Validity of the Hardman Index to Predict Outcome in Ruptured Abdominal Aortic Aneurysm
Article Outline
- Abstract
- Introduction
- Patients and Methods
- Results
- Discussion
- Conclusions
- Acknowledgment
- References
- Copyright
This study assessed the validity of the Hardman index in predicting outcome following open repair of ruptured abdominal aortic aneurysm and whether this scoring system can be used reliably to select patients for surgical repair. Patients undergoing open repair of ruptured abdominal aortic aneurysm in two university teaching hospitals over a 5-year period were identified from a computerized hospital database. Thirty-day mortality was the main outcome measure. Five Hardman index factors were calculated and related to outcome retrospectively. There were 178 patients with a mean age of 73.9 years (range 51–94) and a male to female ratio of 5.4:1. The overall in-hospital mortality was 57.3% (102/178). Univariate analysis of risk factors showed that age >76 years (P = 0.007, odds ratio [OR] 2.34, 95% confidence interval [CI] 1.26-4.37) and electrocardiograghic evidence of ischemia on admission (P = 0.002, OR 3.75, 95% CI 1.57-8.93) were associated with high mortality. However, loss of consciousness (P = 0.155, OR 1.56, 95% CI 0.85-2.86), hemoglobin <9 g/dL (P = 0.118, OR 1.89, 95% CI 0.85-4.22), and serum creatinine >0.19 mmol/L (P = 0.691, OR 1.25, 95% CI 0.42-3.70) were not significant predictors of mortality. Using a multivariate analysis, age >76 years (P = 0.043, OR 2.29, 95% CI 1.03-5.11) and myocardial ischemia (P = 0.029, OR 2.93, 95% CI 1.12-7.67) were again found to be the significant predictors of mortality. The operative mortality was 44%, 46%, 68%, 79%, and 100% for Hardman scores of 0, 1, 2, 3, and 4, respectively. No patient had a score of 5. The Hardman index is not a reliable predictor of outcome following repair of ruptured abdominal aortic aneurysm. High-risk patients may still survive and should not be denied surgical repair based on the scoring system alone. Further evaluation of the risk factors is required to reliably and justifiably exclude those patients in whom the intervention is inappropriate.
Introduction
The prevalence of abdominal aortic aneurysm (AAA) is high, and despite recent advances in surgical, anesthetic, and intensive care techniques, the mortality for open repair remains high. A nation-based prospective study of ruptured AAA (rAAA) from Canada reported an early mortality of 51.4%.1 The community mortality from rAAA is 67-88%.2, 3, 4 In-hospital operative mortality differs from center to center depending on selection criteria. However, most centers report an operative mortality of 40-45%.5, 6 This figure combined with community mortality accounts for an overall mortality of >90%.
It has been suggested that operative mortality could be improved by selecting those patients for surgery who have a reasonable chance of survival. This would allow better use of limited hospital resources. Although some preoperative variables indicate increased mortality, an absolute risk is difficult to work out.7, 8 Various preoperative scoring systems have been proposed to help in this selection process, including POSSUM (Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity),9 the Glasgow Aneurysm Score,10 and the Hardman index.11 However, none of them is robust or validated to predict outcome, and hence, the decision to operate is often subjective.
A successful scoring system would accurately identify the subgroup of patients with 100% mortality, hence avoiding unnecessary surgery and the suffering of grieving relatives. Hardman's index has been popularized because of its simplicity and possible accuracy. This study aimed, firstly, to assess whether the Hardman index accurately predicts mortality in patients with rAAA and, secondly, to see whether this scoring system could be used reliably to select patients for surgery in our practice.
Patients and Methods
All patients undergoing open repair of rAAA over a 5-year period, from November 1999 to October 2004, in two university teaching hospitals in Belfast were included. These tertiary referral centers in vascular surgery provide services to Northern Ireland with a population of 1.7 million.12 Data were retrieved from medical case records and analyzed in a retrospective fashion.
The decision to operate was made by one of the seven consultant vascular surgeons based on individual cases. No patient was turned down for surgery on the basis of specific selection criteria. However, patients with advanced malignancy, dementia, debilitating cardiac or pulmonary conditions, and disabling stroke were not offered surgical repair.
