- Email: [email protected]
Predicting In-Hospital Complications After Anterior Cervical Discectomy and Fusion: A Comparison of the Elixhauser and Charlson Comorbidity Indices
Original Article
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 45 46 47 48 49 50 51 52 53 54 55 56 57 58
Predicting In-Hospital Complications After Anterior Cervical Discectomy and Fusion: A Comparison of the Elixhauser and Charlson Comorbidity Indices Q10 Q11 Q1Q9 Q8
William A. Ranson1, Sean N. Neifert2, Zoe B. Cheung1, John M. Caridi2, Samuel K. Cho1
OBJECTIVE: The objective of this study was to determine the ability of the Elixhauser Comorbidity Index (ECI) and Charlson Comorbidity Index (CCI) to predict postoperative complications after anterior cervical discectomy and fusion (ACDF).
-
METHODS: This was a retrospective study of ACDF hospitalizations in the National Inpatient Sample from 2013 to 2014. The ECI and CCI were calculated, and patients who experienced postoperative complications were identified. The ability of these indexes to predict complications was compared using the c statistic (area under the receiver operating characteristic curve [AUC]). In addition, the CCI and ECI were compared with a base model that included age, sex, race, and primary payer.
-
RESULTS: A total of 261,780 patients were included. Patients who experienced a complication were more often male (P < 0.0001) and older (P < 0.0001). They also had a higher comorbidity burden as assessed by both the ECI (P < 0.0001) and the CCI (P < 0.0001). The ECI was superior in predicting airway complications (AUC, 0.81 vs. 0.75; P < 0.0001), hemorrhagic anemia (AUC, 0.67 vs. 0.63; P [ 0.0015), pulmonary embolism (AUC, 0.91 vs. 0.77; P < 0.0001), wound dehiscence (AUC, 0.80 vs. 0.55; P [ 0.0080), sepsis (AUC, 0.87 vs. 0.82; P [ 0.0001), and septic shock (AUC, 0.94 vs. 0.83; P < 0.0001). The CCI was not found to be superior to
-
Key words ACDF - Anterior cervical discectomy and fusion - Charlson - Comorbidity - Complication - Elixhauser -
Abbreviations and Acronyms ACDF: Anterior cervical discectomy and fusion AKI: Acute kidney injury AUC: Area under the receiver operating characteristic curve CCI: Charlson Comorbidity Index CMS: Centers for Medicare and Medicaid CVA: Cerebrovascular accident DRG: Diagnosis-related group DVT: Deep vein thrombosis HCUP: Healthcare Cost and Utilization Project
WORLD NEUROSURGERY -: e1-e10, - 2019
the ECI for predicting any complications. Both were excellent for predicting mortality (ECI AUC, 0.87; CCI AUC, 0.90). CONCLUSIONS: The ECI was superior to the CCI in predicting 6 of 15 complications in this study. Both are excellent tools for predicting mortality after ACDF.
-
INTRODUCTION
A
nterior cervical discectomy and fusion (ACDF) is a widely used and effective surgical procedure for the treatment of degenerative disease affecting the cervical spine. The most common indications for ACDF include cervical spondylosis and disc herniation refractory to conservative management, as well as certain subtypes of cervical fractures, neoplastic lesions, and infectious processes.1,2 Because cervical spine degenerative disease is a chronic condition that develops over many years, most patients undergoing ACDF are generally older, with more than half of these surgical patients being older than 50 years and a quarter older than 60 years.3,4 A well-recognized challenge to surgeons treating an older patient population is the presence of a relatively greater comorbidity burden compared with younger patients. For all surgical candidates, medical comorbidities are an important consideration because they have been shown to significantly affect the rate of various postoperative complications
ICD-9: International Classification of Diseases, Ninth Revision MI: Myocardial infarction NIS: National Inpatient Sample PE: Pulmonary embolism SSI: Surgical site infection UTI: Urinary tract infection From the Departments of 1Orthopaedics, and 2Neurosurgery, Mount Sinai Hospital, New York, New York, USA To whom correspondence should be addressed: Samuel Kang-Wook Cho, M.D. [E-mail: [email protected]]
Q12
Citation: World Neurosurg. (2019). https://doi.org/10.1016/j.wneu.2019.10.102 Journal homepage: www.journals.elsevier.com/world-neurosurgery Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e1
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
ORIGINAL ARTICLE ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES Q2
WILLIAM A. RANSON ET AL.
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
and also to play a critical role in provider performance assessment, reimbursement, and proposed bundled payment plans.5-14 To better assess surgical patients who may present with any combination of the multitude of possible comorbidities, various scoring measures have been proposed, with 2 specific indexes, the Charlson Comorbidity Index (CCI) and the Elixhauser Comorbidity Index, having attained prominence in orthopedic surgery.15,16 The CCI, originally developed in 1987 to predict 1-year oncologic mortality and subsequently adapted for use with administrative databases in 1992, has become a widely used risk adjustment method for patients undergoing spine surgery.15,17 The CCI consists of 19 medical conditions and has been shown to be a excellent predictor of inpatient mortality, postdischarge all-cause mortality, and health care expenditures in both medical and surgical patient populations.17-21 The Elixhauser Comorbidity Index is a newer risk adjustment method encompassing 31 medical comorbidities; its popularity has grown to rival the CCI, with recent literature suggesting a superior performance in predicting inpatient mortality after gastrointestinal, cardiac, and major orthopedic surgeries.19,22-27 Multiple comorbidities shown to affect perioperative outcomes in the spine surgery population in particular, including hypertension, obesity, coagulopathy, anemia, alcohol use disorder, drug use disorder, and depression, are present in the Elixhauser Index but not in the CCI.28-33 Although the literature has established that both the CCI and the Elixhauser Index are excellent predictors of mortality after orthopedic surgeries in aggregate, little investigation has been carried out into their value in predicting inpatient perioperative complications after spine surgery, with no studies specifically investigating ACDF.22 Using nationally representative data, this study sought to assess and compare the performance of the CCI and Elixhauser Comorbidity Index in predicting mortality and perioperative complications during the inpatient period in patients undergoing ACDF.
