National trends in repair for type B aortic dissection

Abstract Background Thoracic endovascular aortic repair (TEVAR) first gained in popularity for repair of type B aortic dissections (TBADs) in the early 2000's. We aimed to describe patients undergoing open repair, TEVAR, and no repair and analyze factors associated with repair within 14 days of presentation in the contemporary era. Methods We used the MarketScan database to find patients with TBAD between 2014 and 2017. To assess factors associated with early repair, univariable, and multivariable log‐binomial regression were used. Results There were 2613 patients admitted with TBAD between 2014 and 2017 across the United States, of whom 38.4% underwent repair within 14 days of admission (25.3% open repair and 13.1% TEVAR). The incidence of repair within 14 days decreased over the study period (43% of the study cohort in 2014 to 26.4% in 2017) primarily due to a decrease in open repairs from 30.8% of patients in 2014 to 12.5% in 2017. In multivariable analysis, older age, Middle Atlantic location, diabetes mellitus, insulin use, antiplatelet use, and more recent year were associated with lower likelihood of early repair; male sex, peripheral vascular disease, and the presence of extremity ischemia, rupture, shock, and acidosis were associated with higher likelihood of repair. Conclusions Overall, repair of TBAD within 14 days of presentation declined from 2014 to 2017, with a steady rate of TEVAR but declining rate of open repairs. Further investigation into provider‐ and hospital‐specific factors as they relate to likelihood of repair is needed.

and December 31, 2013. Due to data availability, patients were not able to be excluded on the basis of dissection codes prior to 2013; therefore, some patients in the cohort may have been hospitalized due to progression or complications of chronic dissections. Patients with type A aortic dissections were removed by excluding patients with concomitant procedure codes for cardioplegia, cardiopulmonary bypass, valve repair, or operations on the vessels of the heart, following previously-described coding strategies (Appendix A). 9 Patients without 1 year of continuous insurance enrollment (for comorbidity ascertainment) and 90 days of continuous prescription coverage (for medication usage ascertainment) prior to their index admission, as well as 14 days of continuous insurance enrollment post index admission (for repair ascertainment) were excluded ( Figure 1).
Though mortality is not reported in MarketScan, removal of enrollment ID from the dataset may indicate death, loss of health insurance, or discontinuation of an insurance's participation in the MarketScan database.

| Outcomes and covariates
The primary outcome was the presence and type of repair within 14 days of index admission. This time window was chosen in accordance with the 2020 Society for Vascular Surgery and Society of Thoracic Surgeons reporting standards for TBAD. 10 Open surgical repair and TEVAR were defined by ICD-9-PCS, ICD-10-PCS, and CPT codes as given in Table 1.
Other variables of interest included demographics, insurance plan type (divided into those with and without primary care clinician assignments), and comorbidities. Patient race and ethnicity are not available in MarketScan. Comorbidities were ascertained using previously validated coding algorithms in the year prior to, and exclusive of, the index admission date. 11,12 Medication use was assessed using filled medication prescriptions from 90 days prior to, and not inclusive of, the index admission date. Medication types included antihypertensives, anticoagulants, antiplatelets, statins and other lipid lowering agents, and insulin (Appendix B).
Complications during index admission were ascertained using ICD-9-CM and ICD-10-CM codes and included aortic rupture, bowel ischemia, extremity ischemia, acidosis, shock, renal failure, paraplegia, and stroke (Appendix C). Because the diagnosis codes used to determine the presence of complications were captured from the billing claim filed at the time of discharge, it was not possible to determine whether they were present at the time of admission or developed following repair. Therefore, it was not possible to divide the cohort into complicated versus uncomplicated dissections. Because of the lack of mortality data in MarketScan, we were not able to report mortality during or following admission. We were also unable to report morbidity following the index admission, because mortality is a significant competing risk for morbidity in this patient cohort.   Table 3).

| Factors associated with early repair versus no repair
Factors associated with early repair on univariable analysis can be seen in Table 4 repair within 14 days and 62% did not. This is similar to repair frequencies reported in other mixed "complicated" and "uncomplicated" A preference for subacute repair may also have increased over the study period following the 2014 publication of data from the VIRTUE Registry that showed that patients undergoing subacute TEVAR experienced lower mortality than the acute group while retaining the favorable aortic remodeling characteristics associated with acute repair. 16 It is also possible that increased focus on the role of antihy- Hypertension was the most common comorbidity in this cohort (present in 67.7% of patients overall 90 days prior to index), in keeping with its high rate nationally and its known association with aortic dissection. Other authors have reported slightly higher proportions of hypertension in TBAD cohorts ranging from 75.1% to 88.1%. 3,5,8,14,17 It is possible that relatively fewer patients in our cohort had aortic dissection due to hypertension and relatively more had dissection due to trauma or connective tissue disorders because we did not specifically There are several limitations of this analysis. First, we were unable to describe and account for hospital-and clinician-related data, including the TBAD treatment patterns of institutions, because the necessary data fields either had very high rates of missingness or very homogenous data (such as the provider type very frequently being listed as "acute care hospital" rather than a clinician specialty). We believe that institution-level practice patterns likely significantly influence whether or not patients undergo early repair, but are unable to show that based off our data. Second, as noted above, the use of administrative data makes distinguishing between complications of TBAD versus complications of repair impossible, therefore we were not able to analyze repair patterns separately among patients with complicated and uncomplicated TBAD. Third, MarketScan includes only patients with certain insurance types from participating insurance carriers but prior research has shown TBAD to disproportionately affect underinsured patients, suggesting that there is an important group of TBAD patients who were not captured in this analysis. 18 Fourth, Mar-ketScan does not include mortality data. Although we required 14 days of continuous enrollment post-index to ascertain early repair, we cannot be certain absolutely certain whether patients without early repair died before repair could be done, and we were not able to determine mortality following discharge, which prevented us from reporting post-discharge complications. Fifth, because we were limited in the amount of data we had available for a look-back period, this cohort represents a mix of patients with acute, sub-acute, and chronic TBAD. While it therefore cannot illuminate practice patterns specifically among patients with acute TBAD, which is an area of active investigation nationally and internationally, we feel that an updated look at TBAD management patterns overall provides important context to ongoing scholarly activities and debate.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are part of the Mar-ketScan database, a proprietary dataset, and are therefore not able to be shared.