A New Preoperative Predictor of Outcome in Ruptured Abdominal Aortic Aneurysms: The Time Before Shock (TBS)

Article Outline

• Abstract
• Introduction
• Patients and Methods
  • Cohort of Patients and Their Characteristic
  • Statistical Methods
• Results
  • Univariate Analysis
    • TBS
    • Shock
    • Severity of bleeding
    • Patient age
    • Comorbidities
    • Gender
  • Multivariate Analysis
• Discussion
• Conclusion
• References
• Copyright

Background

In patients with ruptured abdominal aortic aneurysm (RAAA) and shock, the time lag between the onset of the symptoms due to RAAA and the presence of a full developed shock syndrome was evaluated to assess its prognostic meaning. This time lag was called time before shock (TBS).

Methods

Ninety-four patients operated on between 2002 and 2007 have been retrospectively analyzed regarding TBS and the following parameters: presence of shock, severity of bleeding, age, comorbidities, and gender. According to TBS, on a 10-hour cutoff value, three groups of patients were distinguished: patients with TBS of 10 or less (short TBS), patients with TBS greater than 10 (long TBS), and patients without shock. The relationship of these variables with intraoperative and 30-day mortality was analyzed by both univariate and multivariate analyses.

Results

In the univariate analysis, patients with short TBS presented with four-fold mortality compared to patients without shock (p=0.000), whereas the increase in mortality of the patients with long TBS was nonsignificant (p=0.448). The mortality in patients with shock (presence of shock) was 3.7 times higher than in patients without shock (p=0.001). The mortality related to massive bleeding was 3.7 times higher than that associated with moderate bleeding (p=0.001). An increased mortality with borderline significance level was observed in patients older than 75 years (p=0.052). The relationship of mortality to the presence of comorbidities and gender was not significant. In the multivariate analysis, the mortality among the patients with short TBS was clearly highest, after either massive or moderate bleeding. In the logistic model with TBS, the Wald test showed as significant both short TBS (p=0.001) and severity of bleeding (p=0.033) but not age (p=0.103) and long TBS (p=0.0401). The model with TBS presented a better performance than that with shock, showing higher sensitivity, higher values of Youden's J, and a greater proportion of the total variation in mortality. Through the model with TBS, two groups of patients (those 75 years or younger with massive bleeding and those older than 75 years with moderate bleeding), both with short TBS, presented with a high risk of death not predicted by the model with shock.

Conclusion

TBS seems to complete the information given by the parameter “presence of shock,” and its evaluation allows a more effective judgment of the risk of death, at emergency admission of patients with RAAA.

Back to Article Outline

Introduction 

Overall mortality from ruptured abdominal aortic aneurysm (RAAA) is still relevant, despite significant improvements in surgery, anesthesia, and postoperative critical care.¹, ², ³, ⁴, ⁵, ⁶, ⁷, ⁸, ⁹ The presence of shock, together with the need of cardiopulmonary resuscitation, has been proved to be the most adverse prognostic factor, able to predict mortality above than expected from RAAA alone. The time lag between the beginning of the symptoms, due to RAAA, and the presence of a full developed shock syndrome has never been recorded in previous studies.

In this paper, we evaluate this time lag, named time before shock (TBS) to correlate early versus late shock onset, in the light of the final outcome of RAAA patients.

Back to Article Outline

Patients and Methods 

In a retrospective study of 94 consecutive patients undergoing open RAAA repair between January 2002 and February 2007, the prognostic meaning of shock (assumed as “presence of shock”) and TBS was evaluated, together with severity of bleeding, age, comorbidities, and gender. Shock was defined as an insufficient perfusion of vital organs with consecutive imbalance between oxygen supply and demand, due to intravascular volume deficiency with critically impaired cardiac preload.

Systolic blood pressure of less than 100 or shock index greater than 1.0 has been used as clinical findings of hypovolemic shock. Other signs were also considered: drop in systolic blood pressure of greater than 40 mm Hg from baseline, pallor, fainting, cool and clammy skin, delayed capillary refill, tachycardia, tachypnea, oliguria, and altered mental status.¹⁰, ¹¹

According to TBS, three groups of patients were distinguished: patients with TBS of 10 hours or less (short TBS), patients with TBS greater than 10 hours (long TBS), and patients without shock. Regarding severity of bleeding, two groups of patients were defined: those with moderate bleeding (moderate retroperitoneal hematoma) and those with massive bleeding (massive retroperitoneal hematoma or hemoperitoneum).¹² The age cutoff value was set at 75 years.

Cohort of Patients and Their Characteristic 

Patients' mean age was 70 years for men (standard deviation, ±7.1; range, 51-88) and 75 years for women (standard deviation, ±3; range, 71-83). There were 88 men (96.6%) and 6 women (6.4%).

Moderate (extending from the renal to the iliac arteries) retroperitoneal hematoma was present in 43 patients (45.8%); massive (from the diaphragm to the pelvis) retroperitoneal hematoma was present in 26 patients (27.6%); and massive retroperitoneal hematoma associated with an arterovenous fistula was present in 3 patients (3.2%)—an aortocaval fistula, an aortoiliac vein fistula, and an aorto (retroaortic) renal vein fistula; and hemoperitoneum was present in 22 patients (23.4%). A hematocrit value lower than 27 was found in 47.3% of patients with moderate retroperitoneal hematoma, in 71.7% of subjects with massive retroperitoneal hematoma, and in 82.9% of patients with hemoperitoneum.

Fifty-one patients (54.0%) had shock. Chronic obstructive pulmonary disease (COPD) affected 25 patients (26.6%); arterial hypertension, 23 patients (24.4%); coronary artery disease, 9 patients (9.6%); coronary artery disease with previous myocardial infarction (MI), 6 patients (6.4%); chronic renal failure, 3 patients (3.2%); coronary artery disease associated with peripheral arterial occlusive disease (PAOD), 3 patients (3.2%); coronary artery disease and COPD, 3 patients (3.2%); and PAOD, 3 patients (3.2%). Nineteen patients had no comorbidity (20.2%) (Table I).

Table I. Mortality and preoperative comorbidities

Preoperative Comorbidities	Total No. of Patients	Mortality	%
COPD	25	6	24.0
Coronary artery disease	9	0	0
Hypertension	23	5	21.7
PAOD	3	1	33.3
CRF	3	0	0
Coronary artery disease+ PAOD	3	2	66.6
Coronary artery disease+ COPD	3	1	33.3
Coronary artery disease + previous Myocardial infarction	6	5	83.3
Absence	19	7	36.8

COPD, chronic obstructive pulmonary disease; PAOD, peripheral arterial occlusive disease; CRF, chronic renal failure.

Statistical Methods 

In univariate analysis, the Fisher exact test (two-tailed p value) was applied. Taylor series and Greenland-Robins 95% confidence limits were used for the risk ratio and the Mantel-Haenszel combined risk ratio, respectively. To compare risk ratios in the stratified analysis, a Chi-squared test for homogeneity was performed.

Multivariate analysis was performed through the multiple logistic regression model, including the variables found to be significant in the univariate analysis. The proportion of the total variation in mortality explained by the complete model was estimated by the pseudo-R². Contribution of each variable to mortality was evaluated through the decrease in the pseudo-R² obtained by a backward elimination of each variable according to its significance level. Significance of each variable was evaluated by the Wald and likelihood ratio test.

The performance of the logistic model was evaluated by its ability to correctly classify the patients as dead or not dead. The classification of patients by logistic regression was performed assuming as dead patients those with a predicted death probability greater than 0.5.

Back to Article Outline

Results 

Morbidity was 22.3% (95% confidence interval [CI]=14.4-32.1, 21 patients); overall mortality was 28.7% (95% CI=19.9-39.0, 27 patients). Causes for death are listed in Table II. The mean amount of transfused blood was 4987±2883 mL in patients not surviving surgery and 1783±987 mL in survivors.

Table II. Postoperative complications and related mortality

ARDS, acute respiratory distress syndrome; MI, myocardial infarction; MOF, multiorgan failure; TIA, transient ischemic accident.

Univariate Analysis 

Results are shown in Table III.

Table III. Summary of results of univariate analysis

Variable	Mortality	%	RR	95% CI	p Value
Age (yr)
>75	9/19	47.4	1.97	1.06-3.68	0.053
≤75	18/75	24.0
Gender
Female	3/6	50.0	1.83	0.77-4.37	0.349
Male	24/88	27.3
Severity of bleeding
Massive bleeding	22/51	43.1	3.71	1.53-8.96	0.001
Moderate bleeding	5/43	11.6
Time before shock
≤10 hours	18/31	58.1	4.99	2.08-11.99	0.000
>10 hours	4/20	20.0	1.72	0.51-5.73	0.448
Absence of shock	5/43	11.6
Shock
Present	22/51	43.1	3.71	1.53-8.96	0.001
Absent	5/43	11.6
Preoperative comorbidities
Present	20/75	26.7	0.72	0.36-1.45	0.404
Absent	7/19	36.8

RR, risk ratio.

Compared to patients without shock, those with short TBS presented a quintupled statistically significant mortality. Increase of mortality in the patients with long TBS was not significant.

