The aim of this study was to evaluate the acceptability, effectiveness, and cost-effectiveness of a treatment previously demonstrated to be efficacious in treating a self-selected sample of IBS patients [14, 15]. Patients in this study were consecutively recruited from a gastroenterological clinic and were randomized to ICBT or a waiting list. There was considerable dropout in the study, with only 77% of the ICBT patients participating in the post-treatment assessment and 43% completing the treatment. Compared to the waiting list, the ICBT participants experienced reductions in IBS symptoms, anxiety related to IBS symptoms, as well as improvements in IBS-related quality of life. The effects were more pronounced on the IBS-specific measures than the general measure of daily functioning, probably owing to low pre-treatment values on this measure. Despite the large dropout from the ICBT condition, the treatment produced societal cost-savings compared to the waiting list. Additionally, the cost-savings offset the treatment costs both in short- and long-term.
In our previous trial where we used a waiting list as comparison group, the between-groups effect on GSRS-IBS (d = 1.21)  was larger than the between-groups effect observed in this trial (d = 0.77). Using conservative estimates, by replacing the missing data with pre-treatment assessments, the between-group effect sizes were even lower. Because of the large proportion of missing data and the fact that patients that dropped out were more impaired than the patients that stayed in the trial, the conservative effect sizes are highly influenced by the dropouts. Still, they indicate that for the whole randomized sample, ICBT was clearly less effective than in our previous studies.
The acceptability of ICBT, as measured by the proportions of patients in the ICBT group that stayed in the study (77%) and completed treatment (43%), also differed between this study and our previous studies. In these, 90%-99% of participants completed the post-treatment assessment [14, 15] and 69% of the participants completed all five treatment steps . In addition, while 14 of 75 included patients (19%) withdrew from this study before randomization, 0%-3% withdrew from our previous studies after inclusion [14, 15].
These differences in effectiveness and acceptability are likely linked to the differences in recruitment method. In our study of group CBT for IBS , gastroenterologists at several clinics referred patients that they judged would benefit from treatment and in our previous studies of ICBT [14, 15], patients were self-referred. It would be reasonable to assume that the patients in this study, who were offered to participate simply because they visited a gastroenterological clinic, would be less motivated to engage in treatment. Exposure treatment demands a lot from patients and ICBT also requires them to work independently with planning and conducting the exposure exercises. This may have been too challenging for some patients who might not have applied for the study if we would have used self-referral as recruitment method.
According to clinical guidelines, IBS patients with anxiety, poor coping strategies, or functional impairment should be offered psychological treatment [7, 57]. Indeed, a recent study showed that improvement after CBT for IBS was mediated through change in avoidance behaviors and negative thoughts about symptoms  and that less adaptive IBS behaviors predicted good outcome . However, in this study we did not observe this association. It seems that patients who displayed comparably low levels of avoidance still experienced symptom improvement by changing these behaviors. The study does suggest that large impairment and severe symptoms may be predictive of treatment dropout. However, our previous studies point in another direction. The average pre-treatment scores on GSRS-IBS and IBS-QOL were markedly higher in those studies than in this study, while participants showed larger improvement and treatment adherence [14, 15]. This is in accordance with research demonstrating that IBS patients recruited through the internet are more impaired than clinical patients . Thus, it seems that the method of recruitment is more predictive of treatment outcome than symptom severity and level of impairment. More research is needed to establish what minimum level of avoidance behaviors that is required to benefit from a treatment that targets these behaviors. Other types of psychological treatments for IBS, such as hypnosis, self-management, psychodynamic therapy, and cognitive therapy, may target other mechanisms and require other selection criteria.
An important limitation in this study is our use of a waiting list as a comparison group. Although this practice is common in trials of psychological treatments it has been criticized as it does not control for non-specific factors such as expectancy of improvement and attention from a caregiver . Several trials of psychological treatments for IBS have failed to show differential effects on symptoms when using attention control conditions [[32, 62–64]], and a recent study showed that IBS patients even respond to open-label placebo . These studies underline the need for an active control group when evaluating psychological treatments for IBS. However, ICBT did show superiority to an active control in our previous study where 195 self-referred IBS patients were randomized to ICBT or internet-delivered stress management .
Another limitation is that although our aim was to evaluate the treatment for a clinical sample, it was not carried out within a clinical context. Patients had to participate in an extensive pre-treatment assessment, including a psychiatric evaluation and four weeks of weekly symptom ratings, and were then randomized to treatment or waiting list. These circumstances are seldom part of clinical care and may have contributed to the high dropout rate. We did not include patients who had severe diarrhea and were therefore judged to require prescription of psychotropic drugs. Eleven of the 131 eligible patients (8%) were excluded from the study for this reason. It would have been preferable to prescribe the drug and after evaluation of its effects possibly offer these patients participation in the study. However, for practical reasons this was not possible, as it would have altered the inclusion procedure and time frame. This further limits the generalizability of the results regarding these patients.
The cost-effectiveness analysis also has some limitations. We did not use the actual salary of the patients but estimated them based on educational levels. This produces errors in the estimated costs but should not affect the ICER as it relies on the randomization. Further, we estimated the cost of treatment based only on the amount of therapist time. This estimation does not include the cost of writing the treatment manual or developing the web platform used for delivering the ICBT, nor does it include the overhead costs of running an ICBT treatment clinic.