Impact of an Electronic Health Record Update on Appropriately Scheduled 5-Year Colorectal Cancer–Screening Colonoscopies at an Academic Medical Center

Abstract

Background and Aims

Prior studies have shown that colorectal cancer screening/surveillance colonoscopies are performed earlier than guideline-recommended intervals. We implemented an electronic health record (EHR)-based intervention to improve the proportion of appropriate 5-year colonoscopies that are scheduled at a tertiary academic medical center.

Methods

This was a single-center pre-post study of patients with a primary care provider at our institution with an active EHR flag for 5-year screening/surveillance colonoscopy due in 2023. We first conducted a chart review of cases scheduled between April 2023 and July 2023 to evaluate indication and appropriateness for 5-year screening based on the US Multi-Society Task Force February 2020 guidelines. Starting at July 2023, if the case was guideline inappropriate, the EHR flag was revised to reflect updated guideline-based recommendations. The postimplementation phase was defined as between October 2023 and January 2024.

Results

Of 17,313 cases scheduled in the 10-month period, 9831 were screening/surveillance colonoscopies. Of the total number of cases, 1746 (10.1%) had a 5-year return flag, with 555 patients in the preimplementation phase, 538 in the transition phase, and 653 in the postimplementation phase. In the preimplementation phase, 69.2% (384/555) were guideline inappropriate, whereas in the postimplementation phase, 56.5% (369/653) were nonguideline. There was a 10.9% increase in appropriately scheduled cases per week after implementation.

Conclusion

This study highlights the effectiveness of an EHR intervention in improving adherence to colorectal cancer–screening protocols. The concomitant decrease in polyp-based guideline-inappropriate cases shows that the intervention reflected updates in polyp guidelines. However, continued efforts are needed to reduce the high proportion of nonguideline cases, to increase capacity for other indicated procedures.

Introduction

In 2020, the US Multi-Society Task Force (USMSTF) updated colorectal screening and surveillance guidelines.¹ However, prior studies have shown that it often takes years for guidelines to be implemented into clinical practice.² Separately, previous retrospective studies based on electronic health data have shown that surveillance colonoscopies are often not performed in alignment with guideline-recommended intervals.³ Recently, the American College of Gastroenterology and American Society of Gastrointestinal Endoscopy updated recommendations for quality indicators in colonoscopies to include screening and surveillance intervals,⁴ highlighting the importance of adhering to guidelines with the effort to minimize unnecessary care. In this single-center pre-post quasi-experimental study, we implemented a quality-improvement intervention to update EHR health care maintenance flags in accordance with current guidelines with the hypothesis that this intervention would improve adherence to updated guidelines. The primary outcome was the proportion of appropriate 5-year screening colonoscopies scheduled, and the secondary outcome was the reason for nonguideline cases.

Methods

This study followed the framework of a quality-improvement project including defining the existing state, implementing an intervention, and then measuring the impact of the intervention.

Baseline Workflow

At this tertiary, academic medical center, primary care providers (PCPs) and endoscopists utilize a health care maintenance flag in the Epic (Epic System Corps) electronic health record (EHR) that signifies when a patient is due for colorectal cancer (CRC) screening or surveillance. The health care maintenance flag is a built-in interface into the institution’s Epic system that defaults the screening colonoscopy interval to a 10-year interval starting at age 45 and automatically updates the frequency based on tests performed (such as a colonoscopy and stool-based tests) or by modifications by providers (PCP and gastroenterology physicians) based on results of risk factors. The flag triggers the PCP to order a colonoscopy for the patient, after which the patient contacts or is contacted by the endoscopy center. The endoscopy order requires an indication and for the procedure to be classified as screening (including surveillance) or diagnostic. Importantly, the institution’s cecal intubation rate has consistently been above 97%, and adenoma detection rates above national standards, suggesting good exam quality. This is an important quality indicator necessary for patients to be scheduled based on appropriate guideline intervals.

Preimplementation

From April 2023 to July 2023, we first assessed the proportion of cases that were ordered as 5-year screening/surveillance colonoscopies. For the denominator, we obtained data for patients scheduled on a weekly basis on the Sunday for the upcoming workweek. For the numerator, we included all scheduled cases with an active EHR flag for a 5-year nondiagnostic colonoscopy and the order indicating a screening or surveillance procedure. Excluded were those with a personal history of CRC and those with multiple conflicting CRC-related flags applied to their EHR.