The relation between shock and mortality was statistically significant. Mortality in patients with shock was 3.7 times greater than in patients without shock.

Mortality was significantly higher in patients with massive bleeding.

Mortality among patients older than 75 years was higher than in younger patients, reaching a borderline significance level.

Table I reports mortality related to each type of comorbidity. Relative mortality was not statistically significant. Even the particularly high mortality related to coronary artery disease associated with PAOD and to coronary artery disease with previous MI did not reach statistical significance.

A higher mortality was observed in females, but the small number of women did not allow an appropriate statistical evaluation.

Multivariate Analysis 

Results of two different multiple logistic regression models with TBS and with shock, respectively, are shown in Table IV. The proportion of total variation in mortality explained by the model including shock (pseudo-R²=0.205) was lower than that including the TBS (pseudo-R²=0.244). In the model with TBS the Wald test showed a significant role for short TBS (p=0.001) and for severity of bleeding (p=0.033), but not for the age (p=0.103) and long TBS (p=0.401).

Table IV. Logistic regression model for mortality with TBS or with shock

	Regression with TBS			Regression with Shock
variable	β	SE (β)	p Value	β	SE (β)	p Value
Age (yr)
>75	1.0429	0.6389	0.103	1.1010	0.6109	0.072
Severity of bleeding
Massive bleeding	1.2829	0.6027	0.033	1.4828	0.5861	0.011
Time before shock
≤10 hours	2.0641	0.6385	0.001
>10 hours	0.6443	0.7666	0.401
Shock
Present				1.5602	0.5911	0.008
Pseudo-R2=	0.244			0.205

Two different logistic regression models are shown:

β = regression coefficient; SE (β)=β standard error; p=Wald test p value.

Estimated regression coefficients and their standard errors (SE) are reported for the variables included in the regression model.

TBS accounted for 71.2% of the total variation in mortality explained by the model; severity of bleeding, for 19.1%; and age, for 9.7%. In the model including the shock variable, the Wald test showed a significant role for shock (p=0.008) and for severity of bleeding (p=0.011), and near significance for age (p=0.071). The shock variable accounted for 52.2% of the total variation in mortality explained by the model; severity of bleeding, for 33.5%; and age, for 14.2%. In the logistic model with TBS no-significant interaction was observed between TBS and severity of bleeding, indicating no effect modification between TBS and severity of bleeding.

Risk ratios of short and long TBS stratified by severity of bleeding were reported in Table V. Mortality in short TBS patients was significantly higher than that of patients without shock: 3.8 times higher after massive bleeding and 3.7 times after moderate bleeding. Mortality in long TBS patients was slightly higher than that of patients without shock: 1.7 times higher after massive bleeding and 1.3 times after moderate bleeding. Nonsignificant heterogeneity was observed between massive and moderate bleeding for the risk ratios of the short and long TBS. The M-H combined risk ratios were 3.76 and 1.57, respectively.

Table V. Mortality related to TBS in both types of bleeding

RR, risk ratio; RR_M-H, Mantel-Haenszel combined risk ratio.

The performances of the two logistic regression models, with TBS and with shock, were evaluated and showed in Table VI. In the model with TBS, the sensitivity and the Youden's J were much greater than those in the model with shock. In the model with shock, the apparent prevalence was significantly lower than the true prevalence for both all patients (p=0.001) and the patients with shock (p=0.004). In the logistic model with TBS (Table VII), two groups of patients with short TBS—those 75 years old or younger with massive bleeding and those older than 75 years with moderate bleeding—presented a high expected risk of death not predicted by the model with the shock.

Table VI. Performance of the two logistic regression models applied to total patients (a) and to patients with shock only (b)

The classification by logistic regression was performed considering as dead patients those with a predicted death probability >0.5.

Table VII. Mortality probability predicted by two logistic regression models

Back to Article Outline

Discussion 

The outcome of patients with rapid onset of shock was worse compared to the group of patients with late shock development, and short TBS was shown to be a significant predictor of death, as shock itself and massive hemorrhage. As for the other variables, the age reached a borderline significance in the univariate analysis, but not in the multivariate analysis, and gender and comorbidities also appeared not significant. Only six patients were female and both comorbidities with higher mortality (coronary disease associated with PAOD and coronary disease with previous MI) included a small number of patients: the small size of these three groups did not allow a fair statistical analysis.

TBS appeared to complete the information given by the presence of shock and the use of TBS allows a more effective judgment of the risk of death. Mortality among patients with short TBS (early shock) was clearly higher than for patients with long TBS (late shock), after either massive or moderate bleeding. Interestingly, the risk of death related to long TBS was not significant in univariate and multivariate analyses.

The logistic regression model with the variable TBS presented a better performance than the similar model with the variable shock, showing a higher sensitivity and higher values of apparent prevalence and of Youden's J.

Indeed, through the model with TBS, two groups of patients with short TBS—those 75 years or younger with massive bleeding and those older than 75 years with moderate bleeding—presented a high risk of death not predicted by the model with shock.

Patients who develop shock typically pass through three stages. In stage I (also called compensated), compensatory mechanisms avoid organ hypoperfusion. In stage II (also called decompensated), despite intense activity of compensatory mechanisms, organ hypoperfusion begins. In stage III, compensatory mechanisms fail, blood pressure decreases and there is clear evidence of organ hypoperfusion.¹³, ¹⁴ We believe that patients with TBS ≤ 10 hours are those with shortest stages I and II because of their own feeble compensatory mechanisms, reasonably due to physiologic and/or pathologic features (e.g., age, limited physiologic reserve of one or more organs, previous severe cardiorespiratory illness).

Back to Article Outline

Conclusion 

The TBS evaluation at emergency admission of patients with RAAA could quickly point out those with a worse prognosis.

Back to Article Outline

References 

1. Crawford ES. Ruptured abdominal aortic aneurysm: an editorial. J Vasc Surg. 1991;13:348–350
   • View In Article
   • Full Text
   • Full-Text PDF (285 KB)
   • CrossRef
2. Kniemeyer HW, Kessler T, Reber PU, et al. Treatment of ruptured abdominal aortic aneurysm, a permanent challenge or a waste of resources? Prediction of outcome using a multi-organ-dysfunction score. Eur J Vasc Endovasc Surg. 2000;19:90–196
   • View In Article
   • Abstract
   • Full-Text PDF (239 KB)
   • CrossRef
3. Johansson G, Swedenborg J. Little impact of elective surgery on the incidence and mortality of ruptured aortic aneurysms. Eur J Vasc Endovasc Surg. 1994;8:489–493
   • View In Article
4. Holt PJ, Poloniecki JD, Gerrard D, et al. Meta-analysis and systematic review of the relationship between volume and outcome in abdominal aortic aneurysm surgery. Br J Surg. 2007;4:395–403
   • View In Article
5. Qureshi NA, Rehman A, Slater N, et al. Abdominal aortic aneurysm surgery in a district general hospital: a 15-year experience. Ann Vasc Surg. 2007;21:749–753
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (236 KB)
   • CrossRef
6. Patel ST, Korn P, Haser PB, et al. The cost-effectiveness of repairing ruptured abdominal aortic aneurysms. J Vasc Surg. 2000;32:247–257
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (65 KB)
   • CrossRef
7. Chen JC, Hildebrand HD, Salvian AJ, et al. Predictors of death in nonruptured and ruptured abdominal aortic aneurysms. J Vasc Surg. 1996;24:614–623
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (973 KB)
   • CrossRef
8. Marty-Ané CH, Alric P, Picot MC, et al. Ruptured abdominal aortic aneurysm: influence of intraoperative management on surgical outcome. J Vasc Surg. 1995;22:780–786
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (702 KB)
   • CrossRef
9. Johnston KW the Canadian Society for Vascular Surgery Aneurysm Study Group. Ruptured abdominal aortic aneurysm: six-year follow-up results of a multicenter prospective study. J Vasc Surg. 1994;19:888–900
   • View In Article
   • Abstract
   • Full Text
   • CrossRef
10. Weil MH, Gazmuri RJ. Shock: new development in the management of shock. Appl Cardiopulm Pathophysiol. 1991;4:103–107
    • View In Article
    • MEDLINE
11. Jones AE, Aborn LS, Kline JA. Severity of emergency department hypotension predicts adverse hospital outcome. Shock. 2004;22:410–414
    • View In Article
    • MEDLINE
    • CrossRef
12. Rabitti G, Creazzo V, Bertoletti G, et al. Surgical treatment of ruptured infrarenal aortic aneurysms: early and late results in a nine-year experience with 65 patients. Angiology. 1982;33:349–356
    • View In Article
    • MEDLINE
    • CrossRef
13. Shoemaker WC. Shock states: pathophysiology, monitoring, outcome prediction, and therapy. In: Textbook of Critical Care. ed 2. Philadelphia, PA: Saunders: The Society of Critical Care Medicine; 1989;p. 977–993
    • View In Article
14. Kumar V, Cotran RS, Robbins SL. Robbins basic pathology. In:  Kumar V,  Cotran RS,  Robbins SL editor. Philadelphia, PA: WB Saunders; 2003;
    • View In Article