Sizing of Polyp Illustrations Differs by Endoscopists’ Gender and Improves With a Measurement Reference

Background

Colonoscopy is a mainstay of colorectal cancer screening, effectively identifying cancer and allowing for the removal of precancerous polyps. When polyps are identified and removed, national guidelines recommend using polyp size, number, and dysplasia to determine appropriate timing of a repeat colonoscopy (Howlader et al., 2016). Specifically, polyps greater than or equal to 10 mm in diameter require closer surveillance (Howlader et al., 2016). It has also been suggested (Lieberman, Moravec, & Holub, 2008; Lieberman et al., 2015) that polyps less than 5 mm in diameter rarely show significantly malignant pathology and can be assessed differently than larger polyps.

These recommendations rely on an accurate assessment of polyp size. Unfortunately, multiple studies (Chaptini et al., 2014; Fennerty, Davidson, & Emerson, 1993; Izzy et al., 2015; Lieberman et al., 2015) have shown significant challenges in measuring polyps, for both pathologists and endoscopists. Pathologists may receive partial or fragmented tissue samples, and polyps can change shape when resected and preserved, making direct measurement unreliable (Lieberman et al., 2015). Endoscopists rely on visual estimation to assess polyp size, and evidence suggests that endoscopists routinely over- or underestimate the size of polyps at colonoscopy (Chaptini et al., 2014; Elwir, Shaukat, Shaw, Hughes, & Colton, 2017; Fennerty et al., 1993; Izzy et al., 2015). The association of endoscopist age, gender, level of experience, or adenoma detection rate (ADR) on polyp sizing accuracy has not been studied. Use of adjacent structures or tools to estimate polyp size is problematic, as endoscopists often don’t know the size of their marker for comparison. Some success has been shown in using graduated snares or forceps to assess polyp size (Gopalswamy et al., 1997; Hyun, Han, Bae, Park, & Eun, 2011); however, these specific tools limit the equipment endoscopists can use and add time and cost to the procedure (Rex & Rabinowitz, 2014). There are data that suggest printed materials may have a benefit on practice outcomes (Giguere et al., 2012), leading to the hypothesis that a printed measuring guide would lead to improvement in endoscopist accuracy. The goal of this study was to assess the accuracy of measuring polyp illustrations at baseline and then following the introduction of a polyp measuring guide and study differences by physician age, gender, level of experience, and ADRs.

Method

Study participants included all St. Paul endoscopists in a large, community-based practice (Minnesota Gastroenterology, PA) in Minneapolis/St. Paul, MN. All included endoscopists perform routine colonoscopy in one of the two ambulatory surgical centers located in St. Paul, MN. Every endoscopist was given a 17-page booklet with a solid, black polyp illustration on each page. Endoscopists were asked to estimate the size of the illustration on the page. The illustrations were of known diameter and ranged in size from 2 to 31 mm. As 5 and 10 mm are important thresholds for determining surveillance, there were three 5 mm and three 10 mm illustrations present in the booklet. No other size was repeated. The illustrations were presented in a random order, and each booklet was identical. The booklet was given to each endoscopist at the same time, in the same location, and endoscopists were asked not to confer with each other.

One month after initial data collection was complete, the process was repeated with an intervention. A polyp-measuring guide was included on each page of the booklet. The polyp-measuring guide was an illustration of a fully opened regular snare (ConMed Small Snare®) containing 5, 10, and 15 mm polyp illustrations within the open snare. The ConMed Small Snare was chosen as a reference because it is the most frequently used device for snare polypectomy for this practice. Accuracy was defined as no difference between estimated and actual value. We also calculated 10% and 20% over- and underestimation in sizing, based on similar studies from the literature (Gopalswamy et al., 1997; Izzy et al., 2015). We used student t test or analysis of variance to compare differences in means and χ² test to compare differences in categorical variables where appropriate. A significance level of p < .05 was used for all comparisons (two-tailed). Estimating size with and without the polyp-measuring guide was assessed for the total group of physicians and then further divided by demographics: male, female, average number of colonoscopies per year, years in practice, and ADR.

Results

A total of 29 endoscopists participated in the study, yielding 986 data points for comparison. Twenty-one (72%) were male, and eight (28%) were female. The average age was 51, and endoscopists had an average of 18.2 years in practice. The total number of colonoscopies completed by each endoscopist, per year, was 682. The average ADR of the endoscopists was 39. At baseline, endoscopists’ measuring accuracy was 15%, with most endoscopists undersizing by 58%. With the reference intervention, accuracy improved to 50%—an improvement of 35% (p < .001; Table 1). When specifically looking at 5 and 10 mm polyp illustrations, baseline accuracy was 21% and 23%, respectively, and improved to 80% and 68%, respectively, with the polyp-measuring guide (p < .001). The average estimation of the 5-mm circles improved from 3.5 to 4.9 mm, while the average estimation of the 10-mm circles improved from 7.7 to 10.1 mm. Following intervention, the majority of measurements were accurate (Table 1).

OPEN IN VIEWER

Table 1.

Accuracy of Measurements.