The operation was performed either by or under the supervision of a consultant vascular surgeon. rAAA was defined as the presence of retroperitoneal or intraperitoneal blood from an AAA at laparotomy. Mortality was defined as death within 30 days of operation. Patients undergoing urgent repair of inflammatory or symptomatic AAA without evidence of rupture were excluded.
The Hardman index was calculated for all patients based on five risk factors recorded at presentation prior to any resuscitation. These include age >76 years, history of loss of consciousness, electrocardiographic (ECG) evidence of ischemia on admission (>1 mm ST segment depression or T-wave changes), hemoglobin <9 g/dL, and serum creatinine >0.19 mmol/L. A patient could score from 0 to 5 depending on the number of positive Hardman criteria on admission.
Statistical Analysis
Data were analyzed using SPSS 12.0.1 for Windows (SPSS, Chicago, IL). The univariate association of the five index factors was assessed by chi-squared or logistic regression analysis. A variable was excluded from analysis if its value was not recorded at the time of admission, and P < 0.05 was considered statistically significant. Multivariate analysis of risk factors was carried out by logistic regression, with variables selected by backward elimination using a significance level of 10%.
Results
Out of a total of 209 consecutive patients presenting with rAAA, 180 underwent open repair in the two hospitals during the 5-year study period. Twenty-nine patients (22 males, 7 females) declined surgical repair for the reasons stated in Table I. Case notes were missing for two patients undergoing surgical repair, and hence they were excluded from further analysis. The mean age was 73.9 years (range 51-94), with a male preponderance of 5.4 to 1. The mean age for men and women was 72.7 ± 7 and 79.9 ± 7.4 years, respectively. Overall in-hospital operative mortality was 57.3% (102/178). There was a slightly lower mortality for women (43%) compared to men (60%) (odds ratio [OR] 2.00, 95% confidence interval [CI] 0.88-4.53, P = 0.092), even though this did not reach statistical significance.
Table I. Reasons for not operating on patients with ruptured aneurysm (n = 29)
| Reason | Number of patients |
|---|---|
| Patient refused surgery | 11 |
| Poor cardiac risk | 6 |
| Hospital death before surgery | 5 |
| Senile dementia | 3 |
| Carcinoma colon | 2 |
| Severe COAD | 1 |
| Disabling stroke | 1 |
Loss of consciousness, hemoglobin, creatinine, and ECG on admission were recorded in 97.8% (174/178), 82.6% (147/178), 81.5% (145/178), and 78.7% (140/178) of cases, respectively. All five indices were recorded in 65% (115/178) at the time of admission. A univariate analysis was performed to assess the significance of each of the five Hardman index factors to predict operative mortality. Only age >76 years (P = 0.007) and myocardial ischemia on admission (P = 0.002) were predictive of mortality. However, loss of consciousness (P = 0.155), hemoglobin <9 g/dL (P = 0.118), and creatinine >0.19 mmol/L (P = 0.691) were not significant predictors of mortality (Table II). The Hardman variables were categorized into three groups, and the mortality for each group is shown in Table III. Age >76 years and evidence of myocardial ischemia on admission were associated with increased mortality compared to those who were not in these categories. Using the multivariate logistic regression analysis, age >76 years (P = 0.043) and myocardial ischemia (P = 0.029) were again found to be significant prognosticators of mortality (Table IV). Operative mortality increased with rising score: 44% (12/27), 46% (18/39), 68% (21/31), 79% (11/14) and 100% (4/4) for scores 0, 1, 2, 3, and 4, respectively (Fig. 1). No patient had a score of 5.
Table II. Univariate association with mortality on logistic regression for the five risk factors
| Variable | Number analyzed | OR | 95% CI | P |
|---|---|---|---|---|
| Age >76 years | 178 | 2.34 | 1.26–4.37 | 0.007 |
| Loss of consciousness | 174 | 1.56 | 0.85–2.86 | 0.155 |
| Hemoglobin <9 g/dL | 147 | 1.89 | 0.85–4.22 | 0.118 |
| Creatinine >0.19 mmol/L | 145 | 1.25 | 0.42–3.70 | 0.691 |
| ECG ischemia | 140 | 3.75 | 1.57–8.93 | 0.002 |
Table III. Mortality by subgroups of the five variables
| Variable | Number of patients | Death | Mortality (%) |
|---|---|---|---|
| Age (years) | |||
 >76 | 77 | 53 | 68.8∗ |
| ≤76 | 101 | 49 | 48.5 |
| History of LOC | |||
| Present | 78 | 49 | 62.8 |
| Absent | 96 | 50 | 52.1 |
| Unknown | 4 | 3 | 75 |
| Hemoglobin | |||
| <9 g/dL | 35 | 24 | 68.6 |
| ≥9 g/dL | 112 | 60 | 53.6 |
| Unknown | 31 | 18 | 58.1 |
| Creatinine | |||
| >0.19 mmol/L | 15 | 9 | 60 |
| ≤0.19 mmol/L | 130 | 71 | 54.6 |
| Unknown | 33 | 22 | 66.7 |
| ECG ischemia | |||
| Present | 38 | 30 | 78.9∗ |
| Absent | 102 | 51 | 50 |
| Unknown | 38 | 21 | 55.3 |
∗Increased mortality (P < 0.050) compared with other variables. |
Table IV. Multivariate analysis of risk factors by logistic regression using a significance level of 10% (n = 121)
| Variable | OR | 95% CI | P |
|---|---|---|---|
| Age >76 years | 2.29 | 1.03–5.11 | 0.043 |
| Hemoglobin <9 g/dL | 2.27 | 0.88–5.86 | 0.089 |
| ECG ischemia | 2.93 | 1.12–7.67 | 0.029 |
Fig. 1
Operative mortality by number of positive Hardman factors.
Discussion
A selective approach for the repair of rAAA should be based on a reliable scoring system, which is simple and able to accurately predict operative mortality. This system should be able to differentiate those patients who have no chance of survival from those who are likely to benefit from surgery in view of ethical and resource issues.
With mounting pressures on limited resources, most surgeons in the United Kingdom are beginning to favor a more selective approach for the repair of rAAA.13 However, such a selective policy must be based on a scoring system that is reliable and robust in its application. Until recently, Hardman's scoring system was regarded as a reliable tool to select patients for surgery in rAAA. The popularity of this system was based on its simplicity and theoretical accuracy. It has been verified by four independent series,14, 15, 16, 17 involving a total of 469 patients from five centers (Table V). Nevertheless, a recent study from Edinburgh18 has cast doubts on the validity of this selection scoring system and has reported a mortality of only 33% with a score of ≥3, in contrast to 100% mortality originally reported in the Hardman et al. study.11 Moreover, the study shows a relatively higher mortality (55%) with a score of 1.
Table V. Percentage operative mortality by number of positive criteria in different studies
| Study group, year | Total number | Score 0 | Score 1 | Score 2 | Score 3 | Score 4 |
|---|---|---|---|---|---|---|
| Sydney, 199611 | (154) | 16 | 37 | 72 | 100 | 100 |
| Plymouth, 199914 | (69) | 18 | 28 | 48 | 100 | 100 |
| Gloucester, 200315 | (188) | 35 | 55 | 74 | 90 | 100 |
| S'ampton/Adelaide, 200316 | (75) | 8 | 24 | 55 | 100 | 100 |
| Manchester, 200417 | (137) | 40 | 46 | 77 | 92 | 100 |
| Edinburgh, 200518 | (82) | 15 | 55 | 38 | 33∗ | |
| Belfast (current) | (178) | 44 | 46 | 68 | 79 | 100 |
∗Mortality figures for a score ≥3. |
This study shows that out of the five index factors only two (age >76 years and myocardial ischemia on admission) are significantly related to mortality. Although overall mortality increased with rising scores, in contrast to the original study by Hardman et al.,11 the current study shows that patients with a score of 3 had 79%, rather than 100%, mortality. If a decision not to operate was made in this group, three out of 14 patients would have been denied a life-saving operation. However, as there were only 18 patients with a score ≥3, the results of this study should be interpreted cautiously. Nonetheless, even in Hardman's study, only eight patients had scores ≥3.11
These results suggest that the Hardman scoring system is unreliable in predicting operative mortality and, hence, repair should not be denied based on this scoring system alone. A randomized study would be ideal for precise prediction of mortality, but obviously this may not be ethical in the setting of rAAA. However, a more elaborative evaluation of the risk factors is required before patients with rAAA can justifiably be excluded from surgery. Until such time, the decision to operate should be made on clinical grounds by the operating surgeon at the time of presentation.
A consistent apprehension with the validity of any scoring system used to decide whether surgery should be offered to patients with rAAA has been the availability of data at the time of patient presentation. In an emergency situation, a full set of the five Hardman variables may not always be available at the time of making a decision to operate. In this study, all five factors were recorded in only two-thirds of the patients. A similar finding was observed by Hardman et al.,11 with loss of consciousness, hemoglobin, creatinine, and ECG trace being available in, respectively, 98.1%, 86.4%, 83.8%, and 72.1% of patients only. Although, there is no mention of the total number of patients for whom all five factors were available, it could be speculated that this will be no more than 72.1%. Unfortunately, despite ECG evidence of myocardial ischemia being the most predictive indicator of mortality, it was available in the least number of patients in both series.
Conclusions
The findings of this study show that the Hardman index is not a reliable prognosticator of outcome following repair of rAAA. High-risk patients may still survive surgical repair, and the scoring system could not be recommended for routine use in clinical decision making. Further prospective evaluation of the risk factors is required to reliably exclude those patients in whom surgical intervention is inappropriate.
The authors thank Dr. Chris Patterson (Department of Medical Statistics, Queen's University Belfast) for help with the statistical analyses and all the consultant vascular surgeons at Belfast City Hospital and Royal Victoria Hospital for contributing cases which led to the successful completion of this project.
References
- . Ruptured abdominal aortic aneurysm: six year follow-up results of a multicentre prospective study. Canadian Society for Vascular Surgery Aneurysm Study Group. J Vasc Surg. 1994;19:888–900
- . Community mortality after ruptured abdominal aortic aneurysm is unrelated to the distance from the surgical centre. Br J Surg. 2001;88:1341–1343
- Resource utilization and outcomes: effects of transfer on patients with ruptured abdominal aortic aneurysms. Ann Vasc Surg. 2005;19:149–153
- . The ruptured abdominal aortic aneurysm in a community hospital. A 5-year study. Am Surg. 1992;58:499–501
- A meta-analysis of 50 years of ruptured abdominal aortic aneurysm repair. Br J Surg. 2002;89:714–730
- Ruptured abdominal aortic aneurysm: 15 years of continued experience in a southern West Virginia community. Ann Vasc Surg. 2005;19:851–857
- Factors increasing the mortality rate for patients with ruptured abdominal aortic aneurysms. Ann Vasc Surg. 2001;15:601–607
- Ruptured abdominal aortic aneurysm: factors influencing early survival. Ann Vasc Surg. 2005;19:29–34
- The post-operative mortality of ruptured abdominal aortic aneurysm repair. Eur J Vasc Endovasc Surg. 2004;27:65–74
- . Glasgow aneurysm score. Cardiovasc Surg. 1994;2:41–44
- Ruptured abdominal aortic aneurysm: who should be offered surgery?. J Vasc Surg. 1996;23:123–129
- Northern Ireland Statistics and Research Agency. Northern Ireland Census 2001 Population Report and Mid-Year Estimates. Belfast: Her Majesty's Stationery Office; 2002;p 7
- . Ruptured aortic aneurysm: the decision not to operate. Ann R Coll Surg Engl. 1998;80:221–225
- Ruptured abdominal aortic aneurysm: selecting patients for surgery. Eur J Vasc Endovasc Surg. 1999;17:129–132
- Comparison of POSSUM scoring and Hardman index in selection of patients for repair of ruptured abdominal aortic aneurysm. Br J Surg. 2003;90:421–425
- Predicting outcome in ruptured abdominal aortic aneurysm: a prospective study of 100 consecutive cases. Eur J Vasc Endovasc Surg. 2003;26:607–611
- Ruptured abdominal aortic aneurysm: is it possible to predict outcome?. Int Angiol. 2004;23:47–53
- Validity of the Glasgow Aneurysm Score and the Hardman index in predicting outcome after ruptured abdominal aortic aneurysm repair. Br J Surg. 2005;92:570–573
Presented to a meeting of the Australian and New Zealand Society for Vascular Surgery, Vascular 2005, Sydney, Australia, September 9-14, 2005.
PII: S0890-5096(06)00020-3
doi:10.1016/j.avsg.2006.08.002
© 2007 Annals of Vascular Surgery Inc. Published by Elsevier Inc All rights reserved.
