Hospital volumes and later year of operation correlates with better outcomes in acute Type A aortic dissection^† Free

Abstract

INTRODUCTION

Acute Type A aortic dissection (ATAAD) represents two-thirds of all acute thoracic dissections and, by definition, involves the ascending aorta and to varying degree the aortic arch and the descending aorta [1]. Death and complications occur due to free rupture of the aorta, cardiac tamponade, acute aortic insufficiency as well as from complications related to malperfusion syndromes such as stroke, myocardial ischaemia and mesenteric ischaemia [1–3]. ATAAD requires immediate surgical intervention, because the in-hospital mortality rate is excessively high without surgery [4]. Contemporary methods of Type A repair include the use of hypothermic circulatory arrest (HCA), suturing the distal anastomosis without the application of aortic cross-clamp (open distal anastomosis), resection of primary tear, hemiarch and/or ascending aortic replacement with tube graft and repair or replacement of the aortic root as well as developments of dedicated centres with aortic dissection programmes [5, 6]. The short-term outcome following ATAAD repair has improved over the last 2 decades, but the specific reasons for this improvement are not well defined [4, 7, 8]. The Nordic Consortium for Acute Type A Aortic Dissection (NORCAAD) is a collaborative effort of 8 hospitals in Denmark, Finland, Iceland and Sweden, where data were collected for patients operated on for ATAAD from 2005 to 2014. The aim of this study was to analyse short-term outcomes following ATAAD repair, postulating that short-term mortality has decreased due to modulation of specific risk factors and operative techniques.

MATERIALS AND METHODS

Data were retrospectively collected for all patients who underwent an operation for acute Type A aortic dissection from 1 January 2005 to 31 December 2014 at the following NORCAAD centres: Aarhus University Hospital, Skejby, Denmark; Karolinska University Hospital, Stockholm, Sweden; Landspitali University Hospital, Reykjavik, Iceland; Orebro University Hospital, Orebro, Sweden; Sahlgrenska University Hospital, Gothenburg, Sweden; Skane University Hospital, Lund, Sweden; Tampere University Hospital, Tampere, Finland; Turku University Hospital, Turku, Finland. Data were not collected for patients admitted to hospital centres not undergoing an operation for ATAAD. All participating sites were cardiac surgery referral centres. The study design has been previously described with a total of 194 clinical variables were collected for each patient [9]. Postoperative complications were generally defined as in the Society of Thoracic Surgeon data set and are listed in Supplementary Methods. The length of hospital stay was recorded in days, both in the intensive care unit (ICU) and on a general ward at the hospital where the operation was performed. Primary outcome measure for the study was 30-day mortality, but the rationale for using 30-day mortality instead of hospital mortality was to circumvent potential difference in discharges to referring hospitals or rehabilitation centres. National identity registries exist in all the Nordic countries allowing for complete capture of survival status so patients were not lost to follow-up after hospital discharge. The study was approved by each of the participating institutions. Patients were coded prior to inclusion into the database. As individual patients were not identified, obtaining individual consent for the study was waived.

Statistical analysis

Continuous variables were expressed as mean ± standard deviation and categorical variables as percentages. Student’s t-test was used for continuous variables and χ² and Fisher’s exact test for categorical variables. Odds ratios for predictors of 30-day mortality were estimated with logistic regression. Independent predictors of 30-day mortality were identified with multivariate logistic regression. Variables were selected into initial model based on the degree of significance from univariate regression and clinical relevance. This included age, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, prior sternotomy, preoperative arrest, cardiac tamponade, hypotension/shock, any malperfusion syndrome, root replacement, total arch replacement, open distal anastomosis, cardiopulmonary bypass time, Penn Class, total hospital volume, annual hospital volume, use of antegrade cerebral perfusion and calendar year of surgery. Cases with missing values for the above variables were excluded from final analysis. There were 117 events of 30-day mortality in the final model. Using forward and backward stepwise selection and multivariate logistic regression model, we selected the final model (Table 5) based on the lowest Akaike information criterion, which was used as a measure of the relative quality of the model. Receiver operating characteristic curve was used to assess discrimination of the logistic model. The final model had the highest area under the curve (0.807) of all tested models. Data were collected using Microsoft Excel^® (Microsoft, Redmond, WA, USA), and statistical calculations were performed using R [10].

RESULTS

There were a total of 1159 operations performed over the 10-year period at the 8 participating institutions (mean 144.9 and range 30–262). The annual number for each hospital ranged from 1 to 44 and has been previously described [9]. This is further delineated in Supplementary Material, Table S1. The total number of operations performed per year increased from 85 in 2005 to 150 in 2014 (Fig. 1).

Figure 1

Number of operations performed for acute Type A aortic dissection each year from 2005 to 2014. Line shows linear regression, and the shaded area indicates 95% confidence interval. Odds ratio 1.130, P < 0.001.

