Racial Disparities in Endovascular Aortic Aneurysm Repair

Accepted Manuscript Racial Disparities in Endovascular Aortic Aneurysm Repair Adam Tanious, MD, MMSc, Nirmani Karunathilake, MS, Joel Toro, BA, Afif AbuHanna, BA, Laura T. Boitano, MD, Timothy Fawcett, PhD, Brian Graves, PhD, Peter Nelson, MD, MS PII:

S0890-5096(18)30844-6

DOI:

https://doi.org/10.1016/j.avsg.2018.11.002

Reference:

AVSG 4082

To appear in:

Annals of Vascular Surgery

Received Date: 9 April 2018 Revised Date:

25 October 2018

Accepted Date: 16 November 2018

Please cite this article as: Tanious A, Karunathilake N, Toro J, Abu-Hanna A, Boitano LT, Fawcett T, Graves B, Nelson P, Racial Disparities in Endovascular Aortic Aneurysm Repair, Annals of Vascular Surgery (2018), doi: https://doi.org/10.1016/j.avsg.2018.11.002. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

October 2018

RI PT

Prepared for: Annals of Vascular Surgery

SC

Racial Disparities in Endovascular Aortic Aneurysm Repair

M AN U

Adam Tanious MD, MMSc1; Nirmani Karunathilake MS2; Joel Toro BA2; Afif Abu-Hanna BA2; Laura T Boitano, MD1; Timothy Fawcett PhD2; Brian Graves PhD3; Peter Nelson MD, MS4

TE D

1 Massachusetts General Hospital, Department of Vascular and Endovascular Surgery 2 University of South Florida, Morsani College of Medicine 3 Tampa General Hospital, Department of Nursing 4 University of Oklahoma College of Medicine, Department of Surgery

EP

Presented at the Vascular And Endovascular Surgical Society 2018 Annual Meeting Vail, Colorado

AC C

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Corresponding Author: Adam Tanious Massachusetts General Hospital Department of Surgery Division of Vascular and Endovascular Surgery 15 Parkman Street | WAC 440 Boston, MA 02114 [email protected]

ACCEPTED MANUSCRIPT

Abstract

46

Objective: Racial and ethnic disparities are a critical issue in access to care within all fields of

47

medicine. We hypothesized that analysis of a statewide administrative dataset would demonstrate

48

disparities based on race with respect to access to this latest technology and the associated

49

outcomes following EVAR.

50

Methods: Utilizing de-identified data from the Florida State Agency for Health Care

51

Administration, we identified patients based on ICD-9 procedure codes who underwent EVAR

52

between the years 2000-2014. We then assigned these procedures with the specialty of the

53

operating physician and then analyzed outcomes based on the race of the patient.

54

Results: We identified 36,601 EVAR procedures during the study period. The average age of

55

the total sample was 73.38 (+/- 9.87), with the majority of the cohort being male (n = 29034,

56

81.2%). Breakdown of patients within each race category were as follows: 17,056 (47.7%) non-

57

Hispanic Whites, 1,630 (4.6%) non-Hispanic African Americans, 16,431 (46.0%) Hispanics, and

58

632 (1.8%) patients identified as “Other”. Data analysis showed significant differences between

59

age at presentation, sex of patient, and comorbidity score of patients at presentation. There were

60

significant differences in outcomes based on race with respect to total hospital charges, length of

61

stay, disposition, and payer status.

62

Conclusion: Racial disparities were discovered with respect to EVAR treatment. African

63

Americans present at younger ages, have the highest percentage of females requiring

64

intervention, have the longest hospital stays, have the highest Medicaid payer source, have the

65

highest in-hospital total charges of any racial group, and are more likely to be treated by

66

academic practitioners. Hispanics present with the highest comorbidity scores as compared to

AC C

EP

TE D

M AN U

SC

RI PT

45

ACCEPTED MANUSCRIPT

their counterparts and, along with African Americans, are more likely to be treated by non-

68

vascular surgeons.

AC C

EP

TE D

M AN U

SC

RI PT

67

ACCEPTED MANUSCRIPT

69 70

Introduction: The population of the United States is in constant flux. Not only are we growing as a nation, but the makeup of our population is radically different now than in any previous

72

generation.1 A decade prior, one-third of our population self-identified as a minority; in 3

73

decades-time, that number is on track to be one-half of our population.1 Compounding these

74

statistics, racial and ethnic disparities are a critical issue within all fields of medicine.2-4 With

75

twenty percent of American households being non-English speaking, the delivery of healthcare

76

will be affected from the level of individual provider to the insurance provider.5

SC

Medical research has begun addressing the question of how one’s race and ethnicity play

M AN U

77

RI PT

71

a role in one’s overall healthcare risk, disease prevalence, access to care, and procedural

79

outcomes. Over the past 2 decades we have learned that those of South-Asian and Afro-

80

Caribbean descent carry increased risk of hypertension and diabetes as compared to socio-

81

economically matched Caucasians.6 There has been a recent push to identify the epidemiology of

82

vascular disease patterns in non-Caucasian populations.5,7-9 Even with this push, there is still a

83

paucity of data regarding the incidence, treatment trends, and outcomes of non-Caucasian

84

populations with abdominal aortic aneurysms (AAA).

EP

85

TE D

78

As vascular and endovascular surgeons, we are privileged to be at the forefront of one of the major technology booms within medicine. Techniques and devices for endovascular aortic

87

aneurysm repair (EVAR) have improved significantly allowing the safe, minimally invasive

88

treatment of more complex anatomies by a wider array of practitioners.10-15 The question

89

becomes whether this new technology in the hands of a wider array of practitioners is being used

90

to treat a wider array of patients with equal outcomes. We hypothesized that analysis of a

AC C

86

ACCEPTED MANUSCRIPT

statewide administrative dataset would demonstrate disparities based on race with respect to

92

outcomes following EVAR.

AC C

EP

TE D

M AN U

SC

RI PT

91

ACCEPTED MANUSCRIPT

93 94

Methods The Agency for Health Care Administration (AHCA) maintains administrative databases of all discharges from all non-federal licensed acute care hospitals and free-standing procedure

96

centers in the state of Florida. As a multidisciplinary, outcomes-based research group, we

97

obtained access to de-identified data through the AHCA after obtaining Institutional Review

98

Board (IRB) approval.

Patients were identified between the years of 2000 to 2014, using the International

SC

99

RI PT

95

Classification of Diseases 9th Revision (ICD-9) procedure codes 39.71, 39.73, and 39.78 which

101

represent “endovascular aneurysm repair.” In patients undergoing EVAR during their index

102

hospitalization, we determined their identified race and stratified patients into the following

103

groups: non-Hispanic White, non-Hispanic African Americans, Hispanics, and other minorities.

104

In addition, the unique physician identification number of the operating physician was used to

105

determine the specialty of that physician (vascular surgeon, nonvascular surgeon) as well as their

106

academic practitioner status (yes or no) based on available state licensing information as well as

107

publically available web-based information. Age adjusted Charlson comorbidity scores were also

108

calculated to help stratify patients and results.

TE D

EP

109

M AN U

100

Outcome measures included being discharged to home (versus secondary institution), inhospital mortality, length of stay, and total charges. Total charges per year were adjusted for

111

inflation to the year 2000 dollar value. Cost data and length of stay data were positively skewed

112

to the right and age was negatively skewed to the left, thus violating the assumption of normal

113

distribution. To account for this the data was log-transformed, resulting in normal distributions

114

(by both histogram and normal probability plots), prior to mixed model analysis.

AC C

110

ACCEPTED MANUSCRIPT

Student's independent sample t-test and Kruskal-Wallis test was used to compare numerical

116

means while Pearson's Chi-square compared categorical variables. A generalized linear mixed

117

model was used to examine patient discharge status using demographic data and age adjusted

118

Charlson comorbidity score. Covariates were considered for inclusion in the model if p<.001 on

119

bivariate analysis and were removed from the model as significant predictors if p>.05. Overall

120

models were considered significant if the model significance was p<.001. Odds ratios were

121

reported with 95% confidence intervals. All statistical analyses were performed with SPSS 22 for

122

Windows (SPSS Inc., Chicago, IL).

AC C

EP

TE D

M AN U

123

SC

RI PT

115

ACCEPTED MANUSCRIPT

124 125

Results We identified 36,601 EVAR procedures performed by 1,786 practitioners during the study period. The average age of the total sample was 73.38 (± 9.87), with the majority of the

127

cohort being male (n = 29,034, 81.2%). Breakdown of patients within each race category were

128

as follows: 17,056 (47.7%) non-Hispanic Whites, 1,630 (4.6%) non-Hispanic African

129

Americans, 16,431 (46.0%) Hispanics, and 632 (1.8%) patients identified as “Other”. Table 1

130

displays the various demographic data stratified by race with associated significance values. Of

131

note, African Americans had a significantly higher female population of patients as well as the

132

highest Medicaid payer source as compared to any other race of patients. Hispanic patients

133

treated in this cohort had significantly higher comorbidity scores as compared to their

134

counterparts.

SC

M AN U

135

RI PT

126

We also found significant differences in the type of practitioner treating the various racial groups as well as the academic status of the practitioner. Table 2 details the specific percentages

137

of practitioner status by race category. African Americans and other minorities were more likely

138

to be treated by academic practitioners as compared to Hispanics and Whites. Vascular surgeons

139

performed the majority of EVARs for all groups studied, though they treated a higher percentage

140

of other minorities as compared to Hispanics and African Americans. Amongst these three

141

groups, the 8% increase in vascular surgeons treating other minorities was mirrored by the 7%

142

increase of cardiothoracic surgeons treating Hispanics and African Americans.

EP

AC C

143

TE D

136

There were significant differences in outcomes based on race with respect to in-hospital

144

mortality, disposition, length of stay, and total hospital charges. Table 3 shows the breakdown of

145

the above analysis with associated significance values. With regard to disposition to secondary

146

facility, the breakdown was as follows: Whites (8.5%), African Americans (19.4%), Hispanics

ACCEPTED MANUSCRIPT

(11.7%), and other minorities (12.7%). In addition to the highest rates of non-home discharge,

148

African Americans had significantly higher lengths of stay (8.79 days) and total hospital charges

149

($172,563). Other minorities suffered the highest mortality rates at 5.1% followed by African

150

Americans at 4%.

151

RI PT

147

A generalized linear mixed model was run as a part of this analysis and was run with both patient specific and operating physician specific data (P < .001 set as the significance

153

requirement for entry into model as well as for the overall model). These predictors included age,

154

age-adjusted Charlson comorbidity index score, sex, race, admission priority, payer source,

155

specialty of the operating practitioner (vascular surgeon, nonvascular surgeon), and academic

156

practitioner status (yes, no) data. Models were run for the outcome of in-hospital mortality and

157

non-home discharge. These models found that the following variables significantly increased the

158

odds of a patient death as well as non-home discharge: age-adjusted Charlson Comorbidity

159

Index, sex, age, and race. Tables 4 and 5 show these variables with their associated odds ratios

160

for each model. Models were also run for the outcomes of length of stay and total hospital

161

charges, though the overall models were not significant.

M AN U

TE D

EP AC C

162

SC

152

ACCEPTED MANUSCRIPT

163 164

Discussion Racial disparities are well documented within the vascular patient population, yet these differences have previously focused on the effects of peripheral arterial disease and outcomes of

166

intervention with regard to race.8,16,17 There is little data on the differences in incidence of aortic

167

aneurysmal disease with regard to race as well as treatment modalities and patient outcomes

168

when stratified by race. Our study took a statewide database and looked at all patients who had

169

undergone endovascular aortic aneurysm treatment over a 14-year period. We sought to

170

understand what demographic characteristics distinguish one racial group from another as well as

171

any patient outcomes that differ in patients being intervened upon using the same treatment

172

modality.

SC

M AN U

173

RI PT

165

It is interesting to first look at the demographic data that distinguishes one group from another. There have been previous reports of differences in patients with peripheral arterial

175

disease, particularly in the Hispanic and African American populations, suggesting higher

176

disease burden as well as morbidity/mortality rates.5,8,9,16,17 We found that while the majority of

177

patients treated were non-Hispanic white patients at 47.7%, they were closely followed by

178

Hispanics at 46%.

EP

TE D

174

Hispanic patients also presented with highest average comorbidity scores of their

180

counterparts. African Americans represent the youngest group of patients to be treated and have

181

the highest cohort of female patients. African Americans also have the highest Medicaid payer

182

status of the patients studied. While it is not possible to utilize this data as a surrogate for

183

socioeconomic factors, these differences do raise our attention to the differences in the

184

populations presenting for aneurysm care.

AC C

179

ACCEPTED MANUSCRIPT

Instead of merely looking at the inferior results experienced by non-white patients, we

186

must begin to ask why these disparities exist. Is it an issue of access to vascular surgeons or a

187

lack of community practitioners willing to accept certain types of insurance? 2,18 Furthermore,

188

we cannot avoid the issue of the dramatic increase of cost of care for all non-white races

189

undergoing EVAR. The data from tables 1-3 must be taken as an amalgam to institute change in

190

a number ways; be it preventative medicine with regard to patient education and access to care or

191

better medical management of the known comorbidities that predispose all patients to

192

aneurysms.2,18

SC

Again, while race did not affect patient outcome in our multivariate analysis, it is hard to

M AN U

193

RI PT

185

separate the distinguishing characteristics of the African American population from the notably

195

worse outcomes suffered by this group. African Americans were less likely to be discharged

196

home after surgery and had a higher rate of mortality as compared to Whites and Hispanics. As

197

a group they also had notably longer hospital stays and total hospital charges.

198

TE D

194

Armed with this knowledge, it behooves our treating physicians to understand that each patients needs are unique and that we cannot apply a one-size fits all model to patient care in any

200

area of vascular surgery. These results are not meant to have us look at patients differently as

201

they come into our offices and hospitals. Physician simply should strive to understand the needs

202

of their patients so that we can not simply perform a technically sound surgery, but help towards

203

a speedy hospital discharge to home.

AC C

204

EP

199

It is interesting to not only think about the current care provided to these patients, but turn

205

our attention towards the future care-providers of our nation. Medical schools have recognized

206

the lack of cultural competency amongst all forms of care-providers and have already instituted a

207

strict change to the curriculum regarding the education of medical students to address this very

ACCEPTED MANUSCRIPT

issue.19 Students are not only taught how to recognize and understand the differences in the

209

patients that may and will treat; they are also taught to understand what biases are inherent in the

210

delivery of medical care through to the level of the individual provider. This should be

211

considered as the medical communities response to this known issue of racial disparities and it’s

212

attempt at “preventative care.”

213

RI PT

208

This study has several limitations. The first is that while we did stratify outcomes based on the racial categorization of the patients treated, we did not and were not able to account for

215

socioeconomic factors that play a role in disease. While we did try to account and understand

216

these differences by looking at the payer source difference as well as the provider differences

217

between groups, due to the limited nature of the database we are unable to parcel out specific

218

socioeconomic factors that would affect the outcomes studied.8 Other limitations are those

219

inherent to the use of a state-wide database in that the racial categorizations are self-reported data

220

and that there is no follow-up data available once the patient has been discharged. Finally,

221

practitioner data was limited to information available from the online state licensing website or

222

public data from the physician’s hospital of employment.

M AN U

TE D

EP

224

AC C

223

SC

214

ACCEPTED MANUSCRIPT

225 226

Conclusion There are clear racial disparities within the vascular population. We discovered specific differences in patient outcomes with respect to EVAR treatment when stratifying by race.

228

African Americans present at younger ages, have the highest percentage of females requiring

229

intervention, have the longest hospital stays, have the highest Medicaid payer source, and have

230

the highest in-hospital total charges of any racial group. Hispanics do present with higher

231

comorbidity scores as compared to their counterparts.

AC C

EP

TE D

SC

M AN U

232

RI PT

227

ACCEPTED MANUSCRIPT

References:

234

1.

Bureau USC. Resources at www.census.gov. 2015.

235 236 237

2.

Peek ME, Wilson SC, Bussey-Jones J, Lypson M, Cordasco K, Jacobs EA, et al. A study of national physician organizations' efforts to reduce racial and ethnic health disparities in the United States. Acad Med. 2012 Jun;87(6):694–700.

238 239

3.

Beavers FP, Satiani B. Diversity does not equal disparity: how cultural competence can overcome. J Vasc Surg. Elsevier Inc; 2010 Apr;51(4 Suppl):1S–3S.

240 241

4.

Like RC. Educating clinicians about cultural competence and disparities in health and health care. J Contin Educ Health Prof. 2011;31(3):196–206.

242 243

5.

Kirksey L. Health care disparity in the care of the vascular patient. Vascular and Endovascular Surgery. 2011 Jul;45(5):418–21.

244 245 246

6.

Whitty CJM, Brunner EJ, Shipley MJ, Hemingway H, Marmot MG. Differences in biological risk factors for cardiovascular disease between three ethnic groups in the Whitehall II study. Atherosclerosis. Elsevier; 1999 Feb 1;142(2):279–86.

247 248

7.

Hobbs SD, Wilmink ABM, Bradbury AW. Ethnicity and peripheral arterial disease. European Journal of Vascular & Endovascular Surgery. 2003 Jun;25(6):505–12.

249 250

8.

Nguyen LL, Henry AJ. Disparities in vascular surgery: is it biology or environment? J Vasc Surg. Elsevier Inc; 2010 Apr;51(4 Suppl):36S–41S.

251 252

9.

Kwolek CJ, Clagett GP. Changing demographics in patients with vascular disease. J Vasc Surg. Elsevier Inc; 2009 Feb;49(2):528–31.

253 254 255 256

10.

IMPROVE Trial Investigators, Powell JT, Sweeting MJ, Thompson MM, Ashleigh R, Bell R, et al. Endovascular or open repair strategy for ruptured abdominal aortic aneurysm: 30 day outcomes from IMPROVE randomised trial. BMJ. 2014;348(jan13 2):f7661–1.

257 258 259

11.

Lobato AC, Camacho-Lobato L. Endovascular treatment of complex aortic aneurysms using the sandwich technique. J Endovasc Ther. SAGE Publications; 2012 Dec;19(6):691–706.

260 261 262

12.

263 264 265

13.

MD KA, MD AM, MD PG, MD VAK, MD PD, MD ADSI. Endovascular Stent-Graft Repair of Ascending Aortic Dissection With a Commercially Available Thoracic Endograft. ATS. Elsevier Inc; 2014 Aug 1;98(2):715–7.

266

14.

Veith FJ, Cayne NS, cayne, Berland TL, Mayer D, Lachat M. EVAR for Ruptured

AC C

EP

TE D

M AN U

SC

RI PT

233

Patel RP, Katsargyris A, Verhoeven ELG, Adam DJ, Hardman JA. Endovascular Aortic Aneurysm Repair with Chimney and Snorkel Grafts: Indications, Techniques and Results. Cardiovasc Intervent Radiol. Springer US; 2013;36(6):1443–51.

ACCEPTED MANUSCRIPT

267

Abdominal Aortic Aneurysms. Endovascular Today. 2011 Mar 23;:1–3. 15.

Greenhalgh RM, Brown LC, Kwong GPS, Powell JT, Thompson SG, EVAR trial participants. Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. Lancet. Elsevier; 2004 Sep;364(9437):843–8.

272 273 274

16.

Hughes K, Seetahal S, Oyetunji T, Rose D, Greene W, Chang D, et al. Racial/Ethnic Disparities in Amputation and Revascularization: A Nationwide Inpatient Sample Study. Vascular and Endovascular Surgery. 2013 Dec 17;48(1):34–7.

275 276 277

17.

Feinglass J, Abadin S, Thompson J, Pearce WH. A census-based analysis of racial disparities in lower extremity amputation rates in Northern Illinois, 1987-2004. J Vasc Surg. 2008 May;47(5):1001–7.

278 279 280

18.

Rosero EB, Kane K, Clagett GP, Timaran CH. A systematic review of the limitations and approaches to improve detection and management of peripheral arterial disease in Hispanics. J Vasc Surg. Elsevier Inc; 2010 Apr;51(4 Suppl):27S–35S.

281 282

19.

LCGME. FUNCTIONS AND STRUCTURE OF A MEDICAL SCHOOL. 2016 Jul pp. 1–39.

283

20. Bureau USC. Resources at https://www.census.gov/quickfacts/FL. 2017

M AN U

SC

RI PT

268 269 270 271

AC C

EP

TE D

284

ACCEPTED MANUSCRIPT

Table 1: Demographic Characteristics by Race Non-Hispanic Whites (n=17,056)

Non-Hispanic African Americans (n=1,630)

Hispanics

Other

(n=16,431)

(n = 632)

Average (+/- SD)

73.87 (8.99)

67.42 (13.59)

73.52 (10.05)

72.06 (11.30)

< 0.0001

(%) Male

83.9%

68.5%

79.8%

78.2%

< 0.0001

(%) Medicare (%) Medicaid Comorbidity severity score Mean (SD)

84.0% 0.8%

66.7% 8.1%

82.4% 2.3%

74.4% 2.7%

< 0.0001 < 0.0001

4.92 (1.55)

4.69 (2.07)

4.92 (1.86)

< 0.0001

Demographic Characteristics

p-Value

5.14 (1.84)

AC C

EP

TE D

M AN U

Payer

SC

Sex

RI PT

Age

ACCEPTED MANUSCRIPT

Table 2: Practitioner Data by Race Non-Hispanic Whites (n=17,056)

Non-Hispanic African Americans (n=1,630)

Hispanics

Other

(n=16,431)

(n = 632)

Yes

29.1%

41.4%

29.0%

42.2%

Vascular Surgeons Interventional Cardiologists Interventional Radiologists Cardiothoracic Surgeons Other Surgeons Other Practitioners

60.0% 2.8% 5.1% 19.5% 10.7% 1.8%

58.5% 3.2% 6.9% 22.0% 7.9% 1.5%

Practitioner Data

p-Value

Academic Practitioner Status

AC C

EP

TE D

RI PT 58.0% 5.3% 6.5% 22.7% 6.5% 0.9%

SC

M AN U

Practitioner Type (%)

<0.0001

66.6% 4.0% 7.3% 15.3% 6.0% 0.8%

< 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001

ACCEPTED MANUSCRIPT

Table 3: Outcome Variables by Race Non-Hispanic Whites (n=17,056)

Non-Hispanic African Americans (n=1,630)

Hispanics

Other

(n=16,431)

(n = 632)

In hospital Mortality In-Hospital Death (%)

2.5%

4.0%

2.8%

5.1%

Disposition (%) Discharged to home Transfer to Other Facility

89.0% 8.5%

76.6% 19.4%

4.38 (7.18) 2

8.79 (11.41) 5

$97,169 (77,685)

$172,563 (140,085)

Mean (SD) Median

AC C

EP

TE D

Mean (SD)

85.5% 11.7%

82.2% 12.7%

<0.0001 <0.0001

4.73 (11.43) 2

6.42 (10.42) 3

<0.0001

$151,949 (115,977)

$148,954 (128,283)

<0.0001

M AN U

Total hospital charges ($)

<0.0001

RI PT

Length of hospital stay (Days)

p-Value

SC

Outcome Variable

ACCEPTED MANUSCRIPT

Table 4: Significant Predictors of Patient In-Hospital Mortality Significant Variables Associated with In-Hospital Mortality

AC C

EP

TE D

M AN U

SC

*All Odds Ratios listed were significant with p-Value < 0.05

1.74 1.61 1.52 1.96

RI PT

Sex (female patients) Age (greater than 80 years of age at surgery) African Americans (Compared to Non-Hispanic White) Other Minorities (Compared to Non-Hispanic White)

Odds Ratio*

ACCEPTED MANUSCRIPT

Table 5: Significant Predictors of Non-Home Discharge Odds Ratio*

Sex (female patients) Age (greater than 80 years of age at surgery) African American Race (Compared to Non-Hispanic White American Race) Charlson Comorbidity Index Score greater than or equal to 5

1.92 1.68 1.64 1.38

RI PT

Significant Variables Associated with Non-Home Discharge

AC C

EP

TE D

M AN U

SC

*All Odds Ratios listed were significant with p-Value < 0.05