In this cohort study of over 3,500 patients across nine centres, pre-existing diagnoses of depression and/or anxiety were significantly associated with poorer self-reported tolerability of ERCP after adjustment for relevant confounders including pre-existing substance use or misuse. Specifically, pre-existing depression and/or anxiety was associated with higher levels of intra-procedural awareness and discomfort as well as higher levels of post-procedural abdominal pain, nausea, and distention at the time of discharge from the ERCP unit.

Our findings align with those previously reported by Jeurnink et al. who reported that up to half of patients undergoing ERCP under conscious sedation experienced peri-procedural and/or post-procedural pain or discomfort, while up to 10% reported significant mental issues such as state anxiety around the time of and after the procedure [10]. However, pre-existing depression and/or anxiety have not been investigated as a potential risk factor for these self-reported outcomes in this or any other study, to our knowledge, either in the context of ERCP or more broadly in the realm of all endoscopic procedures. While state anxiety is undoubtedly an important factor in patients’ peri-procedural experience, knowledge of pre-existing depression and/or anxiety is arguably even more actionable given that these patients could theoretically be identified via pre-screening and supports or other interventions could be provided earlier on.

PREMs and patient-reported outcome measures (PROMs) are valuable tools completed by patients to evaluate their experiences and outcomes following health services [14, 23–25] but to date have been arguably underutilized in advanced endoscopy. As an example of a successfully applied PROM, the recently developed PAN-PROMISE patient-reported scale was originally developed for acute pancreatitis not related to ERCP, and was found to correlate with the resolution of acute pancreatitis, local AEs, and the development of severe pancreatitis [26]. A prospective cohort study evaluating the usefulness of PAN-PROMISE as a predictor for post-ERCP morbidity showed that an elevated score, even in the absence of post-ERCP pancreatitis, strongly correlated with lower physical quality of life scores as well as increased direct and indirect health care costs at seven days after ERCP [27]. Similarly, in our study, we employed PRO-STEP, a previously validated PREM [21] whose scores have been shown to correlate with post-ERCP outcomes [22]. More broadly, our findings validate the importance of and argue for the more widespread use of PREMs in procedural medical specialties. Routine measurement of patients’ subjective experiences and satisfaction scores has been shown to increase patients’ willingness to attend repeat procedures [28]. As a consequence, PREMs have also been shown to reduce misuse of health care services as well as medicolegal claims [28, 29].

Importantly, though the vast majority (> 90%) of control patients reported next to no abdominal pain, distention, nausea, or sore throat following their procedure (and were therefore considered ‘symptom-free’ at the time of discharge), this proportion was significantly lower for patients with pre-existing depression and/or anxiety. Given that higher abdominal pain scores have been associated with increased unplanned hospital encounters, even in the absence of post-ERCP AEs [22], it is worth exploring whether a higher occurrence of abdominal pain, nausea, and/or distension even after discharge among participants with depression and/or anxiety could partially explain this unplanned healthcare usage. If so, targeted interventions could be instituted for patients with underlying depression/anxiety undergoing ERCP, such as pre-procedural psychoeducation or planned phone calls and/or visits on the day(s) following ERCP to assess symptoms and provide reassurance or further action where needed to try to reduce the burden of unplanned healthcare usage [30]. One can also reasonably ask the question of whether our findings should prompt changes in unit-based policy whereby patients with underlying diagnoses of depression and/or anxiety should only undergo ERCP with either deep sedation or under general anesthesia.

