Therapy of Renal Artery Aneurysms in New York State: Outcomes of Patients Undergoing Open and Endovascular Repair

Article Outline

• Abstract
• Introduction
• Methods
• Results
• Discussion
• Conclusions
• References
• Copyright

The purpose of this study was to evaluate changing trends in therapy and determinants of outcomes among patients with a renal artery aneurysm (RAA) undergoing surgical or endovascular repair in New York State (NYS). A retrospective cohort study of patients who underwent therapy for RAA in NYS from October 1, 2000, to December 31, 2006, was identified from the Statewide Planning and Research Cooperative System database. Regression models which included hospital and patient characteristics were created to identify predictors of untoward events following surgical or endovascular intervention. Over this time period 215 patients with RAA repairs were analyzed. In multivariate analysis, preoperative predictors of death included diabetes (adjusted odds ratio [OR] = 57.8, 95% confidence interval [CI] 2.3-1,430.1, p = 0.013), the presence of other aneurysms (adjusted OR = 18.5, CI 1.5-234.4, p = 0.024), and coagulopathy (adjusted OR = 16.9, CI 3.4-393.1, p = 0.03) but not repair type. Perioperative cardiac (adjusted OR = 16.7, CI 1.4-197.1, p = 0.026) and vascular device-related (adjusted OR = 11.1, CI 1.003-123.0, p = 0.049) complications were predictive of mortality. When patients with other aneurysms were excluded from analysis (n = 153), there were no significant predictors of death. Ninety-one endovascular and 124 open surgical repairs were performed with a significant increase in the proportion of endovascular repairs performed over time (p < 0.001), although since 2003 the proportion of both has been roughly equal. Diabetes (15.4% vs. 5.6%, p = 0.018), chronic anemia (5.5% vs. 0.8%, p = 0.04), and emergent admission (48.4% vs. 24.2%, p < 0.001) were more prevalent among those with endovascular repair. Endovascular therapy was associated with a lower incidence of complications, lower median length of stay (4 vs. 7 days, p < 0.001), and lower rates of discharge to skilled nursing facilities (18.9% vs. 39.2%, p = 0.001). There has been an increasing number of treated RAAs in NYS since 2000, with the increase being primarily in those treated by endovascular techniques. Whether this represents a true increase in RAA incidence requiring management or an extension of indications is unknown. Outcomes after endovascular repair were better than those after conventional surgery, although whether this was due to the technique of repair itself or preprocedural selection bias cannot be determined.

Back to Article Outline

Introduction 

Renal artery aneurysms (RAAs) are rare, the spectrum of available and reported therapy is wide, and consensus on appropriate indications for management is lacking. Untreated RAAs may be associated with hypertension, dissection, renal failure, and rupture, although the true incidence of these events is unknown.¹, ², ³, ⁴ Generally accepted indications for intervention include size >2.0 cm, renovascular hypertension, dissection, localized symptoms such as flank pain and hematuria, distal embolization, and female gender within childbearing age.¹, ², ³, ⁴, ⁵, ⁶, ⁷, ⁸, ⁹, ¹⁰ Therapeutic options include observation, endovascular therapy, and open repair. Surgical therapy of RAAs in properly selected patients has been shown to provide excellent long-term clinical outcomes.¹¹

The continued improvement in endovascular techniques and tools over the last two decades has created new options for the management of RAAs. Most data regarding endovascular intervention are derived, however, from case reports; and reliable evidence of the results of endovascular therapy is lacking.¹², ¹³, ¹⁴ We conducted a retrospective cohort study utilizing a statewide discharge database to examine a larger group of patients undergoing endovascular treatment of RAAs to better define current practice and in-hospital outcomes of intervention.

Back to Article Outline

Methods 

New York State (NYS) maintains statewide, patient-level discharge data from all nonfederal health-care institutions through the NYS Department of Health. This unit, known as the Statewide Planning and Research Cooperative System (SPARCS), is legislatively mandated to collect discharge data, which include patient and center characteristics, diagnoses and treatments, services, and charges, with mechanisms in place to allow for data review and correction of errors. Approval was obtained from our institutional research subjects review board for the conduct of this study.

