Trends in Thoracic Aortic Aneurysm Hospital Admissions, Interventions, and Mortality in England between 1998 and 2020: An Observational Study

This observational study spanning a 23 year period revealed that there has been a signi ﬁ cant increase in hospital admissions for thoracic aortic aneurysms in the UK, with a parallel increase in thoracic endovascular aortic repair (TEVAR) uptake and a decline in aneurysm related mortality. In the absence of randomised trial data, the trends observed in this ‘ real world ’ dataset, and the signi ﬁ cant reduction in aneurysm related mortality in the endovascular era vs. the pre-endovascular era, support the clinical ef ﬁ cacy of TEVAR and provide some evidence for its continued use. Objective: To assess trends in thoracic aortic aneurysm (TAA) hospital admissions, interventions, and aneurysm related mortality (ARM) in England, and examine the impact of endovascular repair on mortality for the years 1998 to 2020. Methods: Hospital admission and operative approach (thoracic endovascular aortic repair, [TEVAR] or open surgical repair) using Hospital Episodes Statistics, and ARM data from the Of ﬁ ce for National Statistics for England standardised to the 2013 European Standard Population were analysed using linear regression and Joinpoint regression analyses. ARM was compared between the pre-endovascular era (1998 e 2008) and the endovascular era (2009 e 2019). Results: A rising trend in hospital admission incidence has been observed, mainly due non-ruptured admissions (4.11 per 100 000 in 1998; 95% con ﬁ dence interval (CI) 3.71 e 4.50 to 12.61 per 100 000 in 2020; 95% CI 12.00 e 13.21 in 2020; r 2 ¼ .98; p < .001). Operative interventions increased mainly due to an increase in TEVAR (2.15 per 100 000; 95% CI 1.91 e 2.41 in 2020 vs. 0.26 per 100 000; 95% CI 0.16 e 0.36 in 2006; r 2 ¼ .90; p < .001). Reductions in ARM from TAA were observed for males and females, irrespective of age and rupture status. The greatest reduction in ARM in the endovascular era was observed in females aged > 80 years with ruptured disease (15.26 deaths per 100 000 vs. 9.50 deaths per 100 000; p < .001). Conclusion: A signi ﬁ cant increase in hospital admissions for non-ruptured TAA has been observed in the last 23 years in England, paralleled by a shift towards endovascular repair, and signi ﬁ cant declining trends in ARM, irrespective of sex and age. The signi ﬁ cant reductions in age standardised death rates from ruptured and non-ruptured TAA in the endovascular era, particularly for females aged > 80 years with ruptured disease, af ﬁ rm the positive impact of an endovascular approach to TAA.


INTRODUCTION
Aortic aneurysmal disease is a significant public health concern given a natural history of continued aortic expansion and eventual rupture, an event that is fatal in 80% of people. 1 During the twentieth century, a steady increase in aortic aneurysm mortality was reported. 2 However, several population based studies have demonstrated a decline in both abdominal and thoracic aortic aneurysm (TAA) deaths during the twenty first century. 3e5 This trend was hypothesised to be the result of public health measures targeting cardiovascular risk factors associated with aortic rupture, such as smoking and hypertension, rather than the improved trends in operative mortality from aortic repair, and particularly the increased implementation of minimally invasive thoracic endovascular aortic repair (TEVAR) that was embraced in the 2000s. 5 TEVAR has become firmly established in contemporary vascular practice, supplanting traditional open surgical repair (OSR), and is recommended as a treatment strategy in both the European 6 and US 7 vascular societies' practice guidelines. Furthermore, the adoption of endovascular technology shows no sign of slowing down, with the ongoing evolution of new and expensive devices to treat disease in the aortic arch, in the absence of randomised trial data. 8 The conclusions of prior population based studies may now be outdated and reflect outcomes related to clinical practice at the time; the analysed data in many of these studies was from early time periods when TEVAR technology and practice were immature, and during the development of a learning curve with respect to endovascular skill and patient selection. The primary aim of this study was to undertake a recent analysis of the trends of hospital admissions, operative approach, and aneurysm related mortality (ARM) for TAA for unselected "real world" patients in England, and to examine the potential impact of endovascular repair on mortality.

Study design and participants
This population based cohort study included a retrospective review of the following outcome data for all patients older than 60 years in England who (1) had a hospital admission for a diagnosis of ruptured or non-ruptured TAA (admission data); (2) underwent OSR or TEVAR for TAA (procedural data); and (3) were registered on death certificates to have died from TAA (in or out of hospital ARM data).