Table 1. Summary Demographic Statistics for National Inpatient Sample Patients Undergoing Anterior Cervical Discectomy and Fusion Who Experienced No Complications versus 1 Complication (N ¼ 261,780) No Complication Category Age
‡1 Complication
N
%
N
55.01 (mean)
0.08 (SD)
59.33 (Mean)
%
0.29 <0.0001 (SD) <0.0001
Sex Female
127,750 51.51
6435
46.70
Male
120,250 48.49
7345
53.30 <0.0001
Race Asian
3335
1.34
195
1.42
Black
23,750
9.58
2170
15.75
2778
5.60
1110
8.06
39,838 80.32
9775
70.94
3.16
530
3.85
Commercial/health maintenance organization
11,915 48.35
4400
31.93
Medicaid
19,545
7.88
1590
11.54
Medicare
78,965 31.84
6310
45.79
Uninsured/other
29,575 11.93
1480
10.74
Length of stay
1.85 (Mean)
Hispanic White Other
1567
<0.0001
Payer
0.02 (SD)
11.20 (Mean)
0.32 <0.0001 (SD)
Operation year
METHODS The Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) years 2013e2014 was queried for all hospitalizations during which the patient underwent primary ACDF using the International Classification of Diseases, Ninth Revision (ICD-9) code 81.02. Those undergoing concurrent posterior cervical procedures within the same hospitalization were excluded with the ICD-9 code 81.03. For each patient, the presence of Elixhauser Comorbidity Index and CCI component comorbidities were identified by matching the ICD-9 codes proposed and validated by Quan et al. with the corresponding ICD-9 billing codes found in the NIS data set.15,16,34 Chronicity markers were used to identify only chronic conditions as comorbidities and exclude new diagnoses. Weighting of each individual comorbidity in each of the 2 indexes was accomplished by using the weighting scores proposed by Charlson et al.15 and modified by Deyo et al.17 for the CCI comorbidities and by the HCUP mortality weights for the Elixhauser comorbidities.15-17,34-36 Demographic variables in the present study included age, sex, race (categorized as white, black, Asian, Hispanic, or other), and primary payer (categorized as commercial, Medicaid, Medicare, and other/uninsured). Comorbidity variables included in both the
e2
www.SCIENCEDIRECT.com
P Value
0.2564
2013
125,484 50.60
6760
49.06
2014
122,515 49.40
7020
50.94 <0.0001
Elixhauser Comorbidity Index <0
54,975 22.17
2505
18.18
0
143,095 57.70
3620
26.27
1e4
28,710 11.58
2060
14.95
5
21,220
5595
40.60
8.56
<0.0001
Charlson Comorbidity Index 0
155,790 62.82
5160
37.45
1e2
81,240 32.76
5685
41.26
3e4
8655
3.49
2090
15.17
5
2315
0.93
845
6.13
SD, standard deviation.
Elixhauser Index and CCI can be found in their respective original references. Complications coded with ICD-9 codes were as follows: airway complications requiring reintubation or tracheostomy (procedure codes 03.11, 03.12, 03.129, 960.4, 960.5, 0BH17EZ,
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 2. Univariate Regression of Charlson and Elixhauser Comorbidity Index Components of Patients Undergoing Anterior Cervical Discectomy and Fusion Who Experienced No Complications versus 1 Complication (N ¼ 261,780) ‡1 Complication
No Complication Category
N
%
N
%
P Value
11,490
4.63
2785
20.21
<0.0001
4375
1.76
470
3.41
<0.0001
Elixhauser comorbidities Cardiac arrhythmia Valvular disease
655
0.26
325
1.71
<0.0001
115,460
46.56
8015
58.16
<0.0001
6160
2.48
1390
10.09
<0.0001
39,705
16.01
2965
21.52
<0.0001
3790
1.53
950
6.89
<0.0001
23,835
9.61
1610
11.68
0.0004
Renal failure
5345
2.16
1485
10.78
<0.0001
Liver disease
3155
1.27
440
3.19
<0.0001
Pulmonary circulation disorder Hypertension Neurological disease Diabetes uncomplicated Diabetes complicated Hypothyroidism
Lymphoma
425
0.17
85
0.62
<0.0001
Solid tumor
1045
0.42
260
1.89
<0.0001
Rheumatoid arthritis/collagen vascular disease
7475
3.01
715
5.19
<0.0001
Coagulopathy Obesity
1795
0.72
795
5.77
<0.0001
29,490
11.89
2045
14.84
<0.0001
Weight loss
575
0.23
1275
9.25
<0.0001
Fluid or electrolyte abnormality
85
0.03
90
0.65
<0.0001
Blood loss anemia
290
0.12
100
0.73
<0.0001
Deficiency anemia
0
0.00
0
0.00
Not available
Alcohol use disorder
4220
1.70
1105
8.02
<0.0001
Drug use disorder
3845
1.55
675
4.90
<0.0001
Psychoses
815
0.33
265
1.92
<0.0001
Depression
40,850
16.47
2445
17.74
0.0796
Congestive heart failure
3805
1.53
1110
8.06
<0.0001
Peripheral vascular disease
4510
1.82
665
4.83
<0.0001
Chronic pulmonary disease
Shared comorbidities
40,975
16.52
2790
20.25
<0.0001
Paralysis
3380
1.36
1675
12.16
<0.0001
Human immunodeficiency virus/AIDS
220
0.09
20
0.15
0.3413
Peptic ulcer disease
840
0.34
125
0.91
<0.0001
Metastatic cancer
725
0.29
225
1.63
<0.0001
Myocardial infarction
7145
2.88
705
5.12
<0.0001
Cerebrovascular disease
2725
1.10
790
5.73
<0.0001
Dementia
225
0.09
100
0.73
<0.0001
Rheumatic disease
6840
2.76
570
4.14
<0.0001
Charlson comorbidities
Continues
WORLD NEUROSURGERY -: e1-e10, - 2019
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e3
Q6
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 2. Continued ‡1 Complication
No Complication Category
N
%
N
%
P Value
Mild liver disease
3120
1.26
430
3.12
<0.0001
Severe liver disease
120
0.05
50
0.36
<0.0001
40,240
16.23
3255
23.62
<0.0001
Diabetes complicated
3400
1.37
725
5.26
<0.0001
Renal disease
5355
2.16
1485
10.78
<0.0001
Any malignancy
1735
0.70
360
2.61
<0.0001
Peptic ulcer disease
840
0.34
140
1.02
<0.0001
Diabetes uncomplicated