For 5-year screening colonoscopies, we conducted a retrospective chart review of all available prior colonoscopy findings and provider documentation to evaluate indication and appropriateness for 5-year screening based on USMSTF February 2020 guidelines. Current guidelines for polyp screening were defined as the consensus update from the US Multi-Society Task Force on Colorectal Cancer February 2020 guidelines,¹ an update from the prior recommendations in 2012.⁵ Guidelines referenced for family history were the 2017 Multisociety Task Force on CRC.⁶ A chart review was performed by 3 physicians (M.B., S.P., and M.S.) to confirm EHR flags and appropriateness of indications. Indication was assessed by reviewing prior colonoscopy and pathology reports, polyp number, size, and family history. Nonguideline was defined as having a polyp size, pathology, or number on the prior colonoscopy, which did not meet USMSTF 2020 guidelines, or having a charted family history that did not meet the abovementioned guidelines.

Intervention

Based on the feedback from primary care colleagues, a key aspect of the intervention was to maintain trust between patients and their PCPs. Thus, the intervention focused on not challenging any prior conversations but rather to make available information in the EHR for the PCP to appropriately assess indication for ordering a colonoscopy at the next visit. The population for our intervention were patients marked in the EHR with a PCP in our health system, who had a 5-year CRC screening/surveillance EHR flag due in 2023, a pending PCP appointment scheduled in the future within the year, and no scheduled colonoscopy at the time of assessment. The intervention began by July 2023, where a gastroenterology physician assistant reviewed the patient chart for appropriateness per the aforementioned criteria. If the EHR flag was deemed inappropriate, the flag was revised to reflect updated guideline-based recommendations, and a letter was sent to the patient informing them of the change (Appendix 1). If the EHR flag was deemed appropriate, the chart was marked as reviewed in an internal tracking system, and no changes were made in the EHR.

Statistical Analysis

We assessed the proportion of appropriate 5-year screening and surveillance colonoscopies on a weekly basis and used a control chart to assess change over the phases of the study. We also assessed the proportion of the reasons for nonguideline cases. Given that the average time from scheduling to procedure during this period was 3 months, we expected the impact of the intervention in July 2023 to be seen in October 2023. Accordingly, phases of this study were defined as preimplementation from April 1 to June 30, 2023, transition from July 1 to September 31, 2023, and postimplementation from October 1, 2023, to January 31, 2024. A 2-tailed t-test was used to determine the statistical significance of the change in average appropriateness before and after the intervention, with an alpha value of 0.05.

Results

The 10-month period included 17,313 cases scheduled in the endoscopy suite, of which 9831 were screening or surveillance (nondiagnostic) colonoscopies. Of the total, 1746 (10.1%) had a 5-year return flag, with 555 patients in the preimplementation phase, 538 in the transition phase, and 653 in the postimplementation phase (Table). During the intervention phase, 2731 cases were reviewed, and 1063 EHR flags were adjusted according to updated guidelines. In the preimplementation phase, 69.2% (384/555) were not guideline appropriate (Figure 1). Of those, 87.8% (337/384) were due to polyp-based criteria, 11.7% (45/384) were due to family history–based criteria, and 0.5% (2/384) were due to other reasons such as incomplete prep. During the transition phase, 66.7% (359/538) were nonguideline, with 88.6% (318/359) due to polyp-based criteria, 9.47% (34/359) due to family history-based criteria, and 2% (7/359) as other. In the postimplementation phase, 56.5% (369/653) were nonguideline, with 81.3% (300/369) due to polyp-based criteria, 17.1% (63/369) due to family history–based criteria, and 1.1% (4/369) due to other reasons. There were 3 patients that were recommended for a 3-year interval because of polyp characteristics that were inappropriately flagged for a 5-year interval. The proportion of nonguideline-appropriate colonoscopies decreased between the preimplementation and postimplementation phases, as well as the percent of cases where it was nonguideline because of polyp-based criteria. In terms of communication with patients and PCP, out of the 1063 patients whose EHR flag was adjusted, 4 patients or their PCPs (0.4%) directly reached out to the gastroenterology team in response to the update letter that was sent, to ask clarification questions or to provide further clinical information. No office visits with gastroenterology were requested or scheduled as a result of these communications. Furthermore, in meeting with a representative group of PCPs, there were no concerns expressed following the implementation of this project. Figure 2 shows the percent appropriateness of 5-year surveillance colonoscopies by week throughout the study. There was a statistically significant 10.9% increase in the average percent appropriateness of cases per week before and after the intervention (P < .01).

