Adenoma Detection Rates
Adenoma Detection Rates
Our study shows that there is significant variation in advanced adenoma detection among colonoscopists. The results also showed that increasing patient age, male gender, and colonoscopist were associated with advanced adenoma detection. Furthermore, there was lack of positive correlation by Spearman rank-order analysis between nonadvanced and advanced ADRs. Three of the five colonoscopists with the lowest advanced ADRs were not among the five colonoscopists with the lowest ADRs, and two of the five had ADRs greater than the average. At present, global ADR (including advanced and nonadvanced adenomas) has become the accepted detection rate quality marker. However, because there was a colonoscopist effect, even when adjusting for other variables associated with advanced ADR, and there was absence of positive correlation of the colonoscopists' advanced and nonadvanced adenoma detection in our study, it suggests that advanced adenoma detection might be given further consideration as a distinct quality marker in colonoscopy in addition to the overall ADR.
Concern of utilizing the advanced ADR as a marker focuses on a few issues, the first being that physicians might "game the system"—or size polyps larger than they truly are to define an adenoma as advanced by size. With regard to oversizing of adenomas, or the concern of colonoscopists "gaming the system," multiple studies have shown that advanced ADRs are variable. We demonstrated that colonoscopic oversizing of polyps cannot explain the lack of correlation seen between advanced and nonadvanced ADRs in this study.
The second concern is that the association of colonoscopists' advanced ADR with their ADR is unclear. However, our findings show that the colonoscopists' advanced ADR does not appear to be redundant or entirely reflected in their global ADR. It should be noted that although Kaminski et al. showed ADR was a predictor of interval cancer, no study has looked at advanced adenoma detection impact on interval cancer. Thus, the optimal metric to define reduction of interval cancers is not known. Moreover, it should be noted that in this same study by Kaminski et al., ADRs were quite low (<20%) in the majority of colonoscopists, which is not reflective of most North American colonoscopy practices.
There is a clinical concern that by not considering advanced adenomas as a quality marker, a colonoscopist may miss large, advanced polyps, but be adept at detecting smaller ones. Thus, such a colonoscopist would have an acceptable ADR, whereas he or she may have missed key lesions further along in the dysplastic progression pathway to colorectal cancer. This concern stems from prior studies showing variable advanced adenoma detection miss rates. Although an acceptable threshold of satisfactory adenoma detection has been established at 20% in the colonoscopy screening literature, there is no clearly established threshold for detection of advanced adenomas among patients undergoing screening colonoscopy. There is a minimum reported prevalence of ~5%, and reported advanced ADRs of ~4.3–5.9%. In our study, the overall advanced ADR was 7.97%, and 5 (36%) colonoscopists had rates of advanced adenoma detection <5%. Thus, our study suggests a scenario where a colonoscopist detects adenomas globally at an acceptable rate, but advanced adenoma detection may be suboptimal.
A key issue in considering the utilization of the advanced ADR as a quality marker is that the acceptable advanced ADR has not been defined. Moreover, as advanced adenomas are a less frequent occurrence, a large number of colonoscopies would need to be reviewed to assess an individual physician's advanced adenoma detection capabilities. For instance, for a 5% advanced ADR not to include a 3% detection rate for 95% power with 95% confidence limits, a physician would need to perform colonoscopy on approximately 1,200 patients.
There are some limitations of our study. Although the population per colonoscopist was generally similar, and age, race, and gender were adjusted for in our analysis, our study did not take into account the differences in the number of patients per colonoscopist with a family history of colorectal cancer, a patient's smoking habits, or a patient's aspirin or non-steroidal anti-inflammatory history, all of which have been shown to have predictive value in adenoma risk. Withdrawal times were also not taken into account, which have been shown to significantly affect ADRs and are of undetermined significance at this time with respect to advanced ADRs.
One important strength of this study is the restrictive criteria of selected patients to aid in standardized observation. For example, patients with prior inflammatory bowel disease, gastrointestinal bleeding, and any prior colonoscopy, which could change the nonadvanced to advanced adenoma composition, were excluded. Patients with suboptimal bowel preparations (fair and poor) were also excluded as the ratio of advanced to nonadvanced adenoma detection by colonoscopy is affected by suboptimal bowel preparations, is not clearly defined, and not addressed in this study. The cohort is what would be considered an average risk for colorectal cancer and ideal colonoscopy conditions. Although colonoscopists have been shown to have variable ability to accurately size polyps, we demonstrated that colonoscopic oversizing of polyps cannot explain the lack of correlation seen between colonoscopists' advanced and nonadvanced ADRs in this study.
In conclusion, we found that the colonoscopist affects advanced adenoma detection even when adjusting for patient age, gender, race, year of colonoscopy, gastroenterology fellow participation during colonoscopy, and nonadvanced adenoma detection. We also demonstrated a marked variation in advanced ADRs in an average risk cohort. Moreover, colonoscopists' advanced ADR did not reflect their nonadvanced ADR. Future studies specifically defining acceptable advanced ADRs will need to be conducted to further consider the advanced adenoma as a colonoscopy quality indicator.
Discussion
Our study shows that there is significant variation in advanced adenoma detection among colonoscopists. The results also showed that increasing patient age, male gender, and colonoscopist were associated with advanced adenoma detection. Furthermore, there was lack of positive correlation by Spearman rank-order analysis between nonadvanced and advanced ADRs. Three of the five colonoscopists with the lowest advanced ADRs were not among the five colonoscopists with the lowest ADRs, and two of the five had ADRs greater than the average. At present, global ADR (including advanced and nonadvanced adenomas) has become the accepted detection rate quality marker. However, because there was a colonoscopist effect, even when adjusting for other variables associated with advanced ADR, and there was absence of positive correlation of the colonoscopists' advanced and nonadvanced adenoma detection in our study, it suggests that advanced adenoma detection might be given further consideration as a distinct quality marker in colonoscopy in addition to the overall ADR.
Concern of utilizing the advanced ADR as a marker focuses on a few issues, the first being that physicians might "game the system"—or size polyps larger than they truly are to define an adenoma as advanced by size. With regard to oversizing of adenomas, or the concern of colonoscopists "gaming the system," multiple studies have shown that advanced ADRs are variable. We demonstrated that colonoscopic oversizing of polyps cannot explain the lack of correlation seen between advanced and nonadvanced ADRs in this study.
The second concern is that the association of colonoscopists' advanced ADR with their ADR is unclear. However, our findings show that the colonoscopists' advanced ADR does not appear to be redundant or entirely reflected in their global ADR. It should be noted that although Kaminski et al. showed ADR was a predictor of interval cancer, no study has looked at advanced adenoma detection impact on interval cancer. Thus, the optimal metric to define reduction of interval cancers is not known. Moreover, it should be noted that in this same study by Kaminski et al., ADRs were quite low (<20%) in the majority of colonoscopists, which is not reflective of most North American colonoscopy practices.
There is a clinical concern that by not considering advanced adenomas as a quality marker, a colonoscopist may miss large, advanced polyps, but be adept at detecting smaller ones. Thus, such a colonoscopist would have an acceptable ADR, whereas he or she may have missed key lesions further along in the dysplastic progression pathway to colorectal cancer. This concern stems from prior studies showing variable advanced adenoma detection miss rates. Although an acceptable threshold of satisfactory adenoma detection has been established at 20% in the colonoscopy screening literature, there is no clearly established threshold for detection of advanced adenomas among patients undergoing screening colonoscopy. There is a minimum reported prevalence of ~5%, and reported advanced ADRs of ~4.3–5.9%. In our study, the overall advanced ADR was 7.97%, and 5 (36%) colonoscopists had rates of advanced adenoma detection <5%. Thus, our study suggests a scenario where a colonoscopist detects adenomas globally at an acceptable rate, but advanced adenoma detection may be suboptimal.
A key issue in considering the utilization of the advanced ADR as a quality marker is that the acceptable advanced ADR has not been defined. Moreover, as advanced adenomas are a less frequent occurrence, a large number of colonoscopies would need to be reviewed to assess an individual physician's advanced adenoma detection capabilities. For instance, for a 5% advanced ADR not to include a 3% detection rate for 95% power with 95% confidence limits, a physician would need to perform colonoscopy on approximately 1,200 patients.
There are some limitations of our study. Although the population per colonoscopist was generally similar, and age, race, and gender were adjusted for in our analysis, our study did not take into account the differences in the number of patients per colonoscopist with a family history of colorectal cancer, a patient's smoking habits, or a patient's aspirin or non-steroidal anti-inflammatory history, all of which have been shown to have predictive value in adenoma risk. Withdrawal times were also not taken into account, which have been shown to significantly affect ADRs and are of undetermined significance at this time with respect to advanced ADRs.
One important strength of this study is the restrictive criteria of selected patients to aid in standardized observation. For example, patients with prior inflammatory bowel disease, gastrointestinal bleeding, and any prior colonoscopy, which could change the nonadvanced to advanced adenoma composition, were excluded. Patients with suboptimal bowel preparations (fair and poor) were also excluded as the ratio of advanced to nonadvanced adenoma detection by colonoscopy is affected by suboptimal bowel preparations, is not clearly defined, and not addressed in this study. The cohort is what would be considered an average risk for colorectal cancer and ideal colonoscopy conditions. Although colonoscopists have been shown to have variable ability to accurately size polyps, we demonstrated that colonoscopic oversizing of polyps cannot explain the lack of correlation seen between colonoscopists' advanced and nonadvanced ADRs in this study.
In conclusion, we found that the colonoscopist affects advanced adenoma detection even when adjusting for patient age, gender, race, year of colonoscopy, gastroenterology fellow participation during colonoscopy, and nonadvanced adenoma detection. We also demonstrated a marked variation in advanced ADRs in an average risk cohort. Moreover, colonoscopists' advanced ADR did not reflect their nonadvanced ADR. Future studies specifically defining acceptable advanced ADRs will need to be conducted to further consider the advanced adenoma as a colonoscopy quality indicator.
Source...