| Background, purpose and significanceColorectal cancer is the third most common malignancy worldwide. The increased incidence of colorectal cancer in China in recent years has been attributed to improved living standards and changing dietary habits. It is the third most prevalent and the fifth most deadly cancer in China, posing serious harm to human health. Colorectal adenomas are the most important precancerous lesions of colorectal cancer and affect more than 20 percent of the general population over 50 years of age in Europe. The increased Westernization in China in recent years has contributed to a significant escalation in the incidence of colorectal adenomas. Prompt detection and early endoscopic resection of these adenomas is most effective in preventing colorectal cancer. Colonoscopy is widely recognized as the gold standard for colorectal cancer screening. An abundance of evidence has shown that screening colonoscopy has reduced the overall incidence of colorectal cancer by nearly 80 percent, leading to significantly lower left-sided colon cancer incidence and a decline in right-sided colon cancer incidence by 50%. Moreover, screening colonoscopy has also significantly reduced the incidence of colorectal cancer regardless of clinical stage and age.However, the rate of persistent interval cancers or colorectal cancer derived from missed lesions is closely related to the quality of colonoscopy. Huge differences exist in the quality of colonoscopy screening in different countries. Differences in endoscopic screening practices and programs within the same facility, and among different medical organizations and endoscopists largely relate to the following areas. First, adenoma detection rates (ADR) vary greatly among different screening programs (range 20%-46.5%) and among different endoscopists (range 21%-86%). Second, in screening colonoscopy, adenoma miss rate (AMR) for adenomas≥ 10 mm in size is 2%-12%,10%-15% for adenomas measuring between 6 and 9 mm, and 20%-30% for adenomas not larger than 5 mm in size. Moreover, the miss rate for colorectal cancer with screening colonoscopy is also 2%-6%.Missed adenomas will lead to the development of interval colorectal cancer. Previous studies have found that more than half of the interval cancers are derived from missed adenomas, rather than new lesions. Colorectal cancer attributed to missed lesions following screening endoscopy occurs in 0.5 to 3.5 per 1,000 screened persons. Previous studies have found that the ADR of endoscopists is significantly correlated with the risk of interval colorectal cancer. A lower ADR results in a significantly increased incidence, suggesting that ADR is an independent risk factor of interval colorectal cancer after screening colonoscopy. Therefore, ADR and AMR are considered as indicators of the quality of colonoscopy. Increasing ADR and reducing AMR can ensure the quality of colonoscopy and effectively reduce the development of colorectal cancer.The differences in ADR and AMR depend on the colonoscopy technique and are affected by factors such as medical background of patients, procedure-related parameters of colonoscopy, and clinicopathological features of potential adenomas. Studies have shown that differences in ADR and AMR are associated with the experience and professional background of endoscopists, withdrawal time, cecal intubation rate, auxiliary observation and imaging methods, scale or capacity of colonoscopy operating organizations (hospitals), and bowel preparation, all of which affect the quality of colonoscopy. In addition, a number of objective factors such as sex, age, family history of colon cancer, smoking history, and size and morphology of adenomas also affect adenoma detection and contribute to misdiagnosis.Therefore, a number of screening criteria have been developed to ensure quality control in endoscopy, including:cecal intubation rate≥90%; withdrawal time of negative colonoscopies exceeding 6 min; adequate bowel preparation in at least 90% of screened patients; availability of skilled colonoscopists who are required to maintain an annual experience of at least 150 colonoscopies; ADR should reach at least 20% for screening colonoscopy, including 25% in men and 15% in women. In addition, mandatory criteria exist for polyp retrieval, identifying colonoscopy-related complications, and ensuring patient comfort. Although currently available quality assurance systems for large-scale screening colonoscopy programs simply pool together the results of previous studies, it is undeniable that these systems are somewhat comparable and play an important role in controlling the quality of screening.These systems are based on research findings with detailed operational standards, and are mainly used for the evaluation of overall quality of colonoscopy performed by a few endoscopists and operating organizations over a period of time. However, they have the following challenges. First, indicators such as ADR, bowel preparation and cecal intubation rate are indicators for retrospective evaluation and cannot provide immediate and individual guidance, quality control and quality assessment for a particular colonoscopy. Second, it takes considerable time to train colonoscopists and maintain their skills, which is a