0B110F4, and 0B113F4), postoperative hemorrhagic anemia (285.1), myocardial infarction (MI, 997.1), cardiac arrest (427.5, 997.1), cerebrovascular accident (CVA, 997.02), deep vein thrombosis (DVT, 997.2), pneumonia (997.39), pulmonary embolism (PE, 415.11, 415.19), acute kidney injury (AKI, 997.5, 584.9, 584.6, 584.7, 584.8, 584.9), wound dehiscence (998.32), superficial surgical site infection (SSI, 998.59, 958.3, 998.51), sepsis (995.91), septic shock (998.02), and urinary tract infection (UTI, 599.0). Complications were categorized as major or minor based on the classification of Glassman et al.,37 with major complications including airway complications, MI, cardiac arrest, CVA, DVT, pneumonia, PE, sepsis, septic shock, and death, and minor complications including hemorrhagic anemia, AKI, dehiscence, SSI, and UTI. In addition, a dichotomous variable representing the occurrence of any of the ICD-9 coded complications listed earlier was created. Hospitalizations resulting in death were identified as having a discharge disposition as “Expired.” To assess differences in those who experienced a complication during their hospitalization, descriptive statistics of patient demographics were calculated using c2 for categorical variables and a Student t test for continuous variables. Furthermore, similar analyses were run on rates of Elixhauser Index and CCI comorbidities. To assess the predictive ability of the indices, the area under the receiver operating characteristic curve (AUC) was used as a measure of predictive ability for each individual complication, major complications, minor complications, any complication, and mortality in a logistic regression model. The receiver operating characteristic curve provides a graphic representation of sensitivity and specificity across a wide spectrum of values. Assessing the area under this curve allows for a simple measure of the predictive ability of the logistic regression model. A baseline model taking into account age, sex, race, and payer was used as the control, and separately the Elixhauser Index and CCI were added to this baseline model to assess their additive ability in terms of predictive power. Furthermore, the predictive ability of the Elixhauser Index and CCI in the presence of the baseline model was compared by calculating the relative increase in the AUC of the more predictive model compared with the less predictive model. For example, a difference in the AUC between the CCI and Elixhauser Comorbidity Index scores of 0.7 and 0.8, when the base model AUC is 0.65, corresponds to a 67% relative increase in AUC:
e4
www.SCIENCEDIRECT.com
(0.80e0.65) e (0.70e0.65)/(0.80e0.65) ¼ 0.67.38 The ability of each index to assign a high probability of complication occurrence to those patients who experienced that complication was categorized. Models that produced AUC values <0.7 were categorized as “poor,” values between 0.7 and 0.8 were categorized as “acceptable,” values between 0.8 and 0.9 were categorized as “excellent,” and values >0.9 were categorized as “outstanding.”39 All analyses were performed on Statistical Analysis Software version 9.4 (SAS Inc., Cary, North Carolina, USA) using specific procedures designed to account for the complex survey design of the NIS (e.g., proc surveyfreq and surveylogistic). More information on these procedures can be found on the HCUP distributor Web site.40 RESULTS An estimated 261,780 hospitalizations in which ACDF was performed between 2013 and 2014 were retrieved from the NIS database. Those who experienced 1 complication (N ¼ 13,780, 5.3%) were statistically different from the no-complication cohort (N ¼ 248,000, 94.7%) in that they were predominantly male (53.30% vs. 48.49%; P < 0.0001), had a higher proportion of nonwhite patients (29.06% vs. 19.68%; P < 0.0001), had a higher proportion of both Medicare (45.79% vs. 31.84%; P < 0.0001) and Medicaid (11.54% vs. 7.88%; P < 0.0001) patients, and were on average >4 years older (59.33 years vs. 55.01 years; P < 0.0001). In addition, the complication cohort had a greater percentage of patients with Elixhauser Index 1e4 and 5 as well as CCI score of 1e2, 3e4, and 5, indicating a higher overall comorbidity burden by both measures. The average length of hospital stay for the complication cohort was 11.2 days, whereas for the no-complication cohort it was only 1.9 days (Table 1). Of the comorbidity components used in the Elixhauser Comorbidity Index, the 4 most common conditions in the complication cohort were hypertension (58.2%), uncomplicated diabetes (21.5%), chronic pulmonary disease (20.2%), and cardiac arrhythmia (20.2%). For the no-complication cohort, the 4 most common comorbidities were hypertension (46.6%), depression (16.5%), chronic pulmonary disease (16.5%), and uncomplicated diabetes (16.0%). When the comorbidity components used in the CCI score were examined, the 4 most common conditions in the
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 3. Postoperative Complication Rates and Complication Severity For All Patients Undergoing Anterior Cervical Discectomy and Fusion (N ¼ 261,780) Category
N
%
Severity
Airway
3265
1.25
Major
Hemorrhagic anemia
4705
1.80
Minor
Acute kidney injury
2865
1.09
Minor
Myocardial infarction
650
0.25
Major
Cardiac arrest
1050
0.40
Major
Cerebrovascular accident
55
0.02
Major
Deep vein thrombosis
145
0.06
Major
Pneumonia
535
0.20
Major
Pulmonary embolism
395
0.15
Major
Dehiscence
45
0.02
Major
Superficial surgical site infection
205
0.08
Minor
Sepsis
1110
0.42
Major
Septic shock
350
0.13
Major
Urinary tract infection
3750
1.43
Minor
Death
650
0.25
Major
1 minor complications
10,360
4.00
—
1 major complications
5500
2.10
—
13,780
5.26
—
1 complications
complication cohort were uncomplicated diabetes (23.6%), chronic pulmonary disease (20.2%), paralysis (12.2%), and chronic kidney disease (10.8%), whereas for the no-complication cohort, they were chronic pulmonary disease (16.5%), uncomplicated diabetes (16.2%), previous MI (2.9%), and connective tissue disease (2.8%) (Table 2). The cohort of hospitalizations during which 1 postoperative complication was reported had statistically significantly higher rates of every comorbidity in both measures except for depression (P ¼ 0.0796), human immunodeficiency virus/AIDS (P ¼ 0.3413), and deficiency anemia (unable to assess P value because of incidence of 0.0%). The most common complications in this study were postoperative hemorrhagic anemia (1.8%), UTI (1.4%), airway complications (1.2%), and AKI (1.1%). In addition, 5.3% of records were found to have experienced any complication, whereas 1.9% and 4.0% of records were found to have experienced any minor complication and any major complication, respectively (Table 3). On construction of the base, CCI, and Elixhauser models and analysis of the produced AUC values, the Elixhauser model was found to be statistically superior to the CCI model in the prediction of airway complications (AUC, 0.81 vs. 0.75; P < 0.0001; 45% superior), hemorrhagic anemia (AUC, 0.67 vs. 0.63; P ¼ 0.0015; 41% superior), PE (AUC, 0.91 vs. 0.77; P < 0.0001; 60% superior), wound dehiscence (AUC, 0.80 vs. 0.55; P ¼ 0.0080; 106% superior), sepsis (AUC, 0.87 vs. 0.82; P ¼ 0.0001; 39% superior), and septic shock (AUC, 0.94 vs. 0.83; P < 0.0001; 64% superior) as
