Design
The survey was part of the internal quality assurance of the Department forInternal and Integrative Medicine, Essen, Germany. All data were assessed aspaper-based self-rated questionnaires except for main diagnosis, age, and genderthat were taken from the patient’s medical record. Questionnaires werehanded out at admission and collected in by the receptionist. The information onmedical diagnosis and prior complementary therapies use were also used fordeveloping the individual treatment plan.
Ethics approval
The survey was conducted in compliance with the Helsinki Declaration. No approvalwas required from an ethics committee according to the Germanlawa.
Participants
All patients being referred to a Department for Internal and Integrative Medicinebetween January 2001 and January 2004 were asked to complete the surveyimmediately upon admission. All patients were diagnosed with an internalmedicine condition [21]. All included patients gave written informed consent. All data werede-identified and analyzed anonymously.
Yoga use
Using the Freiburg Questionnaire on Attitudes on Naturopathy it was assessedwhich complementary therapies patients had used before to treat their primarymedical complaint [22]. For this analysis, only the question on yoga use was evaluated: Haveyou ever used yoga for your primary medical complaint? (response choice:yes/no). Those patients that reported to have used yoga for their primarymedical complaint were queried about helpfulness: How helpful was yoga for yourprimary medical complaint? (response choices: helpful/not helpful/harmful).Answers were categorized as helpful (yes, no) and harmful (yes, no).
Potential predictors of yoga use
Sociodemographic characteristics
Linear sociodemographic variables were categorized in order to be able tocompute odds ratios between yoga-users and non-users. Age (18–29,30–39, 40–49, 50–64, ≥65 years), sex, education(less than college, college graduate), employment (full-time, part-time,unemployed), and family status (in relationship, not in relationship), wereused as possible predictors together with health behaviors including smokingstatus (current smokers, past smoker, non-smokers), alcohol intake(abstainers, less than twice weekly, at least twice weekly), fast foodintake (abstainers, less than twice weekly, at least twice weekly).
Clinical characteristics
The patients’ main diagnosis was assessed by the transferring doctoraccording to ICD-10 [23]. For this analysis, diagnoses were categorized as a)osteoarthritis, b) arthritis, c) fibromyalgia, d) spinal pain, e) headache,f) other pain, g) hypertension, h) ischemic cardiac disease, i) irritablebowel syndrome, j) inflammatory bowel disease, k) lung diseases, l) other,more rare conditions.
General health status was assessed on a 5 point scale and categorized asgood, very good or excellent versus poor or fair.
Mental health
Anxiety and depression were assessed using the Hospital Anxiety andDepression Scale (HADS) [24]. Scores ranging from 8 to 10 were defined as subthreshold anxietyor depression while scores > 10 were defined as thresholdanxiety or depression [25]. In this sample, Cronbach’s α was 0.81 for the anxietysubscale and 0.84 for the depression subscale.
Satisfaction with health and life in general (FLZ)
Satisfaction with health and life in general were assessed using one 5-pointLikert item each from the questionnaire for life satisfaction (FLZ) [26]. The endpoints were 1 = very unsatisfied and5 = very satisfied. The higher the score the more satisfied thepatient was with health and life in general. Every patient was categorizedas having either high (i.e. above median) or low (i.e. below median)satisfaction with health and satisfaction with life in general.
Health locus of control
Health locus of control was assessed using the GKÜ [27], a 9-item German modified short-form of the multidimensionalhealth locus of control scale, the most commonly used scale to assess healthlocus of control [17]. Response choices ranged from “strongly disagree” to“strongly agree”. The instrument assesses 3 dimensions of healthlocus of control beliefs (3 items each): internal (high perceived owninfluence on health status), external-social (health status perceived ascontrolled by others) and external-fatalistic (health status perceived asdepending on luck or destiny). Cronbach’s α was 0.74 forinternal, 0.46 for external-social, and 0.53 for external-fatalistic healthlocus of control. For every patient, each dimension was categorized aseither high (i.e. above median) or low (i.e. below median).
Statistical analysis
Chi square tests were used to compare sociodemographic, clinical, andpsychological characteristics between patients who ever used yoga for theirprimary medical complaint and those who did not. A p-value of ≤0.05was considered statistically significant. Independent predictors oftherapeutic yoga use were identified using multiple logistic regressionanalysis. A backward stepwise procedure with a Wald statistic p-value of≤0.05 was used. In order to be able to compare the individualinfluence the different variables had on yoga use, continuous variables werecategorized and adjusted odds raios with 95% confidence intervals werecalculated. Among the potential factors that were entered in the initialregression model, only factors were selected that were associated with yogause at a p-value of ≤0.10 in univariate analysis. Statistical analysiswas performed using IBM SPSS® software (release 20.0, IBM, USA).