Through the glass, darkly: variability in GA use for MRI by race
Bishr Haydar MD, Lynne G, Maxwell MD, and Myron Yaster MD
I’m fairly certain that most readers of the PAAD will notice that today’s PAAD is yet another article on health care disparities, which are “defined as differences in outcomes or services between groups that negatively impact the health of one group compared to another.”1 Today’s article by Gan et al.1 is yet another study from Dr. Tony Anderson’s group of Stanford University which used “Optum’s deidentified Clinformatics Data Mart (CDM) Database (DOI: 10.57761/a5hg-b442) to identify children who underwent MRIs between January 1, 2004 and May 31, 2019 under general anesthesia. Optum’s CDM is derived from a database of administrative health claims for members of large commercial and Medicare Advantage health plans.”1
As you will see in our review, how one defines general anesthesia is fundamental to understanding the results of today’s article. My good friend, former fellow, and colleague, Dr. Allison Kinder Ross, always rolls her eyes when this issue is raised because she knows that, for me, “it’s like waving a red cape in front of a charging bull.” For many, particularly radiologists, propofol general anesthesia is thought of as sedation and not general anesthesia. Unfortunately, we, as a profession, are very much to blame for this misconception. How often is a propofol, natural airway general anesthetic labeled monitored anesthesia care (MAC) or deep sedation, particularly if the propofol is administered by a “sedation service” staffed by non-anesthesiologists? Obviously, we will not be able to give this issue the full consideration it deserves and I will have some of our thought leaders review it in a future PAAD.
Finally, we are fortunate to have Dr. Bishr Haydar, a clinical associate professor at the University of Michigan and communications subcommittee chair of SPA’s committee on diversity, equity, and inclusion join the PAAD’s executive council. Look forward to seeing his byline and reading his thoughts in the weeks and months to come. Myron Yaster MD
Original article
Ziyu Gan, Julia M Rosenbloom, Elizabeth De Souza, T Anthony Anderson. Racial/Ethnic Variability in Use of General Anesthesia for Pediatric Magnetic Resonance Imaging. Anesth Analg. 2023 Jun 1;136(6):1189-1197. PMID: 36857212
Does use of general anesthesia for MRI in children vary by race? This question was examined in a study by Ziyu Gan BS and colleagues at Stanford (senior author T. Anthony Anderson MD PhD). They evaluated the Optum Clinformatics Data Mart, a large billing database of commercial insurance and found 450,000 patients <= 21 years of age who had an MRI between 1/1/04 to 5/31/19. They compared general anesthesia use between different races using a limited set of covariates. Their regression elements included age, sex (only two categories), altered mental status, ADHD and developmental delay. They captured numerous encounter characteristics that may influence the anesthetic choice, including the anatomic area scanned, whether multiple scans were done and use of contrast. They stratified their analysis using the common WHO age categories.
Overall, only 6.4% of patients who had an MRI had it under GA.. As expected, GA use varied widely by age:
Age(y) % GA
0-1 60.1%
2-5 70.7%
6-11 15.5%
12-18 0.8%
19-21 0.2%
Table 3 shows GA rates by race/ethnicity, with a Poisson regression model to adjust for covariates. Ignoring the tiny group of patients aged 19-21, Asian, Black and Hispanic children of all ages had equal or higher GA rate when compared to White children. However, only 1-2 age categories for each race reached statistical significance. The adjusted relative risk of GA for children ranged from 1.00 - 1.14, but were generally less than 1.10 They present a post-hoc analysis that took a different statistical approach with largely the same results. The article included 13 online supplemental tables of data!!
Our thoughts:
1. Studies that illuminate a health care disparity should be a reminder that we all have implicit bias - that these biases have been shown to have a real impact on patients. Equity should always be in the forefront of our minds - am I treating this patient from a marginalized community differently?
2. This type of study design is highly limited, as many/most outcomes and covariates and moderators/mediators are simply not available. Many of the data fields are also limited - patients can only be one of two genders and only one race.
3. Most disparity studies look at negative impacts. However, this study simply looked at GA exposure, but could not examine whether use of GA was indicated or whether there was any downstream negative impact on the patient. The authors highlight the possible adverse effects of general anesthesia but are unable to identify whether the patients who received GA had worse outcomes than those who did not. For future study design, we might look to choose patient-centered outcomes which are described in several recently published papers.2, 3
4. The authors excluded MRI under sedation. We speculate that this is because many types of providers can bill for sedation, making a retrospective study of this type difficult. So, this study examines GA vs "undefined" which can include:
- "fed and swaddled" newborns
- nursing sedation
- Monitored Anesthesia Care
- sedation provided by the ICU or Emergency Departments or possibly others.
So what's better - GA or this "other" group many of which are general anesthesia by another name?
5. We can't determine causation in this kind of study. For example, there could be regional and institutional differences in patient mix and comorbidities that can drive this outcome. For example, obesity may inform the choice of sedation vs GA for MRI, and the prevalence of obesity varies regionally and by race. There are cultural differences by race that may also influence anesthetic choices. For example, premedication and parental presence at induction of anesthesia has been shown to be less common in Black children, and in retrospect at least some of that may reflect differences in parental preference and expectations.4 As the authors point out, the meaning of the variability in care which was identified in this study is uncertain. As they state, “does this variability constitute a health inequity either in terms of health services or health care processes? How and with whose input is sthe decision around GA for a pediatric made?” It is clear that the anesthesiologist providing GA care for the list of patients for MRI on any given day is not responsible for the decision-making that results in the scheduling of patients.
Bottom line: We're peering through an imperfect, cloudy lens (billing database) to try to get some insight. While the associations found in this study are not definitive, it still gives us an opportunity to reflect on the biases we carry and how they may influence that patient sitting in front of us right now. What do you think? Send your responses to Myron who will post in the Friday Reader Response.
References
1. Gan Z, Rosenbloom J, De Souza E, Anderson TA. Racial/Ethnic Variability in Use of General Anesthesia for Pediatric Magnetic Resonance Imaging. Anesthesia & Analgesia. 2023;136(6):1189-1197. doi:10.1213/ANE.0000000000006403
2. Sommerfield A, Sommerfield D, Bell E, et al. Consumer research priorities for pediatric anesthesia and perioperative medicine. Paediatric anaesthesia. Feb 2023;33(2):144-153. doi:10.1111/pan.14564
3. Gill PJ, Bayliss A, Sozer A, et al. Patient, Caregiver, and Clinician Participation in Prioritization of Research Questions in Pediatric Hospital Medicine. JAMA network open. Apr 1 2022;5(4):e229085. doi:10.1001/jamanetworkopen.2022.9085
4. Baetzel A, Brown DJ, Koppera P, Rentz A, Thompson A, Christensen R. Adultification of Black Children in Pediatric Anesthesia. Anesthesia and analgesia. Oct 2019;129(4):1118-1123. doi:10.1213/ane.0000000000004274