WORLD NEUROSURGERY -: e1-e10, - 2019
well as any minor complication (AUC, 0.66 vs. 0.63; P ¼ 0.0016; 41% superior) and any major complication (AUC, 0.74 vs. 0.73; P ¼ 0.0487; 17% superior) (Figure 1). The CCI model was not found to be statistically superior to the Elixhauser model in the prediction of any individual postoperative complication, or in the prediction of any minor complication, any major complication, or any complication (Table 4). Neither the CCI nor the Elixhauser model proved to be statistically superior in the prediction of postoperative AKI (AUC, 0.80 vs. 0.81; P ¼ 0.1896), MI (AUC, 0.69 vs. 0.69; P ¼ 0.8815), cardiac arrest (AUC, 0.70 vs. 0.72; P ¼ 0.2900), CVA (AUC, 0.92 vs. 0.92; P ¼ 0.9726), DVT (AUC, 0.78 vs. 0.82; P ¼ 0.3375), pneumonia (AUC, 0.65 vs. 0.64; P ¼ 0.7182), SSI (AUC, 0.71 vs. 0.72; P ¼ 0.7798), and UTI (AUC, 0.73 vs. 0.73; P ¼ 0.5619). In addition, neither the CCI nor the Elixhauser model proved superior in the prediction of mortality (AUC, 0.87 vs. 0.90; P ¼ 0.1568) or any complication (AUC, 0.70 vs. 0.70; P ¼ 0.1077) after elective ACDF. The ability of the base model and CCI and Elixhauser models to accurately predict the occurrence of any given complication in a patient who subsequently experienced that complication is represented in Table 5. The Elixhauser model proved to be an excellent or outstanding predictor of airway complications, AKI, CVA, DVT, PE, sepsis, septic shock, and death, whereas it was a poor predictor of postoperative hemorrhagic anemia, MI, pneumonia, and any minor complication. The CCI model proved to be an excellent or outstanding predictor of AKI, CVA, PE, sepsis, septic shock, and death, whereas it was a poor predictor of postoperative hemorrhagic anemia, MI, pneumonia, wound dehiscence, any minor complication, and any complication overall (Table 5). DISCUSSION With an ever-increasing focus on surgical quality improvement, the preoperative risk evaluation of surgical candidates continues to be of the utmost importance. One of the most important factors to consider in this evaluation is the patient’s specific comorbidity burden. Although the presence of individual comorbidities has been linked with increased risk of various postoperative adverse events, a patient’s comorbidity burden is often assessed in aggregate, most commonly with the CCI or the Elixhauser Comorbidity Index in orthopedic surgery. Despite the frequency with which they are used, to the best of our knowledge, no literature exists investigating the predictive capabilities of these 2 measures with respect to specific postoperative complications in spine procedures, and more specifically in ACDF. This study aimed to fill this gap in the literature and found the Elixhauser model to be either an excellent or an outstanding predictor of 8 of the 15 complications investigated and to significantly outperform the CCI in the prediction of 6 of these complications. Similar to the results of previous studies on cervical spine surgical patients by Chitale et al.41 and patients with cervical spine fracture by Menendez et Al.,19 our investigation showed the CCI model to be a poor predictor of any complication after ACDF. In addition, the CCI model was found to be a poor predictor of any minor complication and an acceptable predictor of any major complication. The Elixhauser model performed slightly
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e5
523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Figure 1. Receiver operating characteristic (ROC) curves of the Elixhauser Index, Charlson Index, and base model for the prediction of (A) death, (B)
e6
www.SCIENCEDIRECT.com
minor complications, (C) major complications, and (D) any complication.
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 4. Predictive Ability and Comparison of Base Model, Weighted Elixhauser Comorbidity Index Model, and Weighted Charlson Comorbidity Index Model with Respect to Perioperative Complications for All Patients Undergoing Anterior Cervical Discectomy and Fusion (N ¼ 261,780) Q7
Base Model AUC*
Base Model 95% CI*
Elixhauser AUC
Elixhauser 95% CI
Charlson AUC
Charlson 95% CI
Elixhauser vs. Charlson P Value
Superiority
Percent Superior
Airway
0.68
0.67e0.70
0.81
0.79e0.82
0.75
0.74e0.76
<0.0001
Elixhauser
45.06
Hemorrhagic anemia
0.59
0.57e0.60
0.67
0.65e0.68
0.63
0.62e0.65
0.0015
Elixhauser
41.44
Acute kidney injury
0.73
0.72e0.75
0.81
0.80e0.83
0.80
0.79e0.82
0.1896
—
—
Myocardial infarction
0.68
0.65e0.72
0.69
0.65e0.72
0.69
0.66e0.73
0.8815
—
—
Cardiac arrest
0.69
0.66e0.71
0.72
0.70e0.75
0.70
0.68e0.73
0.2900
—
—
Cerebrovascular accident
0.90
0.85e0.95
0.92
0.89e0.95
0.92
0.87e0.97
0.9726
—
—
Deep vein thrombosis
0.75
0.69e0.81
0.82
0.76e0.87
0.78
0.72e0.84
0.3375
—
—
Pneumonia
0.64
0.60e0.67
0.64
0.60e0.68
0.65
0.62e0.68
0.7182
—
—
Pulmonary embolism
0.68
0.64e0.72
0.91
0.89e0.94
0.77
0.74e0.81
<0.0001
Elixhauser
59.90
Dehiscence
0.57
0.44e0.70
0.80
0.69e0.90
0.55
0.41e0.70
0.0080
Elixhauser
105.77
Superficial surgical site infection
0.67
0.60e0.74
0.72
0.66e0.79
0.71
0.64e0.78
0.7798
—
—
Sepsis
0.73
0.71e0.75
0.87
0.85e0.89
0.82
0.80e0.84
0.0001
Elixhauser
38.51
Septic shock
0.77
0.73e0.81
0.94
0.92e0.95
0.83
0.80e0.86
<0.0001
Elixhauser
63.50
Urinary tract infection
0.68
0.67e0.70
0.73
0.72e0.75
0.73
0.71e0.74
0.5619
—
—
Category
Death
0.82
0.80e0.85
0.90
0.87e0.92
0.87
0.85e0.90
0.1568
—
—
Any minor complication
0.64
0.63e0.65
0.70
0.69e0.71
0.68
0.67e0.69
0.0108
Elixhauser
37.06
Any major complication
0.67
0.65e0.68
0.77
0.76e0.79
0.73
0.72e0.74
<0.0001
Elixhauser
76.92
Any complication
0.64
0.63e0.65
0.71
0.70e0.71
0.70
0.69e0.70
0.1077
—
—
AUC, area under the receiver operating characteristic curve; CI, confidence interval. *Base model: age, sex, race, and primary payer.