Data sources
Hospital admission (based on International Classification of Diseases [ICD] version 9 and 10 codes) and procedural data (according to the Office of Population, Censuses and Surveys: classification of interventions and procedures, 4th revision [OPCS-4]) were obtained from Hospital Episodes Statistics (HES), an administrative data warehouse that is publicly and freely available through the National Health Service (NHS) Digital platform. 9 HES contains details of the activity of hospitals within the NHS in England, including all admissions, diagnoses, and treatment procedures. Individual patient level data or hospital identifiers were not available to the researchers. Mortality data were obtained from the Office for National Statistics for England (ONS), which collates cause of death from death certificates (data requested 2 November 2020, reference number 12426). 10 The authors did not have access to the proportion of patients who underwent autopsy to confirm the diagnosis. Both HES and ONS data sources have been used in similar studies and are well validated administrative databases. 11,12 The full list of ICD-9/10 and OPCS codes used are provided in Supplementary Figure S1.

Inclusion and exclusion criteria
Owing to the broad age bands provided in the HES dataset (18 e 59 years) for certain calendar years, and to ensure the inclusion of degenerative aneurysms only (as opposed to post-dissection aneurysms in individuals with connective tissue pathologies), all data extraction was restricted to patients older than 60 years and in five year age bands. Patients who underwent repair of aneurysmal disease of the ascending aorta, aneurysmal aortic surgery where the segment of aorta was not specified ("other specified or unspecified aneurysmal aorta"), "revision" surgery, aortic surgery without the mention of "aneurysm" (likely related to occlusive disease), for both open and endovascular repair, were excluded to increase the sensitivity of the data towards the specified thoracic aortic segment, and aneurysmal pathology rather than steno-occlusive pathology.

Data handling
The patient flow chart is depicted in Figure 1. The number of admissions (according to rupture status), the number of OSRs and endovascular procedures, and the number of deaths for ruptured and non-ruptured TAAs were collated for each year. Admission data for TAA are presented for 1 April 1998 to 31 March 2020. HES procedural data for TAA are presented for 1 April 2000 to 31 March 2020 because of difficulty in obtaining procedural and mortality data prior to 2000. Both admission and procedural data are presented as crude admission and procedure rates, defined as the number of admissions and procedures per number of patients aged 60 years or older living for a given year, respectively. Admission and procedural data according to five year age bands were not available. Sex specific age standardised death rates (ASDRs) are presented for the period 1 April 1998 to 31 March 2019, as 2020 data had not yet been released at the time of writing. ASDRs were calculated using the 2013 European Standard Population over the age of 60 years and stratified into 60 e 79 years and > 80 years for males and females, and for ruptured and nonruptured disease, respectively (see Supplementary Text S1).

Statistical analysis
Trends in admissions and procedures were assessed with linear regression using SPSS version 27 (IBM, Armonk, NY, USA). ASDR trends were assessed using Joinpoint regression modelling (Joinpoint Command Line Version 4.5.0.1) provided by the US National Cancer Institute Surveillance Research Program according to the authors' previous methodology (see Supplementary Text S2). 3 Estimated annual percentage change (EAPC) for each significant change in trend with its confidence intervals (CIs) were calculated. To examine the impact of endovascular repair on ASDR, the ManneWhitney test was used to test the statistical significance of ASDR between the pre-endovascular era (1998 e 2008) and the endovascular era (2009 e 2019). This cutoff was chosen as TEVAR superseded OSR in 2008. An alpha level of < .05 was considered to be statistically significant for all analyses.

Hospital admission trends
The in hospital incidence for total admissions ranged between 5.95 per 100 000 in 1998 (95% CI 5.47 e 6.42) to 14.36 per 100 000 in 2020 (95% CI 13.71 e 15.00), with a mean increase of 6.4% per annum (linear regression r 2 ¼ .97; p < .001). The in hospital incidence for non-ruptured admissions ranged from 4.11 per 100 000 in 1998 (95% CI 3.71 e 4.50) to 12.61 per 100 000 in 2020 (95% CI 12.00 e 13.21) in 2020, with a mean increase of 9.3% per annum (r 2 ¼ .98; p < .001). Ruptured TAA admissions appeared relatively unchanged over the 22 year period (r 2 ¼ .10; p ¼ .082, Fig. 2A). Conversely, OSR appears to have declined over the last 23 years (r 2 ¼ .29, p ¼ .008). Figure 3 shows the results of the Joinpoint regression analysis for trends in mortality in males and females aged 60 e 79 years and > 80 years for ruptured and nonruptured TAA, respectively. The ASDR per 100 000 for each year were proportionally higher for ruptured disease irrespective of sex and age, with the highest ASDR for each year observed in women over the age of 80 years with ruptured disease. Very little difference was observed for ASDR for each year between men and women aged 60 e 79 years for ruptured and non-ruptured disease. A consistent declining trend in aneurysm related ASDRs from TAAs was observed over the studied 23 years for both ruptured and non-ruptured disease, irrespective of sex and age. The EAPCs, shown in Table 1, demonstrate statistically significant reductions in ASDRs from TAA in all groups. In men, the largest single trend of decreasing EAPCs was observed in those aged 60 e 79 years with ruptured disease (EAPC e6.30, 95% CI e5.90 e e6.80), whereas in females this was seen in those aged 60 e 79 years with non-ruptured disease (EAPC e4.40, 95% CI e2.70 e e6.10) and those