We conducted a retrospective cohort study of patients within this database from October 1, 2000, through December 31, 2006. As data are only available for the current admission, outcome events are in-hospital events only. Patients with RAAs were identified using ICD-9 code 442.1 (aneurysm of renal artery). This is the only code for RAA and implicitly excludes renal artery stenosis, fibromuscular dysplasia, renovascular hypertension, and so on. We limited our investigation to patients undergoing intervention only and, thus, also required the ICD-9 code of either 39.79 (other endovascular repair of aneurysm of other vessels) or a surgical code (below). ICD-9 procedure code 39.79 was introduced on October 1, 2000, and does not specify the exact type of endovascular repair that the patient received; it is used after coil embolization or occlusion, endovascular graft placement, liquid tissue adhesive (glue) embolization or occlusion, and other implant or substance for repair, embolization, or occlusion. Surgical codes included 38.46 (resection of vessel with replacement, abdominal arteries), 39.52 (other repair of aneurysm), 39.24 (aortorenal bypass), 55.51 (nephroureterectomy), 38.36 (resection of vessel with anastomosis, abdominal arteries), 38.66 (other excision of vessel, abdominal arteries), 55.61 (renal autotransplantation), and 39.55 (reimplantation of aberrant renal vessel).

Complications were also identified by ICD-9 codes: 997.1 (cardiac surgical complication), 997.3 (respiratory complications of surgical care), 998.11 (hemorrhage complicating a procedure), 997.5 (urinary complications, renal failure/insufficiency due to a procedure), 997.72 (vascular complications of renal artery), 998.5 (postoperative infection), 996.62 (infectious/inflammatory reaction due to vascular device), 996.74 (complication due to vascular device), and 996.1 (malfunction of vascular device/graft). Vascular device complications excluding central line complications are specific for the implanted device or graft.

The Elixhauser comorbidity index, an index that identifies and measures the effects of 30 different comorbid conditions, was used to evaluate preoperative comorbid conditions. This scale separates existing problems from the primary reason for hospitalization based upon diagnosis-related group codes and allows better evaluation of comorbidities derived from an administrative database.¹⁵

The SPARCS database provides charge information (the amount that hospitals billed for services). We used the Healthcare Cost and Utilization Project (HCUP) cost to charge files to convert hospital charges into actual cost.¹⁶

Data management and statistical analysis were performed using SPSS version 15 (SPSS, Inc., Chicago, IL). Chi-squared analysis was used to test bivariate relationships between risk factors and perioperative complications and to determine subsequent entry into multivariate logistic regression models. Adjusted odds ratios (ORs) were calculated using logistic regression; factors were included if they were significant on chi-squared testing at the p < 0.10 level. Nonparametric tests were used where appropriate. Adjusted ORs were reported as 95% confidence intervals (95% CI). Our hypothesis was that endovascular intervention results in lower mortality and morbidity, so results were considered significant at a one-tailed p value of <0.05.

Back to Article Outline

Results 

From 2000 to 2006, 222 patients underwent intervention for RAAs based on these ICD-9 codes. Seven patients presented with a ruptured abdominal aortic aneurysm and were excluded from analysis, leaving 215 patients analyzed in this study. Ninety-one patients underwent endovascular (ENDO) and 124 underwent open surgical (OPEN) repair. Of all repairs, 52.6% were performed at institutions that have a vascular fellowship and 70.2% were performed at institutions that are associated with a general surgery residency program. The percentage of repairs performed endovascularly was 43.7% at teaching institutions vs. 39.1% at nonteaching institutions (p = 0.529) and 44.2% at institutions with a vascular fellowship vs. 40.2% at institutions without a vascular fellowship (p = 0.549). During this period the total number of repairs increased significantly (p < 0.001) while the number of OPEN repairs remained stable (p = 0.10) and the number of ENDO repairs increased (p < 0.001) (Fig. 1). There was also a significant increase from 2000 to 2006 in the proportion of ENDO to OPEN repairs (p < 0.001); however, since 2003 the proportion of ENDO to OPEN repairs has remained roughly equal.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  Incidence of RAAs undergoing repair in NYS between 2000 and 2006. Solid bars are those undergoing endovascular repair (ENDO) and diagonal lines, those undergoing open (OPEN) repair.

Outcome events are in-hospital events only. Repair type was not predictive of mortality (ENDO 1.1% vs. OPEN 3.2%, p = 0.308). Preoperative predictors of death included diabetes (adjusted OR = 57.8, p = 0.013), the presence of other aneurysms (adjusted OR = 18.5, p = 0.024), and coagulopathy (adjusted OR = 36.3, p = 0.003) but not repair type. Perioperative cardiac (adjusted OR = 16.7, p = 0.026) and vascular device-related (adjusted OR = 11.1, p = 0.049) complications were predictive of mortality (Table I). Vascular device complications included thrombosis or hemorrhage of the vascular graft (four patients), infection of the vascular graft (two patients), and malfunction of the vascular graft (one patient).

Table I. Multivariate analysis

Variable	Adjusted OR	95% CI	p
Mortality
Diabetes	57.8	2.3-1,430.1	0.013
Other aneurysms	18.5	1.5-234.4	0.024
Coagulopathy	16.9	3.4-393.1	0.03
Complication mortality
Cardiac	16.7	1.4-197.1	0.026
Device-related	11.1	1.003-123.0	0.049
Need for postdischarge nursing care
OPEN	2.7	1.4-5.3	0.003
Age	1.04	1.01-1.06	0.002
Complication postdischarge nursing
Cardiac	N/A	N/A	N/A
Device	5.1	0.9-28.9	0.063

Predictors of postdischarge nursing care by multivariate analysis. Positive ORs indicate that the stated condition (preoperative comorbidity or perioperative complication) was positively correlated with the complication in question.

Factors that predicted the need for discharge to a skilled nursing facility or home with nursing services were OPEN repair (adjusted OR = 2.7, p = 0.003) and increasing age (adjusted OR = 1.04, p = 0.002) (Table I).

Patients undergoing ENDO repair were more likely to be diabetic (15.4% vs. 5.6%, p = 0.018), to be chronically anemic (5.5% vs. 0.8%, p = 0.04), to have neurological disorders (5.5% vs. 0.8%, p = 0.04), and to have liver disease (0% vs. 3.3%, p = 0.042). Emergent admission occurred more frequently in the ENDO cohort (48.4% vs. 24.2%, p < 0.001). The overall presence of additional aneurysms was similar between the two groups, although trends toward a greater incidence in the OPEN group existed in most cases. Patients with hematuria were more likely to receive an ENDO repair (16.5% of all ENDO repairs vs. 1.6% of all OPEN repairs, p < 0.001), although no other indication was correlated with repair technique. Patients with ENDO repair had a lower median length of stay (4 vs. 7 days, p < 0.001) and a trend toward lower cost ($13,786.18 vs. $15,406.03 p = 0.12). Postdischarge nursing services were required after 18.9% of ENDO and 39.2% of OPEN repairs (p = 0.001) (Table II).

Table II. Demographics OPEN vs. ENDO

Variable	OPEN (n = 124)	ENDO (n = 91)	p
Median age (years)	65 ± 14.67	62 ± 17.02	0.214
Median cost (dollars)	15,406.03 ± 64,543.12	13,786.18 ± 29,504.52	0.120
Median length of stay (days)	7 ± 19	4 ± 9	<0.001
Emergent admission	24.2	48.4	<0.001
Mortality	3.2	1.1	0.308
Postdischarge nursing care (n = 210)	39.2	18.9	0.001
Race			0.793
Black	10.5	11.0
White	73.4	74.7
Other	16.1	14.3
Sex			0.398
Male	52.4	58.2
Female	47.6	41.8
Other aneurysms
Abdominal	25.0	17.6	0.195
Iliac	10.5	9.9	0.887
Thoracoabdominal	2.4	1.1	0.480
Thoracic	0	1.1	0.243
Lower extremity	0	1.1	0.243
Splenic	0.8	2.2	0.391
Visceral	1.6	1.1	0.751
All other aneurysms	32.3	24.2	0.197
Repair of additional aneurysms	19.4	17.6	0.742
Indications for repair
Hematuria	1.6	16.5	<0.001
Renovascular hypertension	1.6	1.1	0.751
Female, age ≤50	10.5	13.2	0.542
Comorbidity
Diabetes	5.6	15.4	0.018
Hypertension	52.4	46.2	0.365
Peripheral vascular disease	56.5	56.0	0.953
Congestive heart failure	6.5	5.5	0.772
Chronic pulmonary disease	16.9	16.5	0.930
Renal failure	2.4	0	0.136
Coagulopathy	5.6	3.3	0.420
Neuro disorders	0.8	5.5	0.040
Hypothyroid	6.5	7.7	0.725
Fluid/electrolyte abnormalities	12.9	13.2	0.951
Chronic blood loss anemia	0.8	5.5	0.040
Deficiency anemia	11.3	6.6	0.243
Liver disease	0	3.3	0.042
Obesity	3.2	2.2	0.652

Overall demographics and outcomes, split by repair technique (OPEN vs. ENDO, as defined by ICD-9 codes). Numbers given are percentages unless otherwise specified.

Patients with an OPEN repair were more likely to have postoperative vascular device (5.6% vs. 0%, p = 0.022) complications and a trend toward more cardiac (4.0% vs. 0%, p = 0.053) and infectious (4.0% vs. 0%, p = 0.053) complications but fewer hemorrhagic complications (3.2% vs. 8.8%, p = 0.08) (Table III). All of these observations either remained significant or continued to show a trend toward significance after exclusion of patients with concomitant aneurysmal disease (Table IV, Table V, Table VI).

Table III. Major postoperative complications OPEN vs. ENDO

Overall complications, split by repair technique (OPEN vs. ENDO, as defined by ICD-9 codes).

Table IV. Multivariate analysis with concomitant aneurysmal disease excluded

Variable	Adjusted OR	95% CI	p
Postdischarge nursing
Age	1.04	1.017-1.061	<0.001
Open repair	3.1	1.636-5.872	0.001
Complication postdischarge nursing
Cardiac	N/A	N/A	N/A
Device	3.3	0.516-20.569	0.209

Predictors of postdischarge nursing care by multivariate analysis. Positive ORs indicate that the stated condition (preoperative comorbidity or perioperative complication) was positively correlated with the complication in question.

Table V. Demographics OPEN vs. ENDO with concomitant aneurysmal disease excluded

Variable	Open (n = 84)	Endovascular (n = 69)	p
Median age (years)	59 ± 14.66	56 ± 17.83	0.566
Median cost (dollars)	15,406.03 ± 59,159.42	12,807.03 ± 25,317.70	0.257
Median length of stay (days)	6 ± 18	4 ± 8	<0.001
Emergent admission	19.0	56.5	<0.001
Mortality	0	1.4	0.270
Postdischarge nursing care	34.5	15.9	0.009
Race
Black	11.9	13.0
White	69.0	72.5
Other	19.0	14.5
Sex			0.580
Male	51.2	56.5
Female	48.8	43.5
			0.344
Indications for repair
Hematuria	2.4	20.3	<0.001
Renovascular hypertension	2.4	1.4	0.680
Female, age ≤50	14.3	15.9	0.776
Comorbidity
Diabetes	7.1	15.9	0.086
Hypertension	48.8	46.4	0.765
Peripheral vascular disease	35.7	42.0	0.426
Congestive heart failure	7.1	7.2	0.980
Chronic pulmonary disease	17.9	10.1	0.178
Renal failure	1.2	0	0.365
Coagulopathy	4.8	4.3	0.903
Neuro disorders	1.2	7.2	0.056
Hypothyroid	6.0	7.2	0.748
Fluid/electrolyte abnormalities	10.7	13.0	0.657
Chronic blood loss anemia	0	7.2	0.012
Deficiency anemia	11.9	4.3	0.096
Liver disease	0	4.3	0.054
Obesity	4.8	1.4	0.253

Overall demographics and outcomes, split by repair technique (OPEN vs. ENDO, as defined by ICD-9 codes). Numbers given are percentages unless otherwise specified.

Table VI. Major postoperative complications OPEN vs. ENDO with concomitant aneurysmal disease excluded

Overall complications, split by repair technique (OPEN vs. ENDO, as defined by ICD-9 codes).

We found that 19.4% of OPEN and 17.6% of ENDO (p = 0.742) patients had concomitant repair of other aneurysms and 32.3% of OPEN and 24.2% of ENDO (p = 0.197) patients had other aneurysms. When these patients were excluded, there remained a significant difference in the median length of stay (6 vs. 4 days, p < 0.001) and the need for further nursing care after discharge (34.5% vs. 15.9%, p = 0.009) despite ENDO patients being more likely to be admitted emergently (19.0% vs. 56.5%, p < 0.001). In multivariate analysis of patients who had no other aneurysms, there were no significant predictors of mortality, likely as a result of the exclusion of all patients who died except one. Predictors of the need for nursing care after discharge remained OPEN repair (adjusted OR = 3.1, p = 0.001) and increasing age (adjusted OR = 1.04, p < 0.001).

Back to Article Outline

Discussion 

RAAs are uncommon, and therapeutic options for repair are diverse. The incidence of RAAs has been described as between 0.1% (autopsy) and 1.3% (renal angiography).¹, ², ³, ⁴ Perhaps due to this low incidence there is little consensus as to therapeutic management, and recommendations for intervention are not supported by strong evidence. While most would intervene for rupture, dissection, size >2.0 cm, renovascular hypertension, localizing symptoms such as flank pain and hematuria, distal embolization, and an aneurysm in a female of childbearing age, the true natural history of asymptomatic aneurysms is unclear and results of treatment are poorly characterized.¹, ², ³, ⁴, ⁵, ⁶, ⁷, ⁸, ⁹, ¹⁰

We found an increase in the total number of RAA repairs performed in NYS from 2000 to 2006. The number of open repairs remained relatively stable, but there was a steady and significant increase in the number of endovascular repairs performed. It is likely that increasing use of diagnostic imaging and public awareness of aneurysmal disease is contributing to this finding, but the obvious question is whether this increase is the result of more aneurysms being diagnosed (concomitant with increasing use of endovascular techniques in general) or of an expansion of indications for repair. It is certainly reasonable to assume that higher-risk patients who would have been poor candidates for open repair are now considered candidates for endovascular therapy. It is clear from these data that patients in NYS undergoing endovascular repair have better outcomes than those undergoing open surgery despite having greater preoperative major comorbidities of diabetes and liver disease. While causality cannot be determined in retrospect, our data are certainly consistent with the hypothesis that the lower risk of endovascular intervention is broadening indications for intervention and increasing numbers (not just percentages). The observation that approximately equal numbers of aneurysms have been repaired using each technique since 2003 raises the interesting conjecture that this is the new and correct set point identified unconsciously by the clinical marketplace.

We found that cardiac and vascular device complications were correlated with mortality, but whether these were the cause or a marker of problems cannot be determined. Preoperative factors that were associated with eventual mortality included concomitant nonrenal aneurysmal disease, diabetes, and coagulopathy. Given the association of age and comorbidity, we can expect a significant proportion of patients undergoing vascular therapy to need postdischarge nursing care. Patients undergoing open repair and those with increased age have a greater chance of needing such care, and it would seem evident to plan ahead if these factors are present. Finally, a trend toward lower overall cost for endovascular repair was evident, persisting even when patients with nonrenal aneurysms were excluded. This is a little counterintuitive as costs for many endovascular procedures (e.g., aortic aneurysm repair) have been shown to be higher than those for the equivalent open procedure. This may be explained by the lower device cost in these small vessels, but our data cannot answer this question.

There are inherent limitations to large coding databases such as SPARCS. As data are only available for the current admission, outcome events are in-hospital events only. Coding data are not perfect, and there are well-recognized limitations.¹⁷ However, the limitations of these types of data have been well-described and vetted and, in general, have been found to provide reliable information when compared to data derived from direct chart review.¹⁸ Primary diagnosis and procedure codes tend to be accurate as coders are responsible for documenting the primary reasons for admission to obtain reimbursement for the institution.¹⁹ Previous studies have shown, however, that while ICD-9 codes have a high specificity, they may have low sensitivity for existing comorbidities, resulting in underestimation of preexisting diseases.²⁰, ²¹ SPARCS also does not contain data that quantify severity of illness. For example, we have no information regarding size, complexity, location, or rupture status of the RAA. There are also no codes for the specific endovascular procedure performed. Analyses of databases such as this remain more useful than single-center analyses, however, for exploring events that are infrequent and poorly characterized, by providing an efficient mechanism of examining larger numbers of patients.

Back to Article Outline

Conclusions 

RAAs are rare. Although this analysis suffers from inherent limitations, it provides helpful information regarding outcomes after intervention for this problem. The number of RAAs undergoing treatment in NYS has risen over the past 7 years, and a trend toward increased endovascular management was seen. Whether this represents a true increase in incidence, better identification, or an extension of indications is unknown. Outcomes after endovascular repair were better than those after conventional surgery, although whether this was due to the technique of repair itself or preprocedural selection bias cannot be determined.

Back to Article Outline

References 

1. Hageman JH, Smith RF, Szilagyi DE, et al. Aneurysms of the renal artery: problems of prognosis and surgical management. Surgery. 1978;84:563–571
   • View In Article
   • MEDLINE
2. Tham G, Ekelund L, Herrlin K, et al. Renal artery aneurysm. Natural history and prognosis. Ann Surg. 1983;197:348–352
   • View In Article
   • MEDLINE
   • CrossRef
3. Eskandari MK, Resnick SA. Aneurysms of the renal artery. Semin Vasc Surg. 2005;18:202–208
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (532 KB)
   • CrossRef
4. Calligaro KD, Dougherty MJ. Renal artery aneurysms and arteriovenous fistulae. In:  Rutherford RB editors. Vascular Surgery. 6th ed.. Philadelphia: WB Saunders; 2005;p. 1861–1870
   • View In Article
5. Poutasse EF. Renal artery aneurysms. J Urol. 1975;43:113
   • View In Article
6. Fry WF. Renal artery aneurysm. In:  Ernst CB,  Stanley JC editor. Current Therapy in Vascular Surgery. Philadelphia: BC Decker; 1987;p. 363
   • View In Article
7. Bastounis E, Pikoulis E, Georgopoulos S, et al. Surgery for renal artery aneurysms: a combined series of two large centers. Eur Urol. 1998;33:22–27
   • View In Article
   • MEDLINE
   • CrossRef
8. Martin RS, Meacham PW, Ditesheim JA, et al. Renal artery aneurysm: selective treatment for hypertension and prevention of rupture. J Vasc Surg. 1989;9:26–34
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (1129 KB)
   • CrossRef
9. Cohen JR, Shamash FS. Ruptured renal artery aneurysms during pregnancy. J Vasc Surg. 1987;6:51–59
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (629 KB)
   • CrossRef
10. Henricksson C, Bjorkerund S, Nilson AE, et al. Natural history of renal artery aneurysm elucidated by repeated angiography and pathoanatomical studies. Eur Urol. 1985;11:244–248
    • View In Article
    • MEDLINE
11. Henke PK, Cardneau JD, Stanley JC, et al. Renal artery aneurysms: a 35-year clinical experience with 252 aneurysms in 168 patients. Ann Surg. 2001;234:454–463
    • View In Article
    • MEDLINE
    • CrossRef
12. Bui BT, Oliva VL, Leclerc G, et al. Renal artery aneurysm: treatment with percutaneous placement of a stent-graft. Radiology. 1995;195:181–182
    • View In Article
    • MEDLINE
13. Klein GE, Szolar DH, Breinl E, et al. Endovascular treatment of renal artery aneurysms with conventional non-detachable microcoils and Guglielmi detachable coils. Br J Urol. 1997;79:852–860
    • View In Article
    • MEDLINE
14. Rundback JH, Rizvi A, Rozenblit GN, et al. Percutaneous stent-graft management of renal artery aneurysms. J Vasc Intervent Radiol. 2000;11:1189–1193
    • View In Article
15. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27
    • View In Article
    • MEDLINE
    • CrossRef
16. Healthcare Cost and Utilization Project. http://www.ahrq.gov/data/hcup.
    • View In Article
17. Fry DE, Pine MB, Jordan HS, et al. The hazards of using administrative data to measure surgical quality. Gen Surg News. Nov. 2006;57–63
    • View In Article
18. Humphries KH, Rankin JM, Carere RG, et al. Co-morbidity data in outcomes research: are clinical data derived from administrative databases a reliable alternative to chart review?. J Clin Epidemiol. 2000;53:343–349
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (59 KB)
    • CrossRef
19. Berman MF, Stapf C, Sciacca RR, et al. Use of ICD-9 coding for estimating the occurrence of cerebrovascular malformations. A J N R Am J Neuroradiol. 2002;23:700–705
    • View In Article
20. Powell H, Lim LL, Heller RF. Accuracy of administrative data to assess comorbidity in patients with heart disease. An Australian perspective. J Clin Epidemiol. 2001;54:687–693
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (61 KB)
    • CrossRef
21. Anderson PL, Gelijns A, Moskowitz A, et al. Understanding trends in inpatient surgical volume: vascular interventions 1980-2000. J Vasc Surg. 2004;39:1200–1208
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (219 KB)
    • CrossRef