Goldring, Taylor, Kemeny & Anton (2002) examined this question in their study of patients suffering from inflammatory bowel disease by also including measures of quality of life and the doctor-patient relationship. By including the notion of quality of life they turn it from - as in much previous research - an outcome, into a cause. In using these criteria the researchers were able to explain 57% of the variation in medication-taking intention. The larger part of this (30%) was that predicted by measures derived from the HBM. In contrast to the findings in previous studies in diabetic patients, 20% of the variance was accounted for by the perceived risk. The final 8% was accounted for by the interaction effect of quality of life with the HBM. Goldring et al. (2002) place significant importance on two parts of the HBM, the costs and the benefits. They found that if the patient's case of IBD was serious they were more likely to pay attention to both costs and benefits while less seriously affected patients were more oriented towards the costs. Their results suggested that other studies that relied on role-playing by healthy participants might not be providing accurate results as participants would not be weighing the costs and benefits of treatments like real patients with real diseases. Apart from these factors, the doctor-patient relationship was also found to be important, in particular the strength of the recommendation of a particular course of treatment was found to predict 7% of the variance. A significant drawback of this particular study should be noted, in that it only measured the intention to adhere to a treatment regime. The relationship between intention to adhere and the actual behaviour was not measured, although Goldring et al. (2002) report that meta-analytical studies have suggested that a 40-50% variance in intention translates into a 19-38% variance in actual behaviour: not such an impressive result.