DISCUSSION
This population based cohort study examined "real world" trends in hospital admissions, operative intervention, and ARM from TAA in England over the last 23 years. The main study findings included a significant increase in hospital admissions for non-ruptured TAAs, accompanied by an increase in operative intervention, specifically a significant rise in the use of TEVAR, and an overall decrease in ARM for ruptured and non-ruptured disease, regardless of sex and age group.
Earlier interrogation of the English HES database (between 1999 and 2010) observed the decline in ARM to occur as early as 1999, well before the widespread implementation of TEVAR. 5 The authors attributed this to improved risk factor management, such as smoking cessation, blood pressure control, and increased prescription of statins, in line with the declining trends in mortality from other cardiovascular diseases. 13 While the prevalence of smoking in England fell substantially in the late 1970s and early 1980s (45% in 1974 to 35% in 1982) with the introduction of antismoking campaigns launched by the Health Education Council in England and Wales in the late 1960s, and other smoking cessation agreements between the tobacco industry and the government, 14 the rate of smoking decline has, in fact, slowed in the last 10 years. 15 Similarly, while the late 1990s to early 2000s saw a rise in statin initiation therapy for the primary prevention of cardiovascular disease in the UK, this was followed by a decline up to 2011, with rates remaining constant thereafter. 16 Furthermore, while awareness and treatment of hypertension have increased over time, much of the improvement occurred before the mid-2000s with awareness, treatment, and blood pressure control plateauing since then. 17 Improved risk factor management cannot wholly explain the ongoing decline in the TAA mortality rate.
Further analysis of the HES database over the 23 years of this study suggests that improved case ascertainment, the increased uptake of prophylactic repair of intact (nonruptured) TAAs, and, specifically, the preferential use of TEVAR may provide additional plausible explanations for the declining ARM rate. This is substantiated by the increased number of admissions for non-ruptured TAAs that has mirrored the sharp rise in TEVAR, alongside declining trends in ruptured (as well as non-ruptured) ARM.
Improved case ascertainment is probably the result of advances and the widespread availability of sensitive cross sectional imaging introduced in the 1970s and early 1980s. Similar rising trends in hospital admissions for TAA have been observed in other contemporary studies from Germany, 18 Canada, 19 and the USA. 20 Improved case ascertainment is likely to have contributed to the prevention of premature rupture related deaths through early identification, monitoring of aortic growth rate through regular surveillance, and prophylactic repair of non-ruptured aneurysms when treatment indication predicated on aortic diameter is met. 21 While there is no screening programme for TAA in the UK (or in other countries), the UK NHS Abdominal Aortic Aneurysms Screening Programme (NAAASP, implemented in 2009) may represent another pathway for the increased detection of synchronous or metachronous TAAs; 19.2% of patients with abdominal aortic aneurysm are known to have TAAs, permitting the (incidental) detection of TAAs upon referral to a vascular service for a "screen detected" abdominal aortic aneurysm. 22 The decline in ARM is perhaps less likely to be related to a decline in ruptured disease incidence due to risk factor management given that admission numbers for ruptured TAAs have been steady for the last 23 years.
The declining ARM for both ruptured and non-ruptured disease in this observational analysis parallels the supplantation of OSR by TEVAR. This finding, together with the significant reductions in ARM in the endovascular era compared with the pre-endovascular era, highlight the influence of minimally invasive endovascular repair in preventing deaths from TAA rupture. The management of TAAs has seen significant changes in past decades, with a shift away from OSR to TEVAR in the UK and globally. A study of 9 518 TEVARs for the treatment of thoracic aortic disease based on VASCUNET, an international vascular registry collaboration of 13 countries, reported that TEVAR has become the primary surgical treatment modality for descending aortic pathologies, with TEVAR constituting between 40% and 72% of TAA repairs in the USA (40%), New Zealand, Sweden, France, Spain, Switzerland, Australia, Italy, Finland, Germany, Serbia, Hungary, and the UK (72%). 23 This is unsurprising given that OSR remains a high risk procedure, even in experienced high volume centres. 24 While there have been no randomised trials to support the use of TEVAR over OSR, the EVAR trials have provided an indirect evidence base, 25,26 with single centre, 27 industry sponsored registries, 28,29 and population based studies 30e33 all confirming the lower complication rate and early survival advantage of TEVAR vs. OSR. While the significantly improved peri-operative mortality of TEVAR has led to increasing adoption of this treatment modality, the longer term outcomes remain less clear, with uncertainty regarding the effectiveness of TEVAR on mid to long term mortality and survival on a population basis. This was driven, in part, by long term data from the EVAR trials, which demonstrated an increased risk of ARM after eight years owing to concerns regarding the durability of the stent graft technology (increased secondary/late intervention and aneurysm rupture with EVAR vs. OSR). 34 Similarly, early analysis of the English HES data (2006 e 2011) 35 and Medicare data (2004 e 2007) 36 both demonstrated significantly worse survival with TEVAR compared with OSR at five years. However, this was mainly due to cardiopulmonary events rather than aortic related mortality, re-affirming the paramount importance of patient selection.
ARM in this dataset represents both in and out of hospital deaths and therefore encompasses peri-operative mortality, as well as long term mortality, irrespective of whether the patient underwent re-intervention. While this dataset does not permit identification of re-interventions, the declining trends in ASDR from ruptured and nonruptured TAA over the last 23 years, and the significant reductions in ASDR between the pre-endovascular era and the endovascular era, provide further support for the clinical efficacy of TEVAR. Similar findings were also observed on further interrogation of the Medicare database (1999 e 2010) with longer follow up whereby TEVAR was associated with a greater mean patient survival than OSR up to nine years, leading the authors to support TEVAR as the first line treatment for thoracic aortic aneurysms. 30  An interesting observation from the present study is the decline in ARM from TAA for two high risk groups; people over the age of 80 years, and in women, with the greatest rate of decline observed in ruptured disease, which parallels the increased use of TEVAR. While TEVAR has increased the overall pool of patients eligible for operative intervention who, in the past, may have been turned down for OSR due to their advanced age and comorbidities, the benefit of TEVAR in octogenarians remains a subject of debate, especially in acute settings, owing to their frail physiology and limited life expectancy. 37e39 Female sex has also been associated with poorer surgical outcomes. 40 This is likely to be multifactorial and, in part, due to the higher age (and therefore greater comorbidities) of women at diagnosis and intervention, and due to the anatomical differences (smaller diameter access vessels) between men and women that make endovascular repair more technically challenging in women. 41e43 Nevertheless, these findings provide further support for the use of endovascular technology in the treatment of TAA. The proportionally higher ASDRs each year supports the need for ongoing efforts to improve the management of TAA in this group of patients.

Limitations
There are some limitations to this study that should be acknowledged. As with all administrative databases, the dataset used in this study is subject to coding errors, despite the rigorous training that clinical coders undertake to translate data, which may have influenced the results. However, in the absence of randomised trial data, the data presented in this study represent the best available evidence to ascertain the efficacy of TEVAR. The trends are also largely in keeping with two HES population based studies of an earlier time period. 5,35 Both ICD-9/10 and OPCS-4 codes are found to be wanting; neither diagnostic nor procedural codes separate the anatomical segments of the aorta sufficiently to reflect contemporary endovascular practice, which has undergone significant evolution to treat specific segments of the aorta, affecting the interpretation of the dataset. The lack of admission and operative data stratified by age and gender, and the lack of mortality data according to operative intervention precluded further understanding of the declining ASDRs and comparative analysis between OSR and TEVAR, respectively. Similarly, the lack of patient level anatomical and clinical data in these publicly available administrative sets precluded analysis of predictors of mortality and potential confounders; hence, in this observational study attributing causality to the findings was avoided. To focus on capturing predominantly degenerative aneurysms, it was elected to exclude patients under the age of 60 years. This will have led an intentional selection bias against the ascertainment of TAA, which is often the result of post-dissection dilatation in younger patients with aortopathies. While the absolute reduction in the number of post mortem examinations being conducted in England and Wales may exaggerate the declining trends in TAA mortality, the proportion of reported deaths requiring a post mortem has remained stable over the last 20 years. 44 Finally, the last 20 years have seen major changes in the delivery and organisation of vascular services in the UK. There has been centralisation and an appreciation of volume outcome relationships, with set standards for all hospitals offering aortic surgery endorsed by the Vascular Society of Great Britain and Ireland, and implementation through a national Quality Improvement Programme underpinned by a multidisciplinary approach that spans the disciplines of surgery, anaesthesia and critical care, and interventional radiology. These complex issues are also likely to have influenced the data presented herein but could not be quantified further due to data availability.

Conclusion
Contemporary analysis of a "real world" dataset has demonstrated a significant increase in hospital admissions for non-ruptured TAA in the last 23 years in England, paralleled by a major shift towards endovascular repair, and significant declining trends in aneurysm related mortality, irrespective of sex and age. The significant reductions in ASDR from ruptured and non-ruptured TAA in the endovascular era, particularly for women aged > 80 years with ruptured disease affirm the positive impact of an endovascular approach to TAA.