Solving the Rubik’s Cube of AAA mortality in Brazil using space-time cube analysis

This is an invited commentary on Franco RL et al. Abdominal aortic aneurysm treatment and hospital mortality in Brazil: A retrospective longitudinal study. Vasc Med 2024; 29: 526-531. ¹

In this issue of Vascular Medicine, Franco and colleagues ¹ present data from a longitudinal analysis of trends in mortality across geographic regions within Brazil and correlate this with spatio-temporal trends in surgical volume. These data reveal several troubling findings regarding aortic aneurysm care within Brazil, including an increasing rate of mortality due to abdominal aortic aneurysmal disease in multiple regions of the country, primarily in the south and west, which was coincident with a decreasing rate of surgical repair. ¹ Endovascular aneurysm repair throughout the country consistently had a significantly lower postoperative mortality rate compared to open aneurysm repair for both intact and ruptured aneurysms, although both exceeded commonly reported rates of mortality in North America and Europe.

Cardiovascular disease accounts for a substantial portion of global deaths. This is particularly true for abdominal aortic aneurysms, peripheral artery disease, and ischemic stroke, with deaths increasing over 100% over the last 30 years in low- and middle-income countries. Moreover, these resource-constrained environments often have a dramatic mismatch between cardiovascular disease and availability of surgical care. ²

Franco and colleagues ¹ used a unique set of data available through the Brazilian Ministry of Health, which provides geographic information that allow for identification of ‘hot’ and ‘cold’ spots, or changes in rates over time for a particular geographic area, using a technique called space-time cube analysis. This technique involves binning data, in this case individual patient data, into geographic regions based on proximity, as well as intermediate geographic regions based on administrative grouping, and following trends in health statistics in these regions over time. The Brazilian Institute of Geography and Statistics provides information on 134 geographic regions throughout the country, allowing for a relatively precise view of outcomes and procedural volume over the 12 years of data included in the study.

Heat maps and geographic data have been instrumental in analyzing variations in procedural volume, outcomes, and workforce issues for many years in the United States. Most researchers and many nonresearchers are familiar with the Dartmouth Atlas of Healthcare, which investigates a variety of geographic relationships in multiple areas of healthcare, including specifically vascular surgery, as published by Cronenwett and Birkmeyer in 2000. ³ One of the many innovations used within the Dartmouth Atlas is the identification of hospital referral regions, representing distinct tertiary medical care markets that are useful in understanding relationships across space and time. The variation across regions identified in the first Dartmouth Atlas report and subsequent revisions over the next 10 years would form the basis for the concept of Accountable Care Organizations, heavily influencing the Affordable Care Act and Medicare Reimbursement across the entire US healthcare system to this day. Following the initial publication of the Dartmouth Atlas, many more publications have used similar geographic bins to understand healthcare utilization and outcomes, leading to substantial changes in healthcare policymaking and resource allocation.^4–9

Spatiotemporal analyses have been used to examine a variety of relationships within vascular surgical care specifically. For example, Goodney et al. used similar techniques to examine the relationships between revascularization for lower-extremity peripheral artery disease and major amputation across the US. ⁵ Similarly, geographic relationships between payment and procedural utilization have been examined for outpatient peripheral vascular interventions, suggesting that payment and procedure utilization may be financially driven rather than by clinical need. This has already led to changes in reimbursement policy in some states and is likely for others as further attention is brought to discrepancies between payment and utilization patterns. With regard to the workforce, Ho et al. showed in 2005 that there was an inverse correlation between vascular surgeon supply and major amputation rates, providing additional rationale for policies targeting physician recruitment, particularly for regions with a mismatch between the incidence of peripheral artery disease and vascular surgeon supply. ⁷ Techniques investigating variation across hospital systems have been used for almost two decades to improve and standardize surgical care through regional quality improvement initiatives, such as the Michigan Surgical Quality Collaborative or the Vascular Study Group of New England, and nationally, such as the National Surgical Quality Improvement Program, Vascular Quality Initiative, and others.

A frequent problem with research using geographic data over time is its visual presentation and interpretation. Too small geographic bins can result in a low signal-to-noise ratio, particularly for regions with few procedures or patients. Too large regions or geographic boundaries created without an understanding of the underlying administration of healthcare can lead to statistically significant correlations, but without a clear path toward utility. In their analysis, Franco and colleagues ¹ have managed to combine a set of data over a time period critical to the development and adoption of endovascular aneurysm repair with a geographic breakdown that has the opportunity to affect a variety of policymaking and resource allocation decisions across the country. ² These findings detail the declining repair rate for abdominal aortic aneurysms and coincident rising mortality as well as the dramatically different rates across geographic region, each with unique resource and training challenges. In the resource-constrained scenario, data such as these can be used to determine where and where not to deploy resources, identify the precise services and educational tools required to impact outcomes most effectively, and provide longitudinal assessment of the efficacy of such interventions. As such, the techniques utilized by Dr Franco and colleagues highlight additional valuable applications of spatiotemporal data analysis in the realm of vascular care and set the stage for many potentially valuable insights into Brazilian healthcare delivery for years to come.