As discussed earlier, various factors explain our findings of poorer tolerability of ERCP among patients with pre-existing depression and/or anxiety, including higher rates of pre- and peri-procedural state anxiety and lower responsiveness to conscious sedation. In addition to these, cognitive biases such as catastrophizing, maladaptive appraisals, and intolerance of uncertainty are common in patients with underlying depression and/or anxiety, and may influence and exacerbate rumination about fearful and threatening stimuli during ERCP procedures [31]. Heightened awareness during surgical procedures can intensify negative emotions, including feelings of helplessness, panic, vulnerability, fear of pain, actual physical pain, feelings of abandonment and betrayal from healthcare providers. This constellation is important given that it can also increase susceptibility to adverse psychological outcomes including post-traumatic stress disorder (PTSD) and dissatisfaction, which may have critical negative downstream consequences on future healthcare encounters [32]. Patients experiencing adverse events during or following an ERCP report more frequent dissatisfaction regarding the attitude and technical expertise of the endoscopists and are less likely to return to the same physician and hospital for a future procedure [33]. Thus, further research is needed to assess whether the reduced tolerability reported by patients with underlying depression/anxiety during ERCP has a broader negative impact on patient satisfaction, willingness to undergo repeat procedures, future sedation levels and/or plans, use of anaesthesia-guided sedation, loss of work and/or school time, and/or direct and indirect healthcare costs. To accomplish these goals, there is a need for future studies that employ validated PREMs and PROMs not only at discharge but also in the days following the procedure. ERCP is an ideal scaffold for this type of inquiry given its established relatively high adverse event rates [8].

Our study has limitations that need to be considered. Firstly, despite the large number of patients undergoing ERCP that were included, the number of participants with depression/anxiety undergoing ERCP under conscious sedation was ultimately less than 500. As a result, the proportion of patients reporting scores greater than 3 and 6 was low overall, which may explain the lack of significance in most outcomes in the sensitivity analysis of ambulatory patients. Although our study may have been underpowered to detect true associations, the low degree of variability in most of our primary estimates when considering the narrow 95% confidence intervals is encouraging that the signals we observed represent the truth. Secondly, despite using highly granular data including hundreds of potential covariates, we were unable to examine the tolerability of ERCP among patients with state anxiety or psychological distress (i.e.: a temporary phenomenon only around the time of the procedure rather than a pre-existing diagnosis of anxiety). Similarly, we did not evaluate the potentially ‘dose-dependent’ effect of the severity of depression and anxiety symptoms, nor did we account for disease chronicity and activity, and medications used, among other important factors. Thirdly, patients without formal diagnoses of depression and/or anxiety in their medical records could have been misclassified as controls if they had undiagnosed depression and/or anxiety. Fourthly, our study focused on depression and anxiety, and therefore, it remains unclear whether these associations are also potentially applicable to other mental health conditions. Additionally, our methodology exposed our results to potential biases, despite our best efforts. For instance, given that research assistants were aware of patients’ mental health histories, confirmation bias could have played a role in our findings. Although all data collection was meant to be objective and protocolized, this could have been a factor. Furthermore, additional biases may have been present in the measurement of the patient-reported outcomes. Given that asking people about negative feelings and peri- and post-procedural side effects could theoretically increase awareness and hypervigilance effects in and of itself [34], the process of requesting patient-reported experiences itself could influence the reported results, especially given that patients undergoing conscious sedation may not remember things clearly, creating a sort of recall bias. Social desirability bias could also play a role in our findings, given that patient-reported experiences were requested verbally by research assistants rather than via anonymized written or web-based forms [35]. Though we made the decision to actively solicit these results verbally to enhance participation and minimize attrition, we acknowledge that this approach could also introduce bias. Finally, we did not evaluate other crucial outcomes that could have arisen because of poorer procedural tolerability, including exacerbation of PTSD, missed time away from work or reduced quality of life.

Our study also has several strengths. To our knowledge, ours is the only study to date that specifically aimed to evaluate multiple intra- and post-procedural patient-reported domains among participants with pre-existing depression and/or anxiety undergoing ERCP. We utilized a PREM specifically validated for use in endoscopy (and shown to correlate with ERCP outcomes) to evaluate the patient’s subjective experience of ERCP under conscious sedation. We consider our data internally valid and all efforts were made to reduce bias, where possible, given that data were prospectively acquired from multiple centers with a very low level of missing data. Thus, we were able to include participants with verified diagnoses of depression and/or anxiety, and we were able to adjust the analyses for several potentially confounding covariates at the patient and procedure levels, including baseline alcohol, opioid, and cannabinoid use. Furthermore, we feel our results are generalizable, given that participants were enrolled at a mix of nine community and academic centres across Canada, the United States, and Europe.