Table.

Number of Cases in Each Phase

Category	Preimplementation	Transition	Postimplementation
Total number of cases	5792	7386	5863
Number of screening/surveillance colonoscopy cases	3303	4189	3299
Number of 5-y screening/surveillance colonoscopy cases	555	538	653
Number of 5-y screening/surveillance cases/outpatient day	9.4 cases/d	7.27 cases/d	9.6 cases/d

Figure 1.

Percent guideline and reason for nonguideline during each phase.

Figure 2.

Change in percent guideline appropriateness across the weeks through different phases.

Discussion

This quality-improvement study consisted of a time-limited intervention in which we assessed the impact on the proportion of patients scheduled for colonoscopy at an appropriate 5-year screening/surveillance interval. The intervention of manually changing EHR flags to reflect updated guidelines led to a statistically significant 10.9% increase in the average percent appropriateness per week, before and after implementation. Given the concomitant decrease in percentage of cases where polyp-based criteria were the reason for inappropriateness, this shows that the intervention improved appropriateness by reflecting updates in polyp guidelines.

Overall, this study highlights the time lapse in implementing guidelines into clinical practice. We suspect that despite guidelines changing almost 5 years ago, many patients will still be recommended colonoscopies at a time interval that is not in line with the 2020 guidelines, which can lead to overuse of colonoscopies.⁷ Our findings are also in line with a 2019 systematic review that summarized that screening colonoscopies are often recommended at shorter intervals than guidelines recommend.⁸ This may also be due to gaps between guidelines and actual clinical practice despite EHR recommendation, ultimately due to shared decision-making between a patient and their provider. These studies suggest a need for implementation of strategies that generally encourage guideline-based care and allow for quicker adaptation of guidelines, particularly as guidelines continue to change in the future.

Our intervention intended to focus on an aspect of the colonoscopy scheduling process that was easily accessible and intervenable but did not deter from the PCP-patient therapeutic relationship. This meant that the intervention focused on interventions that would be reflected in upcoming orders rather than rescheduling already scheduled colonoscopies.

A strength of our study is that the EHR flag and data allowed for thorough investigation of indication and appropriateness of cases. The high number of patients we evaluated limits confounding factors such as preference of specific PCPs. Limitations of our study include the fact that given our intervention was primarily via EHR, there is no guarantee that the patient or the provider referenced this while ordering the colonoscopy. The study population was also limited to patients established with a PCP in our system. We did not assess how often PCPs, despite the flag, ordered a screening colonoscopy earlier than the recommended interval, but we suspect this would be unusual given the common workflow. We also did not assess cancellations or rescheduling of colonoscopies by the patient. We did not aim to address patients scheduled with inaccuracies in family history assessments or no EHR flag (includes patients with no prior colonoscopies in our system and/or PCPs outside the system), as these were often not accurately reflected in charts. Although this was, by no means, a sustainable intervention in long term, an EHR change without the need for direct communication to the PCP, when placed early in the workflow, did show to be an effective method of intervention. Even with implementation of this intervention, the endoscopy suite continued to operate at full capacity. We predict that given it has been 5 years since the updated guidelines, we should continue to see changes reflected in the colonoscopies scheduled in the coming years without a sustained intervention.

Conclusion

Ensuring patients are on the proper interval for CRC screening follow-up is important for minimizing health care harm and optimizing the capacity of centers for necessary procedures. This study suggests our focused EHR related intervention can be effective in decreasing unindicated procedures. Next steps in interventions include more high-cost but high-yield steps such as including functions in the EMR reminding providers of guidelines personalized to a patient’s history, offering point-of-care and updated guidance to PCPs, and allowing for the evaluation of appropriateness before scheduling procedures. This study highlights the need for active interventions to close the gap between guideline suggestions and real-world care provided, as this is applicable in many clinical scenarios.