D0172 - Validation of an Automated Adenoma Detection Rate

Tuesday, October 25, 2022

10:00 AM – 12:00 PM ET

Location: Crown Ballroom

Presenting Author(s)

McKenzie Needham, MS

Wake Forest University School of Medicine
Winston-Salem, NC

McKenzie Needham, MS¹, David Burns, MS¹, Jason Conway, MD², Heather Duncan, RDCS³, Jared Rejeski, MD¹
¹Wake Forest University School of Medicine, Winston-Salem, NC; ²Wake Forest Baptist, Winston-Salem, NC; ³Atrium Health Wake Forest Baptist, Winston-Salem, NC

Introduction: Adenoma detection rate (ADR) is an internationally recognized benchmark in the performance of colonoscopy, representing an important endoscopist quality metric. While ADR is a simple and widely accepted measure, the process of determining ADR is labor intensive. With the advent of endoscopy software within electronic health record (EHR) systems, ADR can be calculated automatically. However, the method of ADR calculation is inaccurate and therefore we aimed to compare automated ADR based on coding parameters from Lumens software within EPIC with manually calculated ADR.

Methods: We performed an IRB approved analysis of colonoscopy data from 2/1/22-5/10/22 at our institution. All outpatient colonoscopies were evaluated for indication, endoscopic findings, and histology. Manual ADR calculation was performed as per the CMS 2019 definition. We also extracted ADR data and the previously validated cecal withdrawal times available within the EHR. Automated ADR was performed within the EHR by extracting exams performed for the z12.11 and z12.12 indications. Data from individual endoscopists was made anonymous then normalized and compared with a single sample T-test; aggregate ADR rates were compared using Chi-squared analysis.

Results: Over this 98-day period, there were 1,737 colonoscopies performed. After identifying screening exams as per CMS guidelines and exclusion of seven endoscopists with less than 30 colonoscopies performed over this period, the manual calculation of ADR included 688 colonoscopies performed in patients ≥45 and 505 colonoscopies in patients ≥50. The automated ADR calculations in the EHR included a total of 503 exams in individuals ≥50. An adequate prep was seen in 92.2% of cases with an average BBPS of 7.3. Average cecal withdrawal time in screening exams was 9 minutes, 37 seconds and was not correlated with ADR. While ADR is reported in individuals ≥50 years as per CMS, the overall ADR between those ≥50 was 4.0% higher when compared to ≥45 (35.8% vs 31.8%, respectively), though this did not meet statistical significance (95%CI -1.4% to 9.4%). Comparison of manual ADR and automated ADR by endoscopist showed no significant differences. A comparison of manual ADR (35.8%) showed no significant difference from the automated ADR calculation (34.9%), p=0.79.

Discussion: An automated ADR will not strictly adhere to CMS definitions of ADR, but this index appears to be an adequate surrogate marker of ADR, simplifying an otherwise time-intensive process.

Endoscopist	True ADR	Automated ADR
1	41.5%	27.0%
2	34.3%	25.0%
3	40.0%	42.4%
4	37.5%	45.5%
5	31.7%	43.2%
6	37.5%	44.8%
7	28.9%	21.7%
8	30.0%	18.8%
9	42.9%	54.2%
10	30.4%	44.1%
11	47.6%	29.4%
12	25.6%	26.4%
Combined	35.8%	34.9%
	T-statistic =0.846 (p = 20.8)

Table: Comparison of the absolute values of manually calculated ADR ("True ADR") with the EHR-derived "Automated ADR". While individual differences are present, these differences were statistically insignificant. Furthermore, when all data was combined, the two values were within one percentage point.

Disclosures:

McKenzie Needham indicated no relevant financial relationships.

David Burns indicated no relevant financial relationships.

Jason Conway: Cook Medical – Consultant.

Heather Duncan indicated no relevant financial relationships.

Jared Rejeski: Nevro Inc – Grant/Research Support.

McKenzie Needham, MS¹, David Burns, MS¹, Jason Conway, MD², Heather Duncan, RDCS³, Jared Rejeski, MD¹. D0172 - Validation of an Automated Adenoma Detection Rate, ACG 2022 Annual Scientific Meeting Abstracts. Charlotte, NC: American College of Gastroenterology.