Assessing clinical generalizability in sedation studies of upper GI endoscopy

Prior endoscopic studies evaluating conscious sedation have recruited only a small proportion of the patients actually approached. The generalizability of such results to an unselected adult ambulatory care population is thus questionable. The objectives of this study were to determine the characteristics of patients who refused inclusion in a randomized controlled trial of sedation and to compare these characteristics to those of patients who agreed to participate.

Patients who refused participation in a randomized controlled trial of sedation completed standardized questionnaires addressing demographic data and possible confounding variables, and they also completed a validated anxiety questionnaire. Characteristics associated with refusal to participate in the randomized controlled trial were assessed by using uni- and multivariate analysis. Exploratory comparisons of outcomes between both groups included self-reported satisfaction with level of comfort and technical adequacy.

Of 302 patients screened, 203 refused to participate in the randomized controlled trial. The most common reason for refusal was the request for no sedation (135/203). A total of 163 were enrolled in this synchronous study. Patients who refused to participate exhibited 3 distinguishing characteristics: prior treatment for an anxiety disorder (risk difference 9.4%: 95% CI[3%, 17%]), use of analgesic medication (risk difference 10.4%: 95% CI[2%, 19%]), and prior experience with EGD (risk difference 17.9: 95% CI[5%, 30%]). Only upper endoscopy experience was predictive of refusal to participate in the randomized controlled trial. No differences in outcomes existed between patients randomized to sedation and those refusing the randomized controlled trial who chose to receive sedative medication. Patients who refused participation in the randomized controlled trial and who underwent upper endoscopy without sedation were more satisfied than patients randomized to placebo (1.33 vs. 2.58; risk difference -1.25: 95% CI[-0.38, -2.1]). Moreover, the procedure in these patients was more adequate technically (4.86 vs. 4.18; risk difference 0.68: 95% CI[0.21, 1.13]).

Characteristics of patients accepting and refusing randomization were highly similar, differing only with regard to upper endoscopy experience. An exploratory analysis of outcomes suggests that randomized controlled trial results may be biased in a direction that may underestimate the benefits of not administering sedative medication in a real-life setting.