Estimated Value	Baseline	With Reference Intervention	p Value
Accurate	15%	50%	.001
Accurate within 10%	25%	67%	.001
Accurate within 20%	47%	89%	.001
Underestimated	58%	19%	.001
Underestimated within 10%	55%	14%	.001
Underestimated within 20%	37%	4%	.001
Overestimated	26%	31%	.06
Overestimated within 10%	20%	20%	.9
Overestimated within 20%	16%	7%	.001
5 mm Measurement
Average value (mm)	3.5	4.9	.001
Accurate	21%	80%	.001
Underestimated	68%	10%	.001
Overestimated	11%	10%	.23
10 mm Measurement
Average value (mm)	8%	10.1	.001
Accurate	23%	68%	.001
Underestimated	60%	13%	.001
Overestimated	17%	19%	.02

Accuracy within 10% margin of error at baseline was 25% and improved to 67% with the polyp-measuring guide (p < .001). Accuracy within 20% within margin of error improved from 47% to 89% (p < .001; Table 1). The number of underestimated measurements decreased by 39% (p < .001), and there was no statistically significant change in the number overestimated.

The effect of the polyp-measuring guide was further assessed based on demographic characteristics of the endoscopists (Table 2). When compared to their baseline, male endoscopists were more accurate than females by 12% (p = .001); males underestimated size 54% of the time, whereas females underestimated 69% of the time. Endoscopists in practice more than 10 years were accurate 18% of the time, while endoscopists with less than 10 years in practice were accurate 11% of the time (p = .04). Physicians with ADR less than 38% (the average ADR for the group) were accurate 20% of the time, and those with ADR greater than 38% were accurate 12% of the time (p = .02). There was no statistically significant difference in accuracy based on number of colonoscopies. Following the polyp-measuring guide intervention, the difference between males and females was no longer statistically significant. Similarly, the differences between endoscopists with different ADRs or years in practice were no longer statistically significant (Table 2).

OPEN IN VIEWER

Table 2.

Accuracy of Measurement by Physician Demographics.

Demographic Group	No Reference Guide			With Reference Guide			% Change
	% Accurate	% Under	% Over	% Accurate	% Under	% Over
Total group	15	58	26	50	19	31	35
Male	19	54	27	52	20	28	33
Female	7	69	24	45	17	38	39
≤10 years in practice	11	65	24	60	9	31	48
>10 years in practice	18	55	27	46	24	31	28
ADR < 38	20	49	32	48	21	30	29
ADR > 38	12	66	22	52	17	31	40

Note. ADR = adenoma detection rate.

Discussion

The aim of this study was to assess endoscopists’ accuracy-measuring polyp illustrations with and without a polyp-measuring guide. We found that 85% of measurements were incorrect at baseline, but the use of a nearby polyp-measuring guide increased measurement accuracy by 35%. Although the literature supports (Chaptini et al., 2014; Fennerty et al., 1993) both over- and undersizing of polyps, we found that endoscopists almost universally underestimated rather than overestimated the size of polyp illustrations. Clinically, 5 and 10 mm measurements are important as they can change surveillance intervals. Overall, accuracy of 5 and 10 mm circles at baseline was incorrect 80% of the time but improved significantly with the polyp-measuring guide. Average measurements of the 5 mm circles improved from 3.5 to 4.9 mm, and the 10-mm circles improved from 7.7 to 10.1 mm.

We also found that female endoscopists compared to male endoscopists were less accurate by 12% and underestimated size significantly more often than male endoscopists, 69% versus 54% (p < .001). This observation needs to be confirmed in other studies, and the underlying factors need to be explored.

Use of a polyp-measuring guide improved endoscopists’ measuring accuracy independent of gender, experience, and ADR. Noteworthy is the 48% improvement in accuracy in physicians with less than 10 years of experience. Improvement in less experienced and female endoscopists suggests that a polyp-measuring guide would be especially helpful for training programs, particularly now that they include a higher percentage of women.

It is possible that informing endoscopists that we were interested in their ability to estimate size had an impact on their behavior. We attempted to limit this effect as much as possible by administering the booklet to all endoscopists at the same time and encouraging them not to confer with each other. There was no other discussion of polyp sizing or its importance in the month interim between the two assessments.

This study is also limited by the assumption that estimating the measurement of an illustration of a polyp is similar to estimating polyp size during endoscopy. Despite the artificial environment, it is encouraging that the accuracy of baseline measurement in our cohort was similar to prior studies (Chaptini et al., 2014; Izzy et al., 2015) showing endoscopists estimating polyp size within a 20% margin of error about 48% of the time. Another limitation is the fact that the study population is comprised of physicians from a single community-based practice. It is possible that this data would change in a hospital or academic center.

Strengths of the study include the objective nature of the measurements. As prior studies have shown, it is very difficult to develop a gold standard baseline polyp measurement. In this study, the estimated illustrations of polyps were of known diameter, allowing for a definitive baseline to judge accuracy. Additionally, this study limits confounding factors between enodoscopists by using individuals as their own controls.

Although measuring illustrations is not directly comparable to measuring actual polyps during colonoscopy, this pilot study provides insight into inherent cognitive and perceptual differences between men and women, young and old. There is a robust and growing literature in the brain sciences (Halpern, 2012) examining then innate differences in the cognitive abilities between genders. Our study adds to this literature and will help guide future endoscopic studies exploring these issues. Our group has already embarked on a large in vivo study assessing the effects of a polyp-measuring guide on the measurements of polyps during colonoscopy in a large community practice; issues of gender and years in practice will be examined.