factor that cannot be tailored to the requirements of any individual colonoscopy. With a growing number of people who require colonoscopy in recent years, there is a relative shortage of colonoscopists. Third, the universal requirement of at least 6 min of withdrawal time for all colonoscopists fails to account for the individual differences of colonoscopists and bowel conditions. Lengthy withdrawal time increases patient discomfort, risk of complications and cost whereas unduly short withdrawal time easily leads to missed diagnosis, which may be unreasonable. Accordingly, existing quality assurance systems are clearly inadequate for individual guidance and quality evaluation.In fact, the overall quality of screening colonoscopy is ensured by the quality of every individual colonoscopy. Therefore, based on previous research, it is necessary to consider some of the factors influencing ADR and AMR of colonoscopy, such as endoscopist’s experience, bowel preparation, and withdrawal time, and provide intervention and guidance in individual colonoscopy. A few pertinent indicators are integrated to establish an evaluation system for the quality of individual colonoscopy in an effort to provide prompt quality assessment. This system is the first of its kind that will help to improve the quality of individual colonoscopy, to provide objective, timely and individual guidance for follow-up and interventions in screened individuals. The approach will enhance the overall quality of screening colonoscopy in the general population, preventing and reducing the incidence of colorectal cancer.. This retrospective study examined large amounts of colonoscopy data from Chinese hospitals at different levels. Based on previous research on independent factors influencing ADR and AMR of colonoscopy, this study was undertaken to determine the reasonable withdrawal time for individual colonoscopy in the context of different endoscopist experience and bowel conditions and develop feasible measures for personalized interventions. This study was also designed to screen for acceptable indicators to establish a quantitative scale of prompt assessment of the quality of individual colonoscopy.Patients and Methods1. Data sourcesThis study included patients who received colonoscopies at Nanfang Hospital, Southern Medical University, Guangzhou, Guangdong Province; Wuxi People’s Hospital Affiliated to Nanjing Medical University, Wuxi, Jiangsu Province; Mianyang Central Hospital, Sichuan; and Longgang Central Hospital of Shenzhen, Guangdong Province. We searched colonoscopy and colonoscopic biopsy data of these four hospitals from September 2009 to September 2011 by logging on their electronic recording and reporting systems on endoscopy and clinical pathology. We constructed different databases by screening patients who received colonoscopy according to the inclusion and exclusion criteria. Inclusion criteria included:(1) age greater than 20 years; (2) detailed clinical pathological and endoscopic data; (3) patients who never underwent chromoendoscopy of the entire colon; (4) colon photographic data fulfilling the following requirements:pictures of the ileocecal valve and appendix hole were taken when the scope reached the cecum; pictures of the rectum were taken at the end of withdrawal, and pictures of various segments of the colon were taken to reflect the whole of the colon. Exclusion criteria included patients in whom the cecum was not intubated in colonoscopy, patients with colon cancer, polyposis syndrome, inflammatory bowel disease, or partial bowel resection, and patients with incomplete information.2. Data collectionResearch data included general information (such as patient ID, colonoscopy number, time of examination, patient name, operating hospitals), patient-related parameters (such as gender, age, symptoms, family history of colon cancer, family history of adenomas, presence of diverticulum), colonoscopy-related parameters (such as endoscopist experience and professional background, withdrawal time, auxiliary observation and imaging methods, bowel preparation, sedative endoscopy) and clinicopathological characteristics of adenomas (such as size, location, shape, pathology, neoplasia grade and risk of adenoma).Definition of endoscopist experience:Endoscopist experience was categorized into four levels according to the number of colonoscopies performed:> 2000; 2000-1000; 1000-500; and< 500. If the number of colonoscopies performed by any given colonoscopist reached the higher level at some point during the study period, the experience would be adjusted accordingly.Definition of withdrawal time:Withdrawal time referred to the actual time between the time when the scope was introduced to the cecum and the time when the scope was withdrawn to the rectum during colonoscopy. The calculation of the two time points was based on the time points indicated on colonoscopy images. Data were identified for staining or biopsy performed during the period. In statistical analysis, withdrawal time of a colonoscopy was defined as the average withdrawal time of negative colonoscopies, i.e., the average withdrawal time of all negative colonoscopies for a certain endoscopist experience level for which a colonoscopist who performed the colonoscopy qualified.Definition of bowel preparation:This metric was classified into three levels. Excellent:there was little or no semi-solid residue; and there was clear liquid residue that could be aspirated and cleared. Good:there was semi-solid residue and clear liquid, which could be removed by washing and aspiration. Poor:there were solid and semi-solid residues which could not be cleared. Excellent or good bowel preparation was defined as adequate bowel preparation whereas poor bowel preparation was defined as inadequate bowel preparation. Definition of auxiliary observation methods:The auxiliary observation methods included observation under direct vision, rectal retroflexion, ascending colon with or without rectal retroflexion.Definition of imaging methods:Imaging methods were divided into ordinary observation, staining and/or NBI of suspicious lesions.ADR= number of patients with at least one detected adenoma/total number of patients who underwent colonoscopy."Per-adenoma" AMR= number of missed adenomas/total number of adenomas."Per-patient" AMR= number of patients with missed adenomas/total number of patients.3. Database creationAn adenoma detection database for colonoscopy was established to examine the potential factors influencing ADR and to investigate individual interventions of individual colonoscopy. A database of adenoma miss was created to investigate the potential influencing factors of AMR and perform individual assessment of the quality of individual colonoscopy.The criteria for patients in the adenoma detection database included:(a) The colonoscope should reach the cecum and colonoscopy was performed for diagnostic purposes, (b) The endoscopist for each colonoscopy should perform at least 100 negative colonoscopies with complete data at his/her level of experience, (c) The data of the first colonoscopy only was included in the database for patients who underwent repeat colonoscopy.The criteria for the database of adenoma miss:(a) An interval of less than 90 days between two colonoscopies for any patient, with the first colonoscopy being diagnostic for adenoma and the second therapeutic for adenoma removal, (b) Bowel preparation was determined as adequate (including excellent or good) in the second colonoscopy. (c) The number of adenomas identified in the second colonoscopy was equal to or more than the number found in the first colonoscopy. (d) Newly identified adenomas in the second colonoscopy were defined as missed adenomas and adenomas identified in the second colonoscopy that were the same as those in the first colonoscopy were defined as diagnosed adenomas. (e) Endoscopists who performed the second colonoscopy were required to have performed more than 1000 colonoscopies previously and at least 150 colonoscopies annually during the study period. The endoscopist for the initial colonoscopy should have performed at least 100 negative colonoscopies with complete data at his/her level of experience.To further establish the scale for individual assessment of the quality of individual colonoscopy, the database of adenoma miss was stratified by hospital, gender, and age based on whether or not adenomas were missed and randomly divided into two databases in a ratio of 4:1:1) a primary construction database used for creating a quality scale of individual colonoscopy; and 2) a validation database of adenoma miss. In the primary construction database, a missed adenoma was randomly chosen from each missed patient to represent the missed patient and an adenoma was randomly selected from each diagnosed patient to represent the diagnosed patient. All the data pertaining to these adenomas constituted a secondary construction database for quality scale of individual colonoscopy.4. Factors influencing ADR of colonoscopyIn the adenoma detection database, factors that affected ADR were screened using univariate analysis. The significant factors were then subjected to multivariate analysis according to the presence or absence of adenomas.5. Individual intervention of individual colonoscopyUsing the adenoma detection database, we further analyzed the relationship between ADR and related independent factors such as endoscopist experience, bowel preparation and the average negative withdrawal time. We explored the synergistic effect of different combinations of the three factors on ADR. A fitted curve was created to calculate individual withdrawal time needed to achieve the ADR levels for high-quality colonoscopy using different combinations of endoscopist experience and bowel preparation. Subsequently, patients in the database of detected adenomas were categorized according to endoscopist experience, bowel preparation and withdrawal time and adenoma detection results were validated.6. Factors influencing AMR of colonoscopyIn the database of adenoma miss, factors that affected AMR of colonoscopy were screened using univariate analysis, followed by multivariate analysis of the significant factors.7. Construction and validation of the evaluation scale for the quality of individual colonoscopyA quality scale of individual colonoscopy was created in the secondary construction database. First, all independent invariable factors including patient profile and pathologic parameters of adenomas were excluded. All the independent factors influencing the procedure were once again subjected to multivariate analysis mandatorily. A scale was constructed by assigning values to all factors according to partial regression coefficients of each factor. Finally, the scale was validated in the validation database of adenoma miss.8. Statistical analysisThe SPSS 17 statistical software was used for statistical analysis. Data are presented as mean ± standard deviation and categorical data are presented as frequency. P values≤ 0.05 were considered statistically significant. Univariate analysis of factors influencing ADR and AMR in colonoscopy was performed using the t test, rank sum test or chi-square test; and multivariate analysis of factors influencing ADR and AMR was conducted using logistic regression analysis. The effects of endoscopist experience, bowel preparation, and withdrawal time on ADR and corrected ADR of each colonoscopist was analyzed using linear or rank correlation, chi-square test and analysis of variance. Adjusted ADR was calculated after the adjustment of confounding factors such as sex and age by logistic regression analysis. The correlation between ADR and withdrawal time of different colonoscopists with different endoscopic experience and with different levels of bowel preparation was analyzed using linear regression. The fitted curve was obtained to further calculate individual withdrawal times required to attain the ADR level (20%) for high quality colonoscopy in the case of different levels of endoscopist experience and bowel preparation. Patients in the adenoma detection database were divided into different groups. The results were validated using the chi-square test. To create the quality scale, procedure-related independent factors affecting the quality of colonoscopy were subjected to multivariate logistic regression analysis, and the generated regression coefficients for each factor were divided by the smallest regression coefficient. The obtained values were rounded off to yield a score for each factor to create the evaluation scale for individual colonoscopy. Finally, scores were generated according to the scale for patients included in the primary construction database and the database used for AMR verification. The scores were divided into different ranges based on different AMRs. The relationship between the scores and AMR was analyzed using linear regression analysis. Differences between score ranges and AMR as well as linear trends were analyzed using the R x C table.Results1. Factors influencing ADRA total of 16,951 patients were included in this study, of which 2,938 were found to have adenomas, with a total of 5,052 adenomas detected (ADR 17.3%). The average number of adenomas detected per colonoscopy was 0.298 and the average number of adenomas detected per positive colonoscopy was 1.7. Logistic regression analysis showed that factors such as older age, male gender, intestinal symptoms, family history of colorectal cancer and adenoma, hospitals at higher levels, richer endoscopist experience, adequate bowel preparation, use of staining and/or NBI, observation with ascending colon retroflexion with/without rectal retroflexion, and longer withdrawal time were associated with higher adenoma detection.2. Individual colonoscopy interventionsRank correlation analysis was performed to examine the correlation between endoscopist experience levels and the ADR and sex-and age-adjusted ADR of colonoscopists at various levels of experience. The results showed that significant correlations existed between ADR and adjusted ADR with experience levels (r= 0.695, r=0.523, both P< 0.001) and that colonoscopists with richer experience tended to have a higher ADR and adjusted ADR. After stratification by excellent, good and poor bowel preparation, the correlations still remained significant (excellent bowel preparation:r=0.376, P=0.006; r=0.355, P=0.01; good:r=0.503, P <0.001; r=0.309, P=0.026; poor:r=0.503, P< 0.001; r=0.360, P=0.009)Linear regression analysis of the ADR and corrected ADR of colonoscopists at various levels of experience and withdrawal time revealed a significant regression between withdrawal time and ADR (r=0.525, r=0.616, both P< 0.001). After stratification by quality of bowel preparation, a significant regression still existed under excellent and good bowel preparation (excellent bowel preparation:r=0.370, P=0.007; r=0.379, P=0.006; good bowel preparation:r=0.438, P=0.001; r= 0.479, P< 0.001). Under poor bowel preparation, there was no regression between the ADR and adjusted ADR of colonoscopists at various experience levels and withdrawal time (r=0.077, r=0.230, both P> 0.05).There was no significant difference in the ADR of colonoscopists at various experience levels between excellent and good bowel preparation (P= 0.616), but both of these ADRs were significantly higher than that under poor bowel preparation (both P<0.001). The quality of bowel preparation was further stratified by adequacy and inadequacy and the following results were obtained:1) if bowel preparation was adequate and the number of colonoscopies performed by the colonoscopist was greater than 2000, a significant linear regression relationship was found between withdrawal time and adjusted ADR (r=0.854, P=0.001) and the