Open in new tabDownload slide

Demographics of the patients are summarized in Table 1. Males were 68% of the cohort, and the mean age was 61.6 ± 2.2 years. The majority of the patients (52%) had a history of hypertension and history of smoking was present in 34% of the patients. Bicuspid aortic valve was present in 6% of patients, 10% had a known history of thoracic aortic aneurysm (TAA), 8% had a family history of aortic dissection and 5% had a family history of TAA.

Symptoms on presentation are summarized in Table 2. Nearly all of the patients (85%) had chest pain on presentation, almost quarter of the patients (24%) were hypotensive or in shock (SBP <90 mmHg) on arrival to a hospital and 17% had cardiac tamponade. Malperfusion syndrome was present in 28% of the patients and most commonly affected lower extremities (16%), the brain (9%) and the heart (9%).

Operative variables are summarized in Table 3. With respect to proximal repair, the root was replaced in 26% of cases, most often by performing a mechanical composite root replacement (68%). With regard to distal repair, either ascending aortic or hemiarch replacement was performed in majority of the cases, while the total arch was replaced in 6% of the patients. Open distal anastomosis performed without application of aortic cross-clamp during HCA was performed in 85% of cases with varying perfusion strategies. The primary tear was identified in majority of patients in the ascending aorta and was excised in 78% of cases.

Postoperative complications are summarized in Table 4. Complications occurred in 75% of the patients who arrived to the ICU. The most common complications were new-onset postoperative atrial fibrillation (39%), prolonged time on a ventilator (33%), acute kidney injury (39%), mental status changes (22%), reoperation for bleeding (22%) and new perioperative stroke (20%). Median stay in the ICU was 4 (mean 5.8 ± 4.8) days and the median total hospital stay was 9.5 (mean 14.2 ± 16.9) days. Throughout the study period, the 30-day mortality was 18% and hospital mortality was 16%, which included deaths occurring in the operating room in 7% of cases in 7% of cases. The 30-day mortality decreased from 24% in 2005 to 13% in 2014 [odds ratio (OR) 0.921, 95% confidence interval (CI) 0.873–0.971, P = 0.003] (Fig. 2). The rate of intraoperative death did not change significantly throughout the study period (OR 0.940, 95% CI 0.866–1.020, P = 0.134).

Figure 2

Thirty-day mortality following operations for acute Type A aortic dissection from 2005 to 2014. Line shows linear regression, and the shaded area indicates 95% confidence interval. Odds ratio 0.921, P = 0.003.

Open in new tabDownload slide

Univariate logistic regression for 30-day mortality was performed for all demographics, medical history, clinical symptoms on presentation, method of diagnosis and operative variables. All significant (P < 0.05) variables associated with 30-day mortality are presented in Supplementary Material, Table S2. Temporal change of variables associated with 30-day mortality and potentially clinically relevant variables were assessed by linear regression analysis. This revealed significant increased use of open distal anastomosis technique (OR 1.117, 95% CI 1.055–1.183, P < 0.001), increased use of antegrade cerebral perfusion (OR 1.214, 95% CI 1.152–1.282, P < 0.001) and increased number of operations performed per year (OR 1.130, 95% CI 1.103–1.157, P < 0.001) for each consecutive year of the study.

Using stepwise selection and multivariate logistic regression analysis, independent positive predictors for 30-day mortality were determined to be preoperative arrest, any malperfusion syndrome, Penn Class C, Penn Class B and C as well as cardiopulmonary bypass time (Table 5). Later calendar year of operation (2004–2015) and total hospital volume, however, predicted improved 30-day mortality (Table 5).

DISCUSSION

The primary finding of this study is that the mortality of ATAAD has decreased significantly over the last 10 years, and this correlates with higher operative volume for ATAAD. In our cohort, the demographics, presentation and clinical condition of the patient prior to surgery are remarkably similar to the International Registry of Acute Aortic Dissection cohort, although history of hypertension and diabetes is less frequent in the present cohort [7]. The patients commonly arrived in a critical condition to the hospital, reflected by almost a quarter of the patients being in shock and almost one-sixth having cardiac tamponade. Malperfusion syndrome was present in one-quarter of patients, which is similar to reports from other centres [3, 11, 12]. As expected, postoperative complications were common, especially reoperation for bleeding, prolonged ventilator requirement and neurological complications. Rates of complications, however, are similar to most comparable studies from Europe and USA, although neurological complications in our study appear to be more frequent than reported in studies from the USA [5, 8, 13–15].

Operations for ATAAD remain high risk with contemporary data reporting operative mortality ranging between 12% and 31% [7, 8, 14, 16]. Although individual centres have reported mortality rates of 12–14%, and even as low as 3%, major registry and national database studies describe higher rates of 20–30%, which probably represent the true mortality rates of operation for ATAAD in most cardiac surgery centres in North America and Europe [5–7, 14, 16, 17]. In the present study, we observed a mean 30-day mortality of 18% over the 10-year study period, which is in accordance with both registry and single-centre studies mentioned previously. The 30-day mortality decreased significantly from 24% in 2005 to 13% in 2014, confirming the general notion among aortic surgeons that the outcome for ATAAD surgery is improving. Our findings support previously reported data from the Inpatient Medicare database, Nationwide Inpatient Sample, International Registry of Acute Aortic Dissection and a recent multicentre study in the USA, indicating reduced surgical mortality for ATAAD [7, 8, 14, 16].

We used stepwise selection and multivariate logistic regression analysis to identify independent predictors for 30-day mortality. Variables associated with higher risk of 30-day mortality included preoperative arrest, any malperfusion syndrome, Penn Class C (generalized ischaemia), Penn Class B and C (localized and general ischaemia) as well as longer time on cardiopulmonary bypass. Variables signifying precarious clinical condition on arrival to the hospital were the strongest determinants of short-term mortality, as has been distinctly outlined by the Penn Class stratification system [18, 19]. On the other hand, variables associated with decreased 30-day mortality were later calendar year of operation and higher total hospital volume.

When we analysed temporal variation (from 2005 to 2014) of risk factors for 30-day mortality, the number of operations performed per year, the use of open distal anastomosis technique and the use of antegrade cerebral perfusion techniques changed significantly during the 10 years of the study. The number of cases per year nearly doubled from 2005 to 2014. The rise in the number of operations for both TAA and dissection is well documented [20]. Whether this increase in the number of ATAAD cases is due to a rising incidence of aortic dissection is difficult to say, as it could also be related to improved diagnostics, increased awareness, more aggressive referral for an operation and/or change in patient selection. NORCAAD only contains data on patients undergoing an operation for ATAAD, and therefore, we can only speculate on that issue. The incidence for thoracic dissection in Sweden and Oxford, UK, appears to have increased over the last decades [20–22]. However, a large study from the US Medicare claim database showed no change in rates of hospitalizations for aortic dissections [16]. This is supported by a recent whole-population study from Iceland, indicating that the true incidence of acute thoracic aortic dissections has not changed significantly from 1992 to 2013; however, it did show a significant increase in the proportion of patients who underwent an operation for ATAAD [23]. Increased annual number of operations consequentially results in increased experience of both the centre and the surgeons treating ATAAD. This is in line with other studies demonstrating that increased experience in treating the condition is associated with better outcomes of ATAAD surgery and, most likely, explains the improved short-term outcome that we observed [14, 24]. As the current version of the NORCAAD database does not contain surgeon-specific information, we were not able to assess individual surgeon’s contribution to short-term outcome following surgery for AATAD. Based on these results, it rather appears that improvements in short-term mortality are dependent on cumulative hospital experience, which could include surgeon’s experience, better ICU care and systematic approaches to complex diseases.

The use of open distal anastomosis technique and antegrade cerebral perfusion increased significantly with every year of the study, but these techniques did not turn out to be independent predictors of 30-day mortality when controlled for total hospital volume in the multivariate logistic model. The rationale for use of open distal anastomosis includes better visualization of the dissected aorta, allowing for more accurate suture placement, proper exclusion of the false lumen as well as identification of primary tears and longitudinal arch tears. Despite being the preferred operative method to treat the distal aorta in ATAAD among aortic surgeons, there is surprisingly sparse evidence of the benefits of suturing the distal anastomosis under HCA without application of aortic cross-clamp, compared with closed anastomosis, where the aortic cross-clamp is left in place [25]. No randomized or prospective trial has studied this issue. The principal report by David et al. [26] showing that open anastomosis was associated with lower hospital mortality, lower risk of stroke, better long-term survival and higher rate of false lumen thrombosis was not large (n = 109 in total, 54 open) but resulted in a paradigm shift in how the distal anastomosis was approached. This study was followed by several studies that showed no difference in outcome when comparing open to closed distal anastomosis [27–30]. Only recently, a report from the USA demonstrated that the operative strategy of open anastomosis, deep HCA and antegrade cerebral perfusion was associated with improved overall survival [31]. Our study, however, seems to indicate that the use of open distal anastomosis by itself does not affect short-term mortality.

The main strength of NORCAAD is the uniform health care systems in Nordic countries with relatively stable homogeneous populations for which follow-up is nearly 100% complete through centralized national identity registries. The weaknesses of NORCAAD are inherent to all retrospective studies, such as incomplete data sets, missing potential confounders and treatment selection biases.

CONCLUSION

In conclusion, surgical mortality for ATAAD remains high but has decreased significantly over the last decade. In our study, this correlated with both increased number of operations performed per year and later year of operation, both representing the effects of increased operative experience on surgical outcomes. It should be expected that further advances in the management of this complex disease will continue to improve the short-term outcome of operative repair for ATAAD.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ACKNOWLEDGEMENTS

The following individuals also contributed to the study: Erik Björklund, Josefine Carrell, Erik Herou, Sigrun H. Lund, Inga H. Melvinsdottir, Johan Sjögren and Khalil Ahmad.

Funding

This work was supported by Reykjavik, University of Iceland Research Fund and Landspitali Research Fund and The Mats Kleberg Foundation, Stockholm.

Conflict of interest: none declared.

REFERENCES

Author notes