German GPs who practice CAM seem to have much in common with GPs not using CAM. However, a few significant differences have been found, in particular regarding personal characteristics. In comparison, CAM using GPs are more female, younger and have a trend towards a healthier life style. Their practices have higher proportions of privately insured patients and are slightly better technically equipped with ultrasound. GPs with a positive attitude had significant better values within the job satisfaction scale and lower working hours per week than GPs with neutral/negative attitude. Significant predictors for CAM use were a positive attitude towards CAM, holding a special qualification in CAM, own CAM use and the availability of an ultrasound in practice.
In further analyses we identified several significant differences between CAM using GPs having a positive attitude and those using CAM in spite of a neutral or negative attitude. Consequently, CAM use itself seems not to be a predictor for a high job satisfaction, but the combination of CAM use and having a positive CAM attitude does.
Studies comparing GPs in dependence of their CAM use are available from Canada, Australia and Switzerland [8–11]. In an Australian survey, CAM practising GPs tended to be male and full time working, but showed no further differences regarding practice form, practice location or patients seen per week . In a Canadian study CAM using GPs were significantly younger, male and working in solo practices. Both studies were smaller ones (< 500 respondents) and published before 2002.
The 2006 published Swiss study from Widmer et al surveyed 650 primary care physicians (response rate 29%) with the following findings: physicians using CAM were more female, working in group practice, had lower consultation rates and a slightly lower workload .
These findings are in good agreement with our results. However, only when comparing our findings concerning the workload of CAM users with a positive attitude. A remarkable difference of Switzerland and Germany can be seen in the numbers of homeopathy qualifications being the most common ones among Swiss physicians but only on the fourth position in Germany followed by acupuncture, naturopathy and manual medicine. Another difference is that ultrasound is present to a lower extent in CAM practices in Switzerland. In contrast, in Germany, ultrasound was more present in GP practices using CAM. About the reasons can only be speculated.
The reasons why GPs become involved in CAM are diverse. Often the reasons are very personal and closely linked to certain life experiences . Our data show, that, on the one side, there seems to be a congruency between the own life style (e.g. attitude towards and own use of CAM, non-smoking) and orientation of medical care towards CAM. On the other side, CAM using GPs have significantly more privately insured patients. This could be either a consequence of the fact that the majority of CAM interventions are not reimbursed by statutory health insurances. However, this economic advantage could also be a reason for providing CAM.
In a very simplistic view one could hypothesize that there are two groups of GPs using CAM: the economists and the believers. This hypothesis is supported by our finding that differences regarding job satisfaction and work load only arise when considering GPs attitude in the analysis. It seems that GPs with a congruent attitude and use of CAM are more satisfied with their job. Possibly they may have found their way of "escaping the treadmill" and may have developed a higher sense of coherence .
A main strength of our study is the nationwide random sample, but because of the low response rate of 34% our results should be regarded with caution. However, also in the Swiss CAM study they had a comparable response rate with 36%  and from preceding studies it is known that the motivation of German doctors to participate in surveys of this type is generally low with rates between 15% and 30% . Comparisons with available data aiming at validating work parameters and demographic parameters at least confirm that respondents were representative of the background population in basic respects such as gender, age and location of practice . Moreover, it seems that GPs using CAM were more cooperative to participate in the survey than GPs not using CAM. This is a common bias in survey studies and should be considered interpreting the results. In addition, this is an exploratory study; significant changes are only descriptive in nature and do not confirm a causative relationship
Characteristics of health care providers such as job satisfaction are gaining increasing importance in health services research. On the one hand, job dissatisfaction is a major cause of GPs turnover  which can lead to a shortage of GPs. To date, in Germany, there is a continuous decline of the number of GPs, especially of those practicing in the rural parts of the country. On the other hand, physicians' dissatisfaction has not only consequences for physicians but also for patients. There is robust evidence that doctors' feelings of discontent have a significant influence on the quality of patient care [14, 15].
At the level of the doctor-patient-relationship, 'time', 'confidence', 'matching', 'balance of power' and 'rituals' seem to play a major role. Believing in their therapies may create a sense of coherence in doctors and may enable them to activate „placebo" in patients and, doing so, to improve therapeutic outcome . Therefore, sufficient time and a coherent medical attitude may be relevant factors for quality of care.
Furthermore, structural characteristics of the practice are important in the assessment of quality in general practice. We have found no significant differences in basic structural characteristics such as practice type, location of practice and number of patients. However, we found that practices of CAM using GPs are slightly better equipped with ultrasound and have more privately insured patients. Whereas the higher proportion of privately insured patients in CAM practices was expected, the association of CAM use with ultrasound is unclear. Since the availability of ultrasound in practice emerged as significant predictor for CAM use this needs to be investigated in further studies.
Another relevant issue in terms of quality of care is the education of the GPs. Only 61% of the CAM using GPs in our sample had an additional CAM qualification. The remaining GPs provide CAM without certified qualification. This is an important point in terms of quality of care and should be a matter of future discussions about the integration of CAM in primary care. In Germany, the present regulation concerning postgraduate education consists of single CAM qualifications for acupuncture, homeopathy etc. Theses regulations should be reconsidered and adapted according to the requirements of the GPs . A possibility would be to establish a postgraduate education scheme combining the most important (evidence-based) CAM methods.
Our findings permit to speculate on the further development of CAM care in the ambulatory setting. With increasing numbers of female doctors and the inclusion of CAM in undergraduate education in Germany , it can be supposed that the use of CAM will further increase in the future. Therefore, within this increasing CAM-'market' with a large range in quality of the different methods studies on quality and cost-effectiveness are urgently needed