April 18, 2016

Correlation does not equal causation.

One fresh demonstration of the truth of this axiom appears in an article in this month’s issue of Gastroenterology (Wernli KJ, Brenner AT, Rutter CM, Inadomi JM. Risks Associated With Anesthesia Services During Colonoscopy. Gastroenterology 2016; 150: 888-894).

The research team, from the Group Health Research Institute in Seattle, performed a prospective cohort study of nationwide claims data from 3,168,228 colonoscopy procedures in adults aged 40 to 64 in the Truven Health MarketScan Research Databases from 2008 to 2011. Moderate sedation was performed in 65.6 percent of the procedures included in the study; deep sedation (in most cases using propofol) was provided by anesthesiologists or nurse anesthetists in 34.4 percent. The study authors found a correlation between use of anesthesia services and a 13 percent higher risk of any complication within 30 days: specifically, higher risk of perforation, bleeding, abdominal pain, complications due to anesthesia, and stroke and other neurological events.

Dr. Wernli et al. attributed the higher risk of anesthesia to clinical factors, noting that “There are plausible reasons as to why deep sedation with anesthesia during a colonoscopy could increase patients’ risks of adverse outcomes, such as aspiration when a sedated patient cannot protect their airway, or perforation when patients are not able to provide feedback to the endoscopist regarding excessive pressure. (Citations omitted.)”

Both rates of use (53 percent in the Northeast vs eight percent in the West) and rates of complication (12 percent in the Northeast vs 60 percent in the West) varied significantly by geographical region. The use of anesthesia services was associated with a higher risk of any complication in all regions except the Southeast, where there was no association between use of anesthesia services and complications from colonoscopy. The researchers raised the question of potential “confounding by comorbidity status” but did not pursue an answer beyond stating that “we did not find systematic differences in patient characteristics by anesthesia services across regions.”

Norman A. Cohen, MD, Clinical Professor of Anesthesiology and Perioperative Medicine at Oregon Health and Science University, offered a more extensive analysis in a personal communication. Dr. Cohen wrote the following:

The [study] authors note a higher relative risk for complications with an anesthesia service compared to moderate sedation. The complications attributable to anesthesia are a bit problematic. Notably, they attribute infection to anesthesia care. One could argue that many infections could as easily be attributable to the procedure itself; however, for the sake of this analysis, let us ignore that and keep the attributions the same. Let us also ignore the fact that a retrospective study like this only demonstrates associations and not causation and that the attribution of complications in administrative claims analyses are statistically problematic compared to a randomized trial or even a registry based analysis.

The data demonstrate geographic variability both in prevalence of anesthesia care and relative risk of complications. The western U.S. had the lowest prevalence but the highest complication relative risk with anesthesia. The northeastern region had high prevalence but relatively low relative risk.

Given the low prevalence in the West, one may hypothesize that patients receiving anesthesia care have greater comorbidities. The numerous medical necessity policies in place in the West during the period of the study support that hypothesis. My anecdotal knowledge does as well. If the patients are more complex medically, then having a higher complication rate in the anesthesia group makes some sense. It also makes sense that the relative risk in the Northeast is lower. Because a much greater percentage of the population is receiving anesthesia care, it follows that a greater number of healthy patients are undergoing anesthesia. The overall risk of complications is likely lower in a healthier population receiving a given treatment.

So we may have increased complications in the anesthesia group because the patients are sicker, an observation that applies to both procedural risk and anesthesia risk—and we may also have an incremental increase in risk due to anesthesia compared to moderate sedation.

While many facilities require anesthesia involvement for patients who are ASA 3 or greater, healthy patients may receive either moderate sedation or an anesthesia service depending on patient or endoscopist preference. These preferences account for most of the regional differences in anesthesia use.

If anesthesia care independently increases risk, then regions with higher anesthesia prevalence would have a higher risk of complications across the entire population of patients having endoscopies. In other words, the Northeast with its higher prevalence of anesthesia care should have a total complication rate higher than the West. This assumes that overall health status is the same.

