This is a post hoc analysis of a longitudinal prospective cohort study embedded within a three armed randomized controlled trial (RCT) with an educational intervention in a primary care setting 12. The present cohort encompasses all patients enrolled in that trial. The primary goal of the RCT was to asses the impact of guideline-oriented treatment on functional capacity in patients with LBP. A predefined secondary goal of the study was to explore the variation of health care services for LBP. The study was conducted in two centers (Marburg, Göttingen). Ethical approval was obtained from both study sites.

We contacted 818 general practices in the geographical area of both study centers. Addresses were obtained from local health authorities. From 118 practices who agreed to participate, 2 dropped out after randomization. The GPs were on average 12.7 years in practice (SD ± 6.9), 48 years old (SD ± 6) (national average 50.4 years) and 42% of them were female (national average 36%). A total of 68 (59%) practices were run by a single GP. The basic demographic data of our sample is not meaningfully different from the national average 13. Of the 116 participating practices, 3 (2%) were accredited in homeopathy, 5 (4%) in SMT, 16 (14%) in naturopathic medicine, and 25 (21%) in acupuncture. Additionally, 22 (19%) practices offer transcutaneous electric nerve stimulation (TENS) and 46 (40%) at least one form of electrotherapy.

During the recruitment period practice nurses asked every patient with LBP to participate in the study. All patients were registered to allow an estimation of the number of screened patients. Inclusion criteria were (1) consulting for LBP, (2) age above 18, (3) ability to read and understand German, and (4) written consent.

After written consent had been obtained, socio-demographic data were collected with a baseline questionnaire prior to the consultation. During the consultation, GPs assessed warning signs for complicated LBP ("red flags"). Those were major trauma, suspicion or history of cancer, suspicion of inflammatory disease, suspicion of osteoporosis, fever, immunosupression and severe neurological deficits. At follow-ups (four weeks, six months and 12 months later) study nurses conducted standardized telephone interviews and patients were asked about their individual health care utilization, e.g. specialist consultations, medication, and non-pharmacological treatments for LBP within the last 6 months or respectively since inclusion (Figure 1). In the interview, study nurses actively presented a list of 42 possible interventions for LBP. Study nurses were trained in conducting standardized interviews and were able to describe each method in more detail if necessary. We also report some interventions which are not considered as CAM but usually are not recommended as first line therapies by back pain guidelines or are used as household remedies.

The Hanover Functional Ability Questionnaire (HFAQ) was used for the assessment of functional capacity 14. The HFAQ is a frequently used instrument for the assessment of back pain disability and a scale with good psychometric properties that are comparable to the Roland & Morris Scale 15. We preferred the HFAQ because it only consists of 12 items without a loss of psychometric quality compared to the Roland & Morris Scale which is advantageous in telephone interviews. The scale ranges from 0 (extreme functional limitation) to 100 (no functional limitation); scores below 70 are considered to represent a significant impairment.

To classify the natural history of LBP, we used a modification of the von Korff procedure as follows 16:

▪ Acute LBP: single episode of LBP of less than 90 days duration

▪ Recurrent LBP: multiple episodes LBP of less then 90 days duration within the last 12 months

▪ Chronic LBP: more than 90 consecutive days of LBP within the last 12 months.

To estimate the proportion of patients with radicular symptoms, we relied on the patients' reported level of pain radiation into the leg, which we considered as an indicator of possible nerve root irritation. Given the absence of reliable methods, this is a frequently used and pragmatic approach for assessing radicular pain in large cohorts 17.

For the assessment of depression, we applied the German version of the Center for Epidemiologic Studies – Depression Scale (CES-D) 18. Scores above 23 are considered as possible indicator of a clinically relevant depression 19.

Consultation of orthopedic surgeons, general surgeons and neurologists was summarized as "specialist consultation".

We performed logistic regression analyses modeled towards receiving a specific health care service with all socio-demographic and disease-related and healthcare-related factors in Table 1. This procedure provides odds ratios and 95% confidence intervals. Continuous data on depression, pain as measured on a numeric rating scale (NRS) ranging from 0 to 10 and functional capacity were dichotomised. For depression (CESD), we used a cut-off score of >23, for pain levels above 5 and for functional capacity (HFAQ) a cut-off score of >70. All p-values are two-sided and the significance level was 5%.

For acupuncture, SMT, TENS and electrotherapy we added a term in the regression model if such service were provided by the GP.

With a selection procedure (score option) we selected the best model retaining the three most significant predictors. Goodness of fit was tested with the Hosmer-Lemeshow-test and the loglikelihood ratio-test. We report only models in which the null hypothesis of goodness of fit was not rejected 20. The software package SAS 9.1 was used for analysis.

Over a period of three months the 116 participating practices invited approximately 3,400 patients with LBP to participate in the study. They recruited on average 11.6 (SD ± 5.8) patients; a total of 1,342 of 1,588 patients who agreed to participate were finally included. Patients' flow and reasons for exclusion are listed in Figure 2. During the follow-up period 127 (9.4%) dropped out and 1,218 patients were followed up for one year. Patient characteristics are given in Table 1.

In the narrowest to the broadest definitions of CAM service, one half to two thirds of all patients received at least one form of CAM during the one-year follow-up. Using a wide definition including all therapies listed in Tables 2 and 3, a total of 928 patients (69%) received at least one form of CAM. With a narrow definition (acupuncture, SMT and massage), 691 patients (51%) received CAM.

In Table 2 the number of patients who received a specific form of CAM and the predictors for receiving this form of CAM are presented. The most popular forms of CAM were local heat, massage and spinal manipulation. The predictors for receiving at least one form of CAM varied, but the consultation of a specialist and a stay in a rehabilitation facility was consistently associated with the use of CAM.

Therapies which were used by less than 40 patients are given in Table 3.

Our study confirms that a large proportion of patients with back pain is at least using one form of CAM, mostly in the form of local heat, massage and spinal manipulation 45. Using CAM was associated with specialist care, chronic LBP and staying in a rehabilitation facility. Receiving spinal manipulation, acupuncture or TENS were associated with consulting a GP who provides these services. Apart from chronicity, disease-related factors like functional capacity or pain only showed weak or no association with receiving CAM.