better, producing acceptable predictions of any complication and any major complication and poor prediction of any minor complication. These results seem to suggest that both models are of relatively little value in predicting postoperative complications. However, when postoperative complications are analyzed individually rather than in aggregate, the data suggest a different interpretation. This study found the Elixhauser Comorbidity Index to be statistically superior to the CCI in predicting airway complications, bleeding, PE, wound dehiscence, sepsis, and septic shock during the inpatient postoperative period and not to be statistically inferior in the prediction of any individual complications. Furthermore, the Elixhauser model proved to be an excellent or outstanding predictor of airway complications, AKI, CVA, DVT, PE, sepsis, septic shock, and death, which collectively comprise 44.68% of all complications recorded in this study. Despite its strong performance in the prediction of various individual postoperative adverse events, the Elixhauser model performed poorly in its predictive capabilities of any minor postoperative complication, with a calculated area under the curve of 0.66. This relatively small value is most likely the result of poor predictive capabilities of superficial SSI, bleeding anemia, and
WORLD NEUROSURGERY -: e1-e10, - 2019
UTI, given that they collectively comprise 74.8% of all minor complications recorded in this study. Existing studies of multiinstitutional data on patients undergoing a variety of orthopedic procedures have identified multiple patient characteristics and comorbidities independently predictive of each of these complications individually.42-47 Although the Elixhauser model used in this study accounts for many of the risk factors identified in the literature, it does not take into account chronic preoperative steroid use. In 2 studies from 2016, Alvarez et al.47 found chronic preoperative steroid use to double the risk of postoperative UTI and De la Garza-Ramos et al.44 reported that it greatly increased the risk of SSI, with an odds ratio of 3.66 in similar patients. In addition, a recent study of 22,903 patients undergoing posterior lumbar fusion surgery48 found chronic preoperative steroid use to be predictive of increased risk of SSI, wound dehiscence, UTI, and PE. Collectively, these studies suggest that the addition of chronic steroid use to the Elixhauser model may improve its ability to predict minor postoperative complications. With the recent trends in health care reform aimed at containing costs in orthopedic surgery (i.e., the proposed implementation of the Surgical Hip and Femur Fracture Treatment bundled payment plan), the results of this study are also
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e7
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 5. Categorization of Predictive Capability of Base Model, Elixhauser Comorbidity Model, and Charlson Comorbidity Index Model in Predicting Inpatient Postoperative Complications Base Model*
Elixhauser Comorbidity Model
Charlson Comorbidity Model
Airway
Poor
Excellent
Acceptable
Hemorrhagic anemia
Poor
Poor
Poor
Acute kidney injury
Complication
Acceptable
Excellent
Excellent
Myocardial infarction
Poor
Poor
Poor
Cardiac arrest
Poor
Acceptable
Acceptable
Cerebrovascular accident
Outstanding
Outstanding
Outstanding
Deep vein thrombosis
Acceptable
Excellent
Acceptable
Pneumonia
Poor
Poor
Poor
Pulmonary embolism
Poor
Outstanding
Excellent
Dehiscence
Poor
Acceptable
Poor
Superficial surgical site infection
Poor
Acceptable
Acceptable
Sepsis
Acceptable
Excellent
Excellent
Septic shock
Acceptable
Outstanding
Excellent
Urinary tract infection
Poor
Acceptable
Acceptable
Excellent
Excellent
Excellent
Any minor complication
Poor
Acceptable
Poor
Any major complication
Poor
Acceptable
Acceptable
Any complication
Poor
Acceptable
Poor
Death
*Poor, c < 0.70; acceptable, c ¼ 0.70e0.80; excellent, c ¼ 0.80e0.90; outstanding, c > 0.90.
applicable to future reimbursement policy. As is well established, bundled payment plans without risk adjustment measures may be inequitable to providers and may restrict access to care for certain patients. To combat this situation, the Surgical Hip and Femur Fracture Treatment model plans to adjust for comorbidities by separating patients into 3 cohorts based on diagnosis-related group (DRG) codes: DRG 482 (no complication or comorbidity), DRG 481 (complication or comorbidity), and DRG 480 (major complication or comorbidity).49 The list of comorbidities and complications that qualify a patient for inclusion in DRG 480 and 481 cohorts is extensive and can be found on the Centers for Medicare and Medicaid Web site.50 However, based on a study of current reimbursement data after surgical repair of hip and femur fractures by Cairns et al.,51 the CCI had a significantly stronger association with the level of reimbursement compared with using solely the DRG to
e8
www.SCIENCEDIRECT.com
determine reimbursement. Because this study shows a significant association between patient comorbidities and total hospital cost, the literature clearly shows a strong association between inpatient complications and total hospital cost, and the current shows that the Elixhauser model is superior to the CCI model in predicting inpatient complications, the Elixhauser model may be superior to both the DRG and CCI models for explaining reimbursement rates.52 Although more research is certainly needed to confirm this hypothesis and extend it to surgical procedures of the spine, Centers for Medicare and Medicaid should consider implementing the Elixhauser Comorbidity Index in place of either the DRG or CCI scores to optimize equitability in their reimbursements and avoid disincentivizing surgical treatment of patients with a higher comorbidity burden. In an era of ballooning costs, when U.S. health care expenditures accounted for nearly 18% of the annual GDP in 2017, this study highlights the strong association between inpatient complications and drastically increased resource use. Specifically, this study shows that in patients who experienced no inpatient complications postoperatively, the average length of stay was 1.8 days whereas those who experienced 1 complication averaged >11 days in the hospital after the procedure. A recent study by Virk et al.53 found that each extra day in the hospital after ACDF accrued an additional cost of $652. Based on these numbers, the patients in our study who experienced 1 postoperative complication averaged a cost increase >$24,000, more than doubling the mean cost of all patients undergoing ACDF ($18,622), as reported by Kalakoti et al.14 Given that 5.3% of patients in this study experienced 1 complication, this finding represents a substantial use of limited health care resources and highlights the important implication of this study: if comorbidity indexes can accurately predict high-risk patients and their potential complications preoperatively, there is potential for implementation of preventive measures, conserving hospital resources, and optimizing patient outcomes. The results of this study should be interpreted after taking into account several potential limitations. First, chronic medical conditions tend to be underreported in large administrative databases even when the number of diagnoses is not limited.54-56 Second, when ICD-9 codes are used to determine patient comorbidities, as in this study, coding errors cannot be completely avoided.57 Third, the NIS reports information only with respect to the patient’s index surgery and subsequent inpatient stay, and thus all complications occurring after discharge from the hospital were not captured. In addition, this study was retrospective and thus subject to recall bias. For example, in patients who experienced an inpatient complication, providers may have been more likely to accurately record all predisposing comorbid risk factors, whereas patients who experienced no complications may have had their comorbidities underreported, resulting in an overestimation of the predictive capabilities of the CCI and Elixhauser models. Also, the CCI and Elixhauser models in this study are limited in their generalizability to populations with different characteristics. Another limitation is that ICD-9 codes may not be readily available or easy to use in typical outpatient clinic settings, thereby potentially limiting their clinical applicability. With respect to the proposal of adding preoperative steroid use to the list of
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Elixhauser comorbidities as mentioned in the Discussion section, the appropriate weighting of this variable and its potential to either positively or negatively affect the performance of the model in predicting the various other postoperative complications is uncertain and requires further investigation. The Elixhauser Comorbidity Index proved to be either excellent or outstanding in predicting 8 of the 15 complications investigated and performed significantly better than the CCI in predicting 6 of these 15 complications. Because postoperative complication risk prediction and avoidance has the potential to
Q3
comorbidities on Medicare diagnosis-related group reimbursement for adult spinal deformity surgery. Neurosurg Focus. 2017;43:E11.