withdrawal time was 4.2 min when the ADR was 20% as extrapolated from the regression line; 2) if bowel preparation was adequate and the number of colonoscopies performed by the colonoscopist ranged between 1000 and 2000, there was a significant linear regression relationship between withdrawal time and adjusted ADR (r=0.573, P= 0.032), and the withdrawal time was 5.7 min when the ADR was 20% as extrapolated from the regression line; 3) if bowel preparation was adequate and the number of colonoscopies performed by the colonoscopist was between 500 and 1000, there was a significant linear regression relationship between withdrawal time and adjusted ADR (r=0.776, P<0.001), and the withdrawal time was 6.1 min when the ADR was 20% as extrapolated from the regression line; 4) if bowel preparation was adequate and the number of colonoscopies performed by the colonoscopist was less than 500, there was a significant linear regression relationship between withdrawal time and adjusted ADR (r=0.871, P=0.002), and the withdrawal time was 8.7 min when the ADR was 20%as extrapolated from the regression line; 5) if bowel preparation was inadequate, the ADR of colonoscopists at all experience levels was all less than 20%. If the number of colonoscopies performed by the colonoscopist was greater than 2000, there was a significant linear regression relationship between withdrawal time and adjusted ADR (r=0.787, P=0.004), and the withdrawal time was 6.2 min when the ADR was 15% as extrapolated from the regression line. If the number of colonoscopies performed by the colonoscopist was between 1000 and 2000, there was a significant linear regression relationship between withdrawal time and adjusted ADR (r=0.542, P=0.045), and the withdrawal time was 6.3 min when the ADR was 15% as extrapolated from the regression line. If the number of colonoscopies performed by the colonoscopist was less than 1000, there was no significant linear regression relationship between withdrawal time and corrected ADR.3. Factors influencing AMR of colonoscopyA total of 2,093 patients were enrolled, including 560 patients in whom missed adenomas were detected in the second colonoscopy ("per-patient" AMR 26.8%, 560/2093). Two colonoscopies found a total of 4,658 adenomas and the second colonoscopy detected 967 new adenomas ("per-adenoma" AMR 20.8%,967/4658). Logistic regression analysis showed that factors like older age, lack of experience, inadequate bowel preparation, absence of staining and/or NBI, absence of retroflexion, short withdrawal time, small adenomas, non-colorectal adenomas, flat adenomas and low-risk adenomas were likely to lead to missed diagnosis.4. Creation and validation of the scale for the quality of individual colonoscopyOf the 2,093 patients,1,673 patients and 420 patients were included in the databases for scale construction and validation, respectively. Procedure-related independent factors influencing adenoma miss included bowel preparation, endoscopist experience, withdrawal time, imaging methods and retroflexion-based observation. According to the relationships among the partial regression coefficients of the above factors obtained using the logistic model, each factor was scored on a scale of 0 to 3, with 3 points for adequate bowel preparation and 0 point for poor bowel preparation; 2 points for an endoscopist who performed more than 1000 colonoscopies and 0 point for an endoscopist who performed less than or equal to 1000 colonoscopies; 2 points for withdrawal time equal to or greater than 6 min,1 point for withdrawal time equal to or greater than 4 min but less than 6 min, and 0 point for withdrawal time less than 4 min; 1 point if staining and/or NBI was performed and 0 point if staining and/or NBI was not performed; 1 point if ascending colon retroflexion and/or rectal retroflexion was performed, and 0 point if the retroflexion observation was not performed. The quality scale ranged from 0-9. In the validation database, linear regression analysis showed a significant linear regression between scores and AMR (r=-0.869, P<0.001, corrected R2=0.725).Conclusions1. Numerous factors affect adenoma detected and missed during colonoscopy, including those related to patient, colonoscopist, procedure and potential pathological features of adenomas, which warrant further study. A complete understanding of these potential factors by colonoscopists will help improve the quality of colonoscopy.2. Colonoscopists at different levels of experience require different withdrawal times in order to obtain satisfactory ADR under adequate bowel preparation. Timely repeat colonoscopy is recommended for patients with poor bowel preparation. Individualized withdrawal times may help improve the quality of individual colonoscopy.3. We successfully established and validated a quality scale for individual colonoscopy that incorporates procedure-related independent factors influencing missed diagnosis of adenomas. The quality scale correlates significantly with AMR and can therefore assess the quality of individual colonoscopy to a certain extent. However, the scale still needs to be independently verified before it can be used in clinical practice to reduce or prevent colorectal cancer by reducing the miss rate for adenomas. |
PDF Full Text Request