The calculations below show a very modest increase in risk that can be attributed to the use of anesthesia services:

The risk of a complication of any type for moderate sedation patients will be called baseRisk with a value of x.

baseRisk=x

We know from the study that the complication rate under anesthesia is 60 percent greater in the West than for moderate sedation, or 1.6 times the baseRisk.

anesRiskWest=1.6*x

Similarly, the risk in the Northeast is 12 percent greater. anesRiskNortheast=1.12*x The prevalence of anesthesia care in the West and Northeast is eight and 53 percent respectively.

rateWest=8%
rateNortheast=53%

The all-patient relative risk for complications compared to the baseline risk with moderate sedation is determined by the following calculations:

popRiskWest=anesRiskWest*rateWest+baseRisk*(1-rateWest)=>1.048x popRiskNorthEast=anesRiskNortheast*rateNortheast+baseRisk*(1-rateNortheast)=> 1.0636x

Dividing the all-patient relative risk for the Northeast by the all-patient relative risk for the West results in the relative risk attributable to anesthesia only since the base risk cancels out.

relativeRisk=popRiskNorthEast/popRiskWest=>1.0149

Accounting for health status, the increased risk attributable to anesthesia alone is thus 1.5 percent. This is certainly clinically insignificant. Not having the data set, I doubt, but cannot affirm, that the risk is statistically significant.

Dr. Cohen’s calculations can be performed for any other region of interest in the study by replacing the prevalence rate and the complication rate for the Northeast with the new region's rate and then using the same formulas. This will relate the new region to the data from the West, which has the lowest moderate sedation and highest complication rates.

Wernli et al. conclude that “the overall risk of complications after colonoscopy increases when individuals receive anesthesia services”—but Dr. Cohen has shown that the fact of anesthesia by itself does not account for more than a minimal share of the risk. Could it be that many endoscopists prefer the involvement of anesthesiologists and/or CRNAs because the deeper level of sedation allows endoscopists to work more quickly? And that working more quickly and aggressively [would] explain the non-life threatening complications” evaluated in the study, a comment attributed to Jeffrey W. Apfelbaum, MD of the University of Chicago? (Ready T. Anesthesia Use for Colonoscopy Attracts New Scrutiny. HealthLeaders Media online, April 13, 2016.)

Anesthesiologists tend to believe that the growth in the volume of colonoscopies performed with anesthesia—the use of anesthesia services for colonoscopy patients rose from approximately 14 percent in 2003 to more than 30 percent in 2009 to close to 50 percent in 2013, according to a series of reports from the Rand Corporation—is driven by endoscopist and patient demand. Many gastroenterologists and quite a few anesthesiologists believe that the profitability of anesthesia for colonoscopy is a strong motivating factor.

That profitability, combined with geographic variations in prevalence, makes anesthesia a target for cost-cutters. The Wernli article could be helpful to health plans seeking to reduce payments for anesthesia for colonoscopy. John A. Martin, MD of the Mayo Clinic commented on the article thus: “we must now ask ourselves and discuss with our patients honestly, not only whether the added cost of anesthesia is reasonable—but also whether the apparent added risk of anesthesia justifies perceived benefits.” (Martin JA. Anesthesia during Colonoscopy May Not Be Worth the Cost. GI & Hepatology News, April 1, 2016.) Dr. Martin, of course, did not have the benefit of Dr. Cohen’s analysis. His comments might be different if he did.

Anesthesiologists who find themselves defending against the study would do well to remember Dr. Cohen’s analysis, including his observation on the limitations of administrative claims data as a tool for assessing outcomes. The use of anesthesia services by itself increases the risk of complications by a mere 1.5 percent. The benefits of anesthesia include patient comfort and satisfaction, rapid sedation followed by quick recovery and faster overall procedure turnaround time. In all likelihood, these outweigh the well-known but minimal risks.