REFERENCES 1. Song KJ, Choi BY. Current concepts of anterior cervical discectomy and fusion: a review of literature. Asian Spine J. 2014;8:531-539. 2. Sugawara T. Anterior cervical spine surgery for degenerative disease: a review. Neurol Med Chir (Tokyo). 2015;55:540-546. 3. Di Capua J, Somani S, Kim JS, et al. Elderly age as a risk factor for 30-day postoperative outcomes following elective anterior cervical discectomy and fusion. Global Spine J. 2017;7:425-431. 4. Weiler C, Schietzsch M, Kirchner T, Nerlich AG, Boos N, Wuertz K. Age-related changes in human cervical, thoracal and lumbar intervertebral disc exhibit a strong intra-individual correlation. Eur Spine J. 2012;21(suppl 6):S810-S818. 5. Pine M, Jordan HS, Elixhauser A, et al. Enhancement of claims data to improve risk adjustment of hospital mortality. JAMA. 2007;297:71-76. 6. Phan K, Wang N, Kim JS, et al. Effect of preoperative anemia on the outcomes of anterior cervical discectomy and fusion. Global Spine J. 2017;7: 441-447. 7. Narain AS, Hijji FY, Haws BE, et al. Impact of body mass index on surgical outcomes, narcotics consumption, and hospital costs following anterior cervical discectomy and fusion. J Neurosurg Spine. 2018;28:160-166. 8. Chughtai M, Sultan AA, Padilla J, et al. Postoperative stroke after anterior cervical discectomy and fusion in patients with carotid artery stenosis: a statewide database analysis. Spine J. 2019;19: 597-601. 9. Kapadia BH, Decker SI, Boylan MR, Shah NV, Paulino CB. Risk factors for cerebrospinal fluid leak following anterior cervical discectomy and fusion. Clin Spine Surg. 2019;32:E86-E90. 10. Zaki O, Jain N, Yu E, Khan SN. 30- and 90-day unplanned readmission rates, causes, and risk factors after cervical fusion: a single institution analysis. Spine (Phila Pa 1976). 2019;44:762-769. 11. Bovonratwet P, Fu MC, Tyagi V, et al. Incidence, risk factors, and clinical implications of postoperative hematoma requiring reoperation following anterior cervical discectomy and fusion. Spine (Phila Pa 1976). 2019;44:543-549. 12. Nunley PD, Mundis GM, Fessler RG, et al. Impact of case type, length of stay, institution type, and
improve patient outcomes and optimize the use of scarce hospital resources, the use of the Elixhauser Comorbidity Index to predict specific inpatient complications after elective ACDF should be considered. Early identification of patients at high risk for inpatient complications may allow for the implementation of intervention strategies to optimize modifiable risk factors. Future research should assess the performance of the Elixhauser Index in predicting long-term complications and optimize the model to more accurately predict the various complications for which it proved to be inadequate.
13. Daley J, Henderson WG, Khuri SF. Risk-adjusted surgical outcomes. Annu Rev Med. 2001;52:275-287. 14. Kalakoti P, Gao Y, Hendrickson NR, Pugely AJ. Preparing for bundled payments in cervical spine surgery: do we understand the influence of patient, hospital, and procedural factors on the cost and length of stay? Spine. 2019;44:334-345. 15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383. 16. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8-27. 17. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45: 613-619. 18. Menendez ME, Neuhaus V, Ring D. Inpatient mortality after orthopaedic surgery. Int Orthop. 2015;39:1307-1314.
survival using administrative health data. Cancer. 2011;117:1957-1965. 25. Grendar J, Shaheen AA, Myers RP, et al. Predicting in-hospital mortality in patients undergoing complex gastrointestinal surgery: determining the optimal risk adjustment method. Arch Surg. 2012; 147:126-135. 26. Chu YT, Ng YY, Wu SC. Comparison of different comorbidity measures for use with administrative data in predicting short- and long-term mortality. BMC Health Serv Res. 2010;10:140. 27. Kim CY, Sivasundaram L, LaBelle MW, Trivedi NN, Liu RW, Gillespie RJ. Predicting adverse events, length of stay, and discharge disposition following shoulder arthroplasty: a comparison of the Elixhauser Comorbidity Measure and Charlson Comorbidity Index. J Shoulder Elbow Surg. 2018;27:1748-1755. 28. Sivaganesan A, Zuckerman S, Khan I, et al. Predictive model for medical and surgical readmissions following elective lumbar spine surgery: A national study of 33,674 patients. Spine (Phila Pa 1976). 2019;44:588-600. 29. Dobran M, Marini A, Nasi D, et al. Risk factors of surgical site infections in instrumented spine surgery. Surg Neurol Int. 2017;8:212.
19. Menendez ME, Ring D, Harris MB, Cha TD. Predicting in-hospital mortality in elderly patients with cervical spine fractures: a comparison of the Charlson and Elixhauser Comorbidity Measures. Spine. 2015;40:809-815.
30. Seicean A, Seicean S, Alan N, et al. Preoperative anemia and perioperative outcomes in patients who undergo elective spine surgery. Spine. 2013;38: 1331-1341.
20. Farley JF, Harley CR, Devine JW. A comparison of comorbidity measurements to predict healthcare expenditures. Am J Manag Care. 2006;12:110-119.
31. Bronheim RS, Oermann EK, Cho SK, Caridi JM. Coagulation profile as a risk factor for 30-day morbidity following cervical laminectomy and fusion. Spine. 2018;43:239-247.
21. Poses RM, McClish DK, Smith WR, Bekes C, Scott WE. Prediction of survival of critically ill patients by admission comorbidity. J Clin Epidemiol. 1996;49:743-747.
32. Goyal A, Elminawy M, Kerezoudis P, et al. Impact of obesity on outcomes following lumbar spine surgery: a systematic review and meta-analysis. Clin Neurol Neurosurg. 2018;177:27-36.
22. Menendez ME, Neuhaus V, van Dijk CN, Ring D. The Elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat Res. 2014;472:2878-2886.
33. Dias CGP, Roberto BB, Basaglia L, Lenza M, Nicolau RJ, Ferretti M. Interaction of demographic factors with the results of the surgery for degenerative disease of the cervical spine: a retrospective evaluation. Rev Bras Ortop. 2017;52:575-581.
23. Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data. Med Care. 2004;42:355-360.
34. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9CM and ICD-10 administrative data. Med Care. 2005;43:1130-1139.
