Role

Colonoscopy reports serve various different functions and are used by various different parties. First and foremost, the colonoscopy report is a clinical document. For the referring provider and other clinicians, the report renders diagnoses and recommended therapies, and the timing and interval for future examinations. The report may also help endoscopists prepare for future examinations through providing critical information about sedation and instrument requirements, response to bowel cleansing strategies, patient tolerance, and any complications. Second, the colonoscopy report can be used for quality assurance and improvement processes. Quality measurement information may be obtained, such as cecal intubation, quality of the preparation, and withdrawal time. Third, the colonoscopy report may have medicolegal implications. Fourth, colonoscopy reports are important for reimbursement. Inaccurate documentation of procedure indications or techniques used might translate into more time and effort spent obtaining reimbursement from payers.

Current State

Only a few studies have rigorously examined the quality of endoscopy reports, and most have shown incomplete reporting for several items. Robertson and colleagues examined reports from 122 endoscopy centers and found a substantial proportion of reports were missing key data, such as bowel preparation adequacy (70.5%), complications (41.8%), polyp size (26.2%), postprocedure recommendations (31.1%), sedation dose (27.1%), patient history (42.5%), and demographics (30.4%). More recently, Lieberman and colleagues queried 438,521 colonoscopy reports and also found that many were missing key data, including bowel preparation quality (13.9%), comorbidity classification (10.1%), description of cecal landmarks (14.7%), and polyp size (4.9%). Across practices, significant variation was seen in the inclusion of certain elements. For instance, the bowel preparation quality was missing in more than 20% of procedures at 14 of the 52 endoscopy centers. In another recent study, only 33.5% of reports included follow-up recommendations.

Colonoscopy Reporting and Data System

Recognizing that the absence of comprehensive and standardized colonoscopy reports has actually impeded quality measurement for colonoscopy, the Quality Assurance Task Group of the National Colorectal Cancer Roundtable developed the colonoscopy reporting and data system (CO-RADS) in 2007. The group modeled CO-RADS after work performed to standardize the reporting of other screening tests (mammography and CT colonography) and was developed with the expertise of individuals in gastroenterology, radiology, primary care, and health care delivery. The group had the goal of creating a system that included a colonoscopy report with standardized elements and terminology that would allow measurement of quality within and across practices. For each of the following items, CO-RADS identifies important subitems that should be specifically addressed.

Elements of a Quality Report

Defining and Measuring Quality

According to one industry survey, almost 90% of gastroenterologists already collect quality information or intend to do so soon. Assessment of quality and performance measures (or quality indicators), however, depends on how one defines quality. Health care professionals have traditionally defined quality in terms of the technical excellence of medical care and the nature of provider–patient interactions. Technical excellence consists of two aspects: the appropriateness of care and the skill with which care is provided. In essence, it is “doing the right thing right.” The nature of provider–patient interactions depends on clear communication, trust, concern, and sensitivity.

Although quality can be defined in various ways, quality measures have only three basic types: structure, process, and outcomes. Structure deals with the organizational infrastructure, including the facility and provider characteristics (eg, physician training and specialty). Processes are the components of care or steps taken by a provider to achieve some end point (eg, performing random biopsies for chronic diarrhea). Finally, outcomes refer to health status and are the result of structure and processes (eg, colonic perforation or adenoma detection).

Process Measures

Several features make process measures appealing. First, process measures are often clearly defined and simple, making them easier to measure than outcomes. Second, unlike outcome measures, process measures typically do not need to be adjusted for differences in patient characteristics. Third, when a good process measure is identified, it highlights a specific action or technique on which a provider can focus to improve health outcomes. Fourth, process measures are often more sensitive than outcome measures.

For example, two endoscopists may have dramatically different rates of cecal intubation but very similar rates of colon cancer detection. As this example illustrates, an outcome measure sometimes will not detect poor quality care even when poor quality care occurs. This scenario is more common when certain outcomes (eg, colon cancer detection) are relatively infrequent and a large number of encounters are required to calculate reliable estimates for their frequency.

Process measures also have limitations; they often have no intrinsic worth. A process measure is only useful if a clear link exists between the process and a clinically relevant outcome. Process measures face a test of validity and are not always accepted as meaningful by patients, providers, or payers. Additionally, individual process measures usually only assess one specific element of care, but do not yield a comprehensive evaluation of care.

Outcome Measures

Outcomes are more intuitive measures; they typically do have intrinsic worth and are often the purpose for the health care processes. These measures usually matter most to individual patients and purchasers of care. One of the major limitations of outcomes measures is the fact that patient characteristics can profoundly affect health outcomes. Thus, comparison of outcome measures across providers or groups often requires risk adjustment to ensure that differences in patient populations can be accounted for and valid comparisons can be made. For instance, adenoma detection can be influenced by age, gender, and quality of bowel preparation, among other things.

Because outcome measures are influenced by several factors other than a provider’s actions, providers will often have some degree of concern about the validity of outcome measures. Additionally, although health outcomes are the ultimate goal of care, measuring outcomes does not always provide physicians with specific targetable actions or steps they can address to improve. Also, certain outcomes are rare or a substantial delay occurs between care and health outcomes. For colonoscopy, perforations and colorectal cancers are too rare to offer meaningful comparisons until large sample sizes are amassed. Finally, although processes can often be collected easily during routine clinical care, outcomes data analysis often requires additional steps outside of the clinical encounter.

Ideal Measures

Rising health care costs and an emphasis on quality improvement have spurred discussion about characteristics of ideal quality measures across medicine, including gastroenterology. Colonoscopy quality measures are considered desirable if they are closely linked to relevant outcomes, evidence-based, auditable, easily integrated into practice, aimed at quality improvement (not cost-containment), and developed by a consensus process involving gastroenterology societies. Ideal measures should also be resistant to unintended consequences, and collection of measures should not be overly time-consuming. In 2006, the ASGE and the ACG formed the ASGE/ACG Task Force on Quality in Endoscopy and proposed a list of quality indicators for colonoscopy. These indicators are applied to three periods: preprocedure, which encompasses all contacts with the patient until sedation or scope insertion; intraprocedure, which encompasses scope insertion through withdrawal; and postprocedure, which encompasses recovery through follow-up.

Quality Measures for Colonoscopy

Reasons for Variation in Quality

Many studies have focused on documenting the existence of variation in colonoscopy quality, but the reasons for this variation have not been clear. Retrospective studies have shown associations between cecal intubation and several factors, including bowel preparation quality, sedation type, colonoscopy setting, and endoscopist factors such as volume and years in practice. Others have retrospectively analyzed polyp detection rates and found associations with bowel preparation quality and sedation, among other factors. Some of the variability is certainly due to patient-related factors (eg, demographics, family history) and procedure-related factors (bowel preparation quality, sedation), but a substantial fraction of the variability is also related to endoscopist performance. As a result, quality improvement projects have investigated patient-related changes, organizational arrangements, and endoscopist-related interventions.

Impact of Measurement: Internal Audits and Organizational Changes

In 2004, Ball and colleagues reported one of the first prospective quality improvement projects for colonoscopy. Hoping to improve cecal intubation rates, members of an endoscopy unit in the United Kingdom audited colonoscopy reports over two periods to determine reasons for incomplete examinations. The strategy during the first audit included increasing appointment duration from 20 to 30 minutes, and hospitalization of frail patients for their bowel preparation. This practice led to an increase in cecal intubation rate from 60% to 71.2%. The strategy developed during the second audit included assigning more colonoscopies to the endoscopists with higher cecal intubation rates and having less-successful endoscopists receive further training or stop performing colonoscopies. The group increased their completion rate to 88.1% of all procedures and 93.8% of procedures without a poor bowel preparation, stricture, or obstruction. Not only did the department as a whole improve, but also all of the individual endoscopists who continued to do procedures improved. Although many of the changes were organizational, the authors thought the process of auditing itself also influenced performance.

Imperiali and colleagues used a similar approach to that used by Ball and colleagues, examining a quality improvement project in Italy. They performed six monthly audit cycles from 2001 to 2005. The audits included departmental meetings to discuss standards, review the results of the audit, and develop action plans. Individual endoscopists were informed about their own performance only. One intervention included giving preprocedure sedation instead of on-demand sedation for patients. An approach that differed from the one used by Ball and colleagues was the practice of having more endoscopy sessions for endoscopists with lower performance rates or less experience. The less-skilled endoscopists were also randomly supervised by more-experienced endoscopists, and endoscopists with low polyp detection rates had a personal interview with the head of the unit to discuss the importance of polyps and technical aspects of examination. Additionally if an endoscopist was unable to intubate the cecum, a second endoscopist was instructed to attempt to intubate the cecum, if possible.

Over the study period, the cecal intubation rate increased from 84.6% to 93.1%. The overall polyp detection rate remained about the same, but the range in polyp detection among individual endoscopists decreased. They also acknowledged the possibility that the audit process itself may have been responsible for much of the improvement and reported a slight decrease in cecal intubation during a period when endoscopists were unaware that examinations had been reviewed.

Colonoscopy Technique and Feedback

Harewood and colleagues investigated the issue of feedback on cecal intubation, insertion time, and withdrawal time. On a quarterly basis for almost a year, 58 endoscopists received e-mails communicating their performance and rank compared with their colleagues. The cecal intubation rate increased from 95.3% to 96.2% and insertion time decreased from 10.6 to 9.5 minutes, whereas median withdrawal times did not change significantly. In addition to the e-mails, the group postulated that discussions at divisional meetings may have reinforced the importance of slow withdrawal and careful inspection. They also raised the issue of durability, noting that some of the trends started to reverse toward the end of the study period.

In 2008, Barclay and colleagues investigated a novel quality improvement intervention that involved an emphasis on both withdrawal time and inspection technique. To achieve an 8-minute withdrawal time, they used a digital stopwatch that emitted an audible sound signifying 2-minute increments. To address the issue of inspection technique, the group discussed an article describing techniques (insufflation, suction, inspection of flexures, proximal sides of folds, withdrawal time) and solicited the advice of endoscopists in the group with high ADRs. After the intervention, endoscopists increased the rate of neoplasia detection from 23.5% to 34.7%, which included an approximately 50% increase in the percentage of subjects with at least one adenoma and a 45% increase in the number of advanced adenomas per subject screened. Despite the apparent improvements, the authors found that the endoscopists with the highest ADRs had only intermediate withdrawal times. A retrospective study by other authors failed to identify an increase in polyp detection, using a 7-minute withdrawal protocol.

In one of the largest studies to address the issue of performance improvement, Shaukat and colleagues found that various interventions did not appear to change endoscopists’ adenoma detection rates. The authors prospectively collected data on 47,253 screening colonoscopies over 3 years, while instituting five interventions. Interventions included ADR review (blinded and unblinded), education about ADR benchmarks and withdrawal time, discussions between poor performing endoscopists and practice leaders, and financial consequences of not achieving a 6-minute withdrawal time in greater than 95% of colonoscopies.

Although some of the studies above show some improvement, uncertainty still exists regarding the optimal approach to improve endoscopist performance. Elements of inspection thought to impact neoplasia detection include ample insufflation, meticulous evaluation of flexures and proximal sides of folds, adequate suctioning of retained fluid, and withdrawal duration. Recently, operator motivation and fatigue have been suggested as additional factors affecting the yield of colonoscopy. Aside from withdrawal times, most of these features are too ambiguous or difficult to quantify and measure consistently. Furthermore, even if a measure is selected, the optimal frequency and format for feedback is unclear. Finally, if improvements are made, the improvements may not be durable or long-lasting.

High-priority Measures

Although the perfect quality measure does not exist, several measures seem to be particularly useful for improving the quality of colonoscopy: (1) cecal intubation rate, (2) adenoma detection rate, (3) withdrawal time, (4) preparation quality, (5) follow-up recommendations, and (6) ASA classification.