributable to obesity exists in the subsample with physical morbidity. Though severely obese individuals do not have significantly higher odds for GP contact than those with normal weight [OR = 1.34, and 1.35 in model 2, i.e. when adjusted for the physical FCI (both n.s.)], the moderately obese group has double odds when compared to their normal-weight peers, even if adjusted for morbidity (OR = 2.09, p < 0.01). Among users, it is the severely obese group who report excess use. Here, severe obesity is associated with an IRR of 1.73 compared to the normal-weight group (p < 0.05). Also, this association is only partly mediated by morbidity: the latter, which itself predicts number of consultations (IRR = 1.84, p < 0.10), reduces this coefficient only slightly to 1.59 (p < 0.10). In other words, morbid and severely obese users of GP tend to revisit GP with comparatively high frequency.

Discussion

The results can be summarized as follows. First, among healthy respondents, i.e. with no morbidity on the physical FCI, obese respondents had no significantly higher odds for GP use compared to those with normal weight. Also, no higher numbers of consultations were observed among users. Second, and in contrast to the healthy group, obese respondents among those who had reported any morbidity showed excess utilization of GP in a twofold manner. On the one hand, moderately obese adults had significantly higher odds of any contact to GP than the group in the normal weight range. On the other hand, among GP users, the severely obese group had reported excess consultations. Also, this latter effect was only partly mediated by severity-weighted physical morbidity: while among the morbid group, the physical FCI itself tended to be associated with higher numbers of consultations, severe obesity did so even if morbidity was adjusted for. All told, any excess utilization of GP by obese adults in terms of consultations reported earlier (von Lengerke et al. 2005) can be attributed to those segments of the population who are affected not only by obesity, but at least to some extent by some kind of physical (co-)morbidity as well.

A number of limitations of the present study have to be noted. First, utilization was assessed via self-report, thus falling short of medical record data as the gold standard in health services utilization research. However, the survey strategy also holds an advantage, namely to assess utilization in unselected population samples, and thus circumvent problems of merging data from different sources. Notably, this holds in Germany, i.e. depends on organizational aspects of health care systems; in the Netherlands, for example, where all inhabitants are listed in a general practice, extant and absent utilization can be studied via sampling enrolled patients (van Dijk et al. 2006; another advantage of utilization research via surveys is that psychosocial variables, which are practically unavailable to routine and insurance data, can be assessed; e.g. von Lengerke et al. 2006). Second, the severity-weighted morbidity index used in this study—the physical FCI based on the Functional Comorbidity Index by Groll et al. (2005)—shares the complicacies of any generic index of morbidity (Cohen-Mansfield et al. 2001; de Groot et al. 2003). Besides, it did not include hypertension due to the fact that no weight is available for this disorder (Groll et al. 2005). While this may be seen as a problem, as hypertension is a common comorbidity of obesity relevant to primary care (Bramlage et al. 2004a), its relative lack of symptoms makes it unlikely to contribute largely to utilization behaviour as elicited by potential patients. Also, states of mental ill health (e.g. depression) were excluded from the version used here and warrant future consideration. However, given the original FCI has been shown to be particularly valid for physical aspects of health (Fortin et al. 2005), at least this domain can be considered operationalised with sufficient validity. Third, and most critically, gender- and/or age-stratified analyses were neither the focus of our study nor comprehensively possible due to methodological constraints such as subsample sizes and relatively infrequent outcomes (e.g. high use). Thus, while explorative effect modification analyses indicated no conspicuous role of gender, the need for further subgroup analysis is acknowledged. Nevertheless, analyses such as the above in our view do contribute to health care utilization research, not least by taking into account that utilization, to an extent, occurs even in the absence of need factors such as obesity (Hurley et al. 1995). That is, by the excess use rationale employed in the present study, we think that opportunities for GP to interact with obese individuals can be modelled. As noted before, what happens in physicians’ practices is something else (Bramlage et al. 2004a,b) and has to take account of health psychology approaches (Wiesemann et al. 2006).

With these notes of caution in mind, the results of this study may be interpreted as follows. Obviously, generally healthy but obese adults are not driven significantly more strongly to GP than those in the normal weight range. This implies that GP do not have proportionally surplus opportunities to manage obesity in this group, where primary prevention is still possible, and which, due to its comparable youth, is listed in general practice guidelines as in need of treatment (Rebhandl et al. 2006). Also, this finding may indicate that the affected population does not view obesity as a disease, or at least not a disease-like state in need of treatment. Alternatively, this group might not have the resilience to consult with their GP, for instance because they may fear that primary care physicians may share broader society’s negative stereotypes about the personal attributes of obese persons, and view obesity treatment as less effective than treatment of most other chronic conditions (Foster et al. 2003). In contrast, among morbid obese groups, GP do have surplus chances for counselling both moderately and severely obese groups and may use these as windows of opportunity, not least given probably higher motivation induced by morbidity. For instance, obese groups often consult with GP due to impaired health-related quality of life associated with musculoskeletal diseases (Lauterbach et al. 1998). In fact, in the present data, both obese groups did show higher rates of attrition of the vertebral column and intervertebral discs. At the same time, long-term treatment by GP of severely obese patients seems to be relatable specifically to an excess burden of disease induced by diabetes mellitus.

All told, two tentative conclusions may be justified. First, under the present conditions of GP use by obese adults in the general population in Germany, this group of physicians seems to have the most opportunities for secondary and tertiary prevention, given this specific group of patients. This finding is particularly promising because the potentials of primary care-based programmes to be successful in morbid patients has been shown (Bowerman et al. 2001).

Second, primary prevention efforts, i.e. those directed towards obese adults who are as yet by and large healthy, may be viable not so much in primary care, but other, probably community-based settings. Most likely, health behaviour change strategies on multiple levels of preventive action will be of avail here. At the same time, this assertion roots in the status quo of utilization as observed in the present study. That is to say, if, and to what extent, GP are disposed and in a position to further develop their role in primary prevention in the context of obesity will have to be determined by themselves, their professional organizations, and prevention policy and research in the future. Notably, the scope of benefits in the German statutory health insurance system at present does not include provision of individual services such as nutritional consulting through physicians which aim at weight control or reduction in obese patients whose obesity is not judged by the physician as in need of treatment, but “only” a risk factor. Thus, stronger GP integration in obesity management would require changes in the legal basis capable of strengthening their role in primary prevention. For instance, legislation’s currently chary position on obesity treatment is reflected in that, besides interventions such as nutritional consulting, anti-obesity drugs (e.g. orlistat and sibutramine) may not be prescribed on account of statutory health insurances, despite evidence of at least moderate effectiveness in promoting weight loss as an adjuvant to behavioural interventions (Padwal et al. 2004). Thus, any attempts to strengthen the role of GP in this context, which do not include notable add-ons in remuneration for these components of obesity care (together with an increase in overall budget, and thus leeway for interventions), are probably bound to fail. Finally, and despite the potentials to improve general practice in regard to primary obesity care, the question remains as to what extent general practice can contribute to controlling the obesity epidemic in the first place (i.e. in terms of efficacy). In our view, any imbalanced focus on health education without corresponding social-ecological, policy and environmental measures (including reduction or elimination of barriers hindering dietary and physical activity-related change) is not necessarily promising. As Hill and Wyatt (2002) have noted, effective weight management will require providing appropriate physical environments for obese patients, which has major implications for primary obesity care. At the same time, among medical subfields, general medicine may be (one of) the most pertinent when it comes to multimodal prevention, given its self-defined mode of operation taking into account psycho-social, socio-cultural, and ecological determinants of health and risk factors. Among other things, this might include work in inter- and transdisciplinary contexts to further develop systems for categorising diseases according to preventable determinants (Syme 2006).

Notes

Acknowledgements

This study is part of the KORA Survey S4 1999/2001, a project conducted within the KORA research platform (Cooperative Health Research in the Region of Augsburg). This platform was initiated by the GSF-National Research Center for Environment and Health (Neuherberg, Germany), which is funded by the German Federal Ministry of Education and Research and the State of Bavaria. The authors wish to thank all present and former members of the KORA Study Group, especially Rolf Holle, Christian Janßen, Andreas Mielck, Hannelore Nagl, Peter Reitmeir, Walter Satzinger, and Andrea Wulff. Also, our sincere appreciation goes to the originator of the Functional Comorbidity Index (FCI), Dianne Groll (University of Ottawa, Canada), for her prompt, thorough and cooperative support of our adaptation of the instrument.

Competing interests

The authors declare they have no relevant associations that might pose a conflict of interest.

References

  1. Bowerman S, Bellman M, Saltsman P, Garvey D, Pimstone K, Skootsky S, Wang HJ, Elashoff R, Heber D (2001) Implementation of a primary care physician network obesity management program. Obes Res 9:321S–325SPubMedGoogle Scholar
  2. Bramlage P, Pittrow D, Wittchen HU, Kirch W, Böhler S, Lehnert H, Höfler M, Unger T, Sharma AM (2004a) Hypertension in overweight and obese primary care patients is highly prevalent and poorly controlled. Am J Hypertens 17:904–910PubMedCrossRefGoogle Scholar
  3. Bramlage P, Wittchen HU, Pittrow D, Kirch W, Krause P, Lehnert H, Unger T, Hofler M, Kupper B, Dahm S, Bohler S, Sharma AM (2004b) Recognition and management of overweight and obesity in primary care in Germany. Int J Obes Relat Metab Disord 28:1299–1308CrossRefGoogle Scholar
  4. Bray GA (2004) Obesity is a chronic, relapsing neurochemical disease. Int J Obes Relat Metab Disord 28:34–38PubMedCrossRefGoogle Scholar
  5. Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 40:373–383PubMedCrossRefGoogle Scholar
  6. Cohen-Mansfield J, Pawlson G, Lipson S, Volpato S (2001) The measurement of health: a comparison of indices of disease severity. J Clin Epidemiol 54:1094–1102PubMedCrossRefGoogle Scholar
  7. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM (2003) How to measure comorbidity: a critical review of available methods. J Clin Epidemiol 56:221–229PubMedCrossRefGoogle Scholar
  8. Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY (1999) Methods for analyzing health care utilization and costs. Annu Rev Public Health 20:125–144PubMedCrossRefGoogle Scholar
  9. Fortin M, Hudon C, Dubois M-F, Almirall J, Lapointe L, Soubhi H (2005) Comparative assessment of three different indices of multimorbidity for studies on health-related quality of life. Health Qual Life Outcomes 3:74PubMedCrossRefGoogle Scholar
  10. Foster GD, Wadden TA, Makris AP, Davidson D, Sanderson RS, Allison DB, Kessler A (2003) Primary care physicians’ attitudes about obesity and its treatment. Obes Res 11:1168–1177PubMedGoogle Scholar
  11. German Obesity Association, German Diabetes Association, German Nutrition Association, & German Nutritional Medicine Association (2006) Evidenzbasierte Leitlinie: Prävention und Therapie der Adipositas-Version 2006 (in German). German Obesity Association, Hamburg. http://www.dge.de/pdf/ll/Adipositas-LL-2006.pdf (accessed July 03, 2006)
  12. Groll DL, To T, Bombardier C, Wright JG (2005) The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 58:595–602PubMedCrossRefGoogle Scholar
  13. Heusch G (2006) Obesity—a risk factor or a RISK factor for myocardial infarction? Br J Pharmacol 149:1–3PubMedCrossRefGoogle Scholar
  14. Hill JO, Wyatt H (2002) Outpatient management of obesity: a primary care perspective. Obes Res 10:124S–130SPubMedCrossRefGoogle Scholar
  15. Hoffmann W, Terschüren C, Holle R, Kamtsiuris P, Bergmann M, Kroke A, Sauer S, Stang A, Latza U (2004) The problem of response in epidemiologic studies in Germany (part II) (in German). Gesundheitswesen 66:482–491PubMedCrossRefGoogle Scholar
  16. Holle R, Happich M, Löwel H, Wichmann H-E, MONICA/KORA Study Group (2005) KORA—a research platform for population based health research. Gesundheitswesen 67:S19–S25PubMedGoogle Scholar
  17. Hurley SF, Bond LM, Carlin JB, Evans DB, Kaldor JM (1995) A method for estimating baseline health care costs. J Epidemiol Community Health 49:525–531PubMedCrossRefGoogle Scholar
  18. Jöckel K-H, Babitsch B, Bellach B-M, Bloomfield K, Hoffmeyer-Zlotnik JHP, Winkler J, Wolf C (1998) Empfehlungen der Arbeitsgruppe “Epidemiologische Methoden” in der Deutschen Arbeitsgemeinschaft Epidemiologie der Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (GMDS) und der Deutschen Gesellschaft für Sozialmedizin und Prävention (DGSMP) zur Messung und Quantifizierung soziodemographischer Merkmale in epidemiologischen Studien (in German). In: Ahrens W, Bellach B-M, Jöckel K-H (eds) Messung soziodemographischer Merkmale in der Epidemiologie (RKI-Schriften 1/98). MMV-Verlag, Munich, pp 7–38Google Scholar
  19. Lauterbach K, Westenhöfer J, Wirth A, Hauner H (1998) Adipositas Leitlinie: Evidenz-basierte Leitlinie zur Behandlung der Adipositas in Deutschland (in German). University of Cologne, Cologne. http://www.westenhoefer.de/infos/adipositasleitlinie_expertenversion.pdf (accessed October 17, 2006)
  20. Löwel H, Döring A, Schneider A, Heier M, Thorand B, Meisinger C, MONICA/KORA Study Group (2005) The MONICA Augsburg Surveys—basis for prospective cohort studies. Gesundheitswesen 67:S13–S18PubMedCrossRefGoogle Scholar
  21. Mielck A (2000) Soziale Ungleichheit und Gesundheit (in German). Hans Huber, BernGoogle Scholar
  22. Padwal R, Li SK, Lau DC (2004) Long-term pharmacotherapy for obesity and overweight. Cochrane Database Syst Rev 3:CD004094Google Scholar
  23. Rebhandl E, Rabady S, Mader F (eds) (2006) Evidence based medicine—Guidelines für Allgemeinmedizin (in German). Deutscher Ärzte-Verlag, CologneGoogle Scholar
  24. Sander B, Bergemann R (2003) Economic burden of obesity and its complications in Germany. Eur J Health Econ 4:248–253PubMedCrossRefGoogle Scholar
  25. Syme SL (2006) Preventing disease and promoting health: the need for some new thinking. Soz Praventivmed 51:247–248CrossRefGoogle Scholar
  26. van Dijk L, Otters HB, Schuit AJ (2006) Moderately overweight and obese patients in general practice: a population based survey. BMC Fam Pract 7:43PubMedCrossRefGoogle Scholar
  27. von Lengerke T, Happich M, Reitmeir P, John J, KORA Study Group (2005) Utilization of out- and inpatient health services by obese adults: a population based study in the Augsburg region, Germany. Gesundheitswesen 67:S150–S157CrossRefGoogle Scholar
  28. von Lengerke T, John J, KORA Study Group (2006) Use of medical doctors, physical therapists, and alternative practitioners by obese adults: does body weight dissatisfaction mediate extant associations? J Psychosom Res 61:553–560CrossRefGoogle Scholar
  29. Werner A, Reitmeir P, John J (2005) Switching sickness funds and risk compensation mechanisms in the statutory health insurance system in Germany—empirical results from the Cooperative Health Research in the Region of Augsburg (KORA) (in German). Gesundheitswesen 67:S158–S166PubMedGoogle Scholar
  30. WHO (2000) Obesity: preventing and managing the global epidemic [Technical Report Series No. 894]. WHO, GenevaGoogle Scholar
  31. Wiesemann A, Barlet J, Engeser P, Kuth N, Müller-Bühl U (2006) Obesity: behavior change promotion in primary care (in German). Z Allg Med 82:103–107CrossRefGoogle Scholar

Copyright information

© Springer-Verlag 2007

Authors and Affiliations

  • Thomas von Lengerke
    • 1
    • 2
  • Jürgen John
    • 2
  • KORA Study Group
  1. 1.Medical Psychology Unit (OE 5430)Hannover Medical SchoolHannoverGermany
  2. 2.Institute of Health Economics and Health Care Management (IGM)GSF-National Research Center for Environment and HealthNeuherbergGermany

Personalised recommendations