24. Lieffers JR, Baracos VE, Winget M, Fassbender K. A comparison of Charlson and Elixhauser comorbidity measures to predict colorectal cancer
35. Moore BJ, White S, Washington R, Coenen N, Elixhauser A. Identifying increased risk of readmission and in-hospital mortality using hospital
WORLD NEUROSURGERY -: e1-e10, - 2019
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e9
987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
administrative data: the AHRQ Elixhauser Comorbidity Index. Med Care. 2017;55:698-705. 36. Available at: https://www.hcup-us.ahrq.gov/ toolssoftware/comorbidity/comindex2012-2015.txt. Accessed January 14, 2019. 37. Glassman SD, Hamill CL, Bridwell KH, Schwab FJ, Dimar JR, Lowe TG. The impact of perioperative complications on clinical outcome in adult deformity surgery. Spine. 2007;32:2764-2770. 38. Tseng VL, Yu F, Lum F, Coleman AL. Risk of fractures following cataract surgery in Medicare beneficiaries. JAMA. 2012;308:493-501. 39. Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64:749-759. 40. HCUP Methods Series Calculating National Inpatient Sample (NIS) Variances for Data Years 2012 and Later. Available at: https://www.hcup-us.ahrq.gov/ reports/methods/2015_09.jsp. Accessed December 31, 2018. Q4 Q5
41. Chitale R, Campbell PG, Yadla S, Whitmore RG, Maltenfort MG, Ratliff JK. International classification of disease clinical modification 9 modeling of a patient comorbidity score predicts incidence of perioperative complications in a nationwide inpatient sample assessment of complications in spine surgery. J Spinal Disord Tech. 2015;28:126-133. 42. Sayeed Z, Anoushiravani AA, Chambers MC, et al. Comparing in-hospital total joint arthroplasty outcomes and resource consumption among underweight and morbidly obese patients. J Arthroplasty. 2016;31:2085-2090. 43. Anoushiravani AA, Sayeed Z, Chambers MC, et al. Assessing in-hospital outcomes and resource utilization after primary total joint arthroplasty among underweight patients. J Arthroplasty. 2016; 31:1407-1412.
e10
www.SCIENCEDIRECT.com
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
44. De la Garza-Ramos R, Abt NB, Kerezoudis P, et al. Deep-wound and organ-space infection after surgery for degenerative spine disease: an analysis from 2006 to 2012. Neurol Res. 2016;38:117-123.
54. Powell H, Lim LL-Y, Heller RF. Accuracy of administrative data to assess comorbidity in patients with heart disease. J Clin Epidemiol. 2001;54: 687-693.
45. Echt M, De la Garza Ramos R, Nakhla J, et al. The effect of cigarette smoking on wound complications after single-level posterolateral and interbody fusion for spondylolisthesis. World Neurosurg. 2018;116:e824-e829.
55. Preen DB, Holman CDJ, Lawrence DM, Baynham NJ, Semmens JB. Hospital chart review provided more accurate comorbidity information than data from a general practitioner survey or an administrative database. J Clin Epidemiol. 2004;57: 1295-1304.
46. Nahhas CR, Hu K, Mehta AI. Incidence and risk factors of wound complications in long segment instrumented thoracolumbar spinal fusions: a retrospective study. J Spine Surg. 2018;4:233-240. 47. Alvarez AP, Demzik AL, Alvi HM, Hardt KD, Manning DW. Risk factors for postoperative urinary tract infections in patients undergoing total joint arthroplasty. Adv Orthop. 2016;2016:7268985.
56. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. Comorbidities, complications, and coding bias. Does the number of diagnosis codes matter in predicting in-hospital mortality? JAMA. 1992;267:2197-2203. 57. DeFrances CJ, Lucas CA, Buie VC, Golosinskiy A. 2006 national hospital discharge survey. Natl Health Stat Report. 2008;5:1-20.
48. Ranson WA, White SJW, Cheung ZB, et al. The effects of chronic preoperative steroid therapy on perioperative complications following elective posterior lumbar fusion. Global Spine J. 2018;8: 834-841. 49. Surgical Hip and Femur Fracture Treatment Model. 50. Draft ICD-10-CM/PCS MS-DRGv28 Definitions Manual. Available at: https://www.cms.gov/ icd10manual/fullcode_cms/P0370.html. Accessed May 4, 2019. 51. Cairns MA, Ostrum RF, Clement RC. Refining risk adjustment for the proposed CMS surgical hip and femur fracture treatment bundled payment program. J Bone Joint Surg Am. 2018;100:269-277.
Conflict of interest statement: J.M.C. has financial relationships with Zimmer Biomet (paid consultant). S.K.C. has financial relationships with Corentec (paid consultant), Globus (paid consultant), Medtronic (paid consultant), and Zimmer (paid consultant; research support). None of the remaining authors has any financial interests or affiliations with institutions, organizations, or companies relevant to the article. None of the authors received payment or support in kind for any aspect of the submitted work. Received 21 July 2019; accepted 17 October 2019 Citation: World Neurosurg. (2019). https://doi.org/10.1016/j.wneu.2019.10.102
52. Healy MA, Mullard AJ, Campbell DA, Dimick JB. Hospital and payer costs associated with surgical complications. JAMA Surg. 2016;151:823-830.
Journal homepage: www.journals.elsevier.com/worldneurosurgery
53. Virk SS, Phillips FM, Moffatt-Bruce SD, Khan SN. What factors influence reimbursement for 1 to 2 level anterior cervical discectomy and fusion procedures? Spine. 2019;44:E33-E38.
1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.
Available online: www.sciencedirect.com
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160
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 45 46 47 48 49 50 51 52 53 54 55 56 57 58
Predicting In-Hospital Complications After Anterior Cervical Discectomy and Fusion: A Comparison of the Elixhauser and Charlson Comorbidity Indices Q10 Q11 Q1Q9 Q8
William A. Ranson1, Sean N. Neifert2, Zoe B. Cheung1, John M. Caridi2, Samuel K. Cho1
OBJECTIVE: The objective of this study was to determine the ability of the Elixhauser Comorbidity Index (ECI) and Charlson Comorbidity Index (CCI) to predict postoperative complications after anterior cervical discectomy and fusion (ACDF).
-
METHODS: This was a retrospective study of ACDF hospitalizations in the National Inpatient Sample from 2013 to 2014. The ECI and CCI were calculated, and patients who experienced postoperative complications were identified. The ability of these indexes to predict complications was compared using the c statistic (area under the receiver operating characteristic curve [AUC]). In addition, the CCI and ECI were compared with a base model that included age, sex, race, and primary payer.
-
RESULTS: A total of 261,780 patients were included. Patients who experienced a complication were more often male (P < 0.0001) and older (P < 0.0001). They also had a higher comorbidity burden as assessed by both the ECI (P < 0.0001) and the CCI (P < 0.0001). The ECI was superior in predicting airway complications (AUC, 0.81 vs. 0.75; P < 0.0001), hemorrhagic anemia (AUC, 0.67 vs. 0.63; P [ 0.0015), pulmonary embolism (AUC, 0.91 vs. 0.77; P < 0.0001), wound dehiscence (AUC, 0.80 vs. 0.55; P [ 0.0080), sepsis (AUC, 0.87 vs. 0.82; P [ 0.0001), and septic shock (AUC, 0.94 vs. 0.83; P < 0.0001). The CCI was not found to be superior to
-
Key words ACDF - Anterior cervical discectomy and fusion - Charlson - Comorbidity - Complication - Elixhauser -
Abbreviations and Acronyms ACDF: Anterior cervical discectomy and fusion AKI: Acute kidney injury AUC: Area under the receiver operating characteristic curve CCI: Charlson Comorbidity Index CMS: Centers for Medicare and Medicaid CVA: Cerebrovascular accident DRG: Diagnosis-related group DVT: Deep vein thrombosis HCUP: Healthcare Cost and Utilization Project
WORLD NEUROSURGERY -: e1-e10, - 2019
the ECI for predicting any complications. Both were excellent for predicting mortality (ECI AUC, 0.87; CCI AUC, 0.90). CONCLUSIONS: The ECI was superior to the CCI in predicting 6 of 15 complications in this study. Both are excellent tools for predicting mortality after ACDF.
-
INTRODUCTION
A
nterior cervical discectomy and fusion (ACDF) is a widely used and effective surgical procedure for the treatment of degenerative disease affecting the cervical spine. The most common indications for ACDF include cervical spondylosis and disc herniation refractory to conservative management, as well as certain subtypes of cervical fractures, neoplastic lesions, and infectious processes.1,2 Because cervical spine degenerative disease is a chronic condition that develops over many years, most patients undergoing ACDF are generally older, with more than half of these surgical patients being older than 50 years and a quarter older than 60 years.3,4 A well-recognized challenge to surgeons treating an older patient population is the presence of a relatively greater comorbidity burden compared with younger patients. For all surgical candidates, medical comorbidities are an important consideration because they have been shown to significantly affect the rate of various postoperative complications
ICD-9: International Classification of Diseases, Ninth Revision MI: Myocardial infarction NIS: National Inpatient Sample PE: Pulmonary embolism SSI: Surgical site infection UTI: Urinary tract infection From the Departments of 1Orthopaedics, and 2Neurosurgery, Mount Sinai Hospital, New York, New York, USA To whom correspondence should be addressed: Samuel Kang-Wook Cho, M.D. [E-mail: [email protected]]
Q12
Citation: World Neurosurg. (2019). https://doi.org/10.1016/j.wneu.2019.10.102 Journal homepage: www.journals.elsevier.com/world-neurosurgery Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e1
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
ORIGINAL ARTICLE ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES Q2
WILLIAM A. RANSON ET AL.
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
and also to play a critical role in provider performance assessment, reimbursement, and proposed bundled payment plans.5-14 To better assess surgical patients who may present with any combination of the multitude of possible comorbidities, various scoring measures have been proposed, with 2 specific indexes, the Charlson Comorbidity Index (CCI) and the Elixhauser Comorbidity Index, having attained prominence in orthopedic surgery.15,16 The CCI, originally developed in 1987 to predict 1-year oncologic mortality and subsequently adapted for use with administrative databases in 1992, has become a widely used risk adjustment method for patients undergoing spine surgery.15,17 The CCI consists of 19 medical conditions and has been shown to be a excellent predictor of inpatient mortality, postdischarge all-cause mortality, and health care expenditures in both medical and surgical patient populations.17-21 The Elixhauser Comorbidity Index is a newer risk adjustment method encompassing 31 medical comorbidities; its popularity has grown to rival the CCI, with recent literature suggesting a superior performance in predicting inpatient mortality after gastrointestinal, cardiac, and major orthopedic surgeries.19,22-27 Multiple comorbidities shown to affect perioperative outcomes in the spine surgery population in particular, including hypertension, obesity, coagulopathy, anemia, alcohol use disorder, drug use disorder, and depression, are present in the Elixhauser Index but not in the CCI.28-33 Although the literature has established that both the CCI and the Elixhauser Index are excellent predictors of mortality after orthopedic surgeries in aggregate, little investigation has been carried out into their value in predicting inpatient perioperative complications after spine surgery, with no studies specifically investigating ACDF.22 Using nationally representative data, this study sought to assess and compare the performance of the CCI and Elixhauser Comorbidity Index in predicting mortality and perioperative complications during the inpatient period in patients undergoing ACDF.
Table 1. Summary Demographic Statistics for National Inpatient Sample Patients Undergoing Anterior Cervical Discectomy and Fusion Who Experienced No Complications versus 1 Complication (N ¼ 261,780) No Complication Category Age
‡1 Complication
N
%
N
55.01 (mean)
0.08 (SD)
59.33 (Mean)
%
0.29 <0.0001 (SD) <0.0001
Sex Female
127,750 51.51
6435
46.70
Male
120,250 48.49
7345
53.30 <0.0001
Race Asian
3335
1.34
195
1.42
Black
23,750
9.58
2170
15.75
2778
5.60
1110
8.06
39,838 80.32
9775
70.94
3.16
530
3.85
Commercial/health maintenance organization
11,915 48.35
4400
31.93
Medicaid
19,545
7.88
1590
11.54
Medicare
78,965 31.84
6310
45.79
Uninsured/other
29,575 11.93
1480
10.74
Length of stay
1.85 (Mean)
Hispanic White Other
1567
<0.0001
Payer
0.02 (SD)
11.20 (Mean)
0.32 <0.0001 (SD)
Operation year
METHODS The Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) years 2013e2014 was queried for all hospitalizations during which the patient underwent primary ACDF using the International Classification of Diseases, Ninth Revision (ICD-9) code 81.02. Those undergoing concurrent posterior cervical procedures within the same hospitalization were excluded with the ICD-9 code 81.03. For each patient, the presence of Elixhauser Comorbidity Index and CCI component comorbidities were identified by matching the ICD-9 codes proposed and validated by Quan et al. with the corresponding ICD-9 billing codes found in the NIS data set.15,16,34 Chronicity markers were used to identify only chronic conditions as comorbidities and exclude new diagnoses. Weighting of each individual comorbidity in each of the 2 indexes was accomplished by using the weighting scores proposed by Charlson et al.15 and modified by Deyo et al.17 for the CCI comorbidities and by the HCUP mortality weights for the Elixhauser comorbidities.15-17,34-36 Demographic variables in the present study included age, sex, race (categorized as white, black, Asian, Hispanic, or other), and primary payer (categorized as commercial, Medicaid, Medicare, and other/uninsured). Comorