“Despite these limitations, this study still calls attention to the importance of preserving access to routine propofol use in patients undergoing screening and surveillance colonoscopy.”

In this month’s issue of Anesthesiology, Quaye et al. report findings of their investigation into differences in type of sedation and the detection of polyps during colonoscopy for screening and surveillance.  Their study uses data from the New Hampshire Colonoscopy Registry to compare patients who received propofol for their procedure versus moderate sedation with benzodiazepines and opioids. Although the authors hypothesized that the use of propofol would improve the detection of polyps that might progress to colon cancer, in their final analysis, they found no association between sedation type and overall detection of adenomas (adjusted odds ratio, 1.00; 95% CI, 0.95 to 1.05) or any neoplasia (adjusted odds ratio, 1.03; 95% CI, 0.98 to 1.08). However, they reported a modest statistically and clinically significant association between propofol use and detection of serrated polyps (adjusted odds ratio, 1.13; 95% CI, 1.07 to 1.19).

Serrated polyps comprise a heterogeneous family of polyps grouped based on a histologic feature, including both hyperplastic polyps and the less common—but clinically more concerning for malignant transformation—serrated adenomas. The latter are typically located in the right colon and are flat and more difficult to detect, particularly with suboptimal bowel preparation. Given the heterogeneity of these lesions, serrated polyps as a class are a secondary consideration in current colonoscopy quality metrics, which emphasize greater importance of adenoma detection rates over serrated lesion detection. Small hyperplastic polyps are specifically excluded from these definitions. Hyperplastic polyps larger than 10 mm should be removed during colonoscopy, and their presence may impact subsequent screening intervals. 

While previous work has demonstrated improvements in patient satisfaction and recovery times when propofol is used for sedation, the literature around adenoma detection is mixed making a robustly conducted study an important contribution to the literature. Unfortunately, because the current analysis grouped both serrated adenomas and hyperplastic polyps within the single serrated polyp variable, it is not possible to infer whether propofol is associated with increased detection of serrated adenomas. Future analyses should separate these specific lesion types.

Nevertheless, the analysis contains a number of important design features. The first is the setting of the study within a statewide registry. This design is generally positive, as it ensures that a breadth of clinical practice is captured and provides important definitional consistency of key covariates and outcomes across sites. Registry designs attempt to limit selection bias by including all or a specific sample of procedures across a group of facilities rather than those in a specific setting. As such, they can identify differences in care outcomes that may be attributable to differential care practices.

Second, the authors have taken a causal inference–informed approach to their experimental design. Causal inference as a field has emerged simultaneously from multiple fields including economics, computing science, and epidemiology. This field uses observed data, including practice variation as the basis of natural experiments, to make statements around causation. Fundamental to this approach is the careful assessment of the proposed causal pathway: How could propofol use increase the rate of serrated polyp detection? The authors surmised that deeper sedation achieved with propofol might lead to improved procedural conditions, allowing the endoscopist to more effectively detect serrated polyps. 

Third, informed by this framework, the authors employed a statistical technique known as inverse probability of treatment weighting to try to address differences in the likelihood of receiving the propofol, a source of residual confounding. In an ideal experiment, patients would be randomly assigned to sedation at the same facility with the same equipment and endoscopists to receive either propofol or moderate sedation with a benzodiazepine and opioids. Inverse probability of treatment weighting creates a balanced analytic population where the patients are at equipoise of receiving each treatment and increases the rigor of the analysis.

Some important limitations of this study remain. We note that propofol sedation may not always equate to anesthesiologist-led care. The study’s authors also avoided making this assumption. While the Food and Drug Administration (Silver Spring, Maryland)–approved package labeling for propofol recommends administration “only by persons trained in the administration of general anesthesia and not involved in the conduct of the surgical/diagnostic procedure,” nonanesthesiologist-administered propofol is widely accepted among gastroenterologists. Despite the medicolegal considerations multiple gastroenterology professional societies have released position statements supporting this practice. Therefore, we cannot conclude with certainty that anesthesiologist-led care improves polyp detection.

In addition, propofol use may be associated with providers and systems that are inherently different from those using moderate sedation. The initial study cohort included a large number of facilities that showed exclusive use of propofol and a few centers with exclusive use of moderate sedation. The authors subsequently identified facilities in which both sedation types were used during the study period for a restricted analysis; however, even in these centers, proceduralists might prefer propofol for medically complex patients, who are more likely to have comorbid conditions such as diabetes, smoking, and obesity, which have been linked to a greater risk of polyps. Proceduralists may also be motivated by quality metrics or financial incentives associated with polyp removal. Given this context, after applying propensity methods to adjust for treatment assignment and clustering patients by the endoscopist performing the procedure in this restricted cohort, there was a substantial attenuation in the effect size of receiving propofol sedation from an adjusted odds ratio of 1.51 (95% CI, 1.46 to 1.57) to an adjusted odds ratio of 1.13 (95% CI, 1.07 to 1.19). While the resultant effect size still suggests a clinically meaningful improvement in serrated polyp detection, the results must be interpreted with caution due to the inherent limitations present when analyzing registry data mentioned above.

Despite these limitations, this study still calls attention to the importance of preserving access to routine propofol use in patients undergoing screening and surveillance colonoscopy. The marked increase in the provision of anesthesiologist-led sedation for endoscopic procedures has been widely documented in both the Veterans Affairs and U.S. commercial insurance market.  In recent years, payers have proposed to stop reimbursing anesthesiologists providing sedation for healthy patients undergoing low-risk procedures such as cataract surgery and screening colonoscopy.  To date, these proposals have been unsuccessful due to opposition from patients and physicians. In the case of screening colonoscopy, these coverage decisions could adversely impact early diagnosis, patient outcomes, and access to care due to reduced clinical throughput; this is alarming in the context of the increased care demand driven by recent consensus guidelines that lower the recommended age for initial screening.  Additional rigorously conducted studies evaluating the impact of anesthesia care on outcomes will be important to determine the value of anesthesia care for patients undergoing screening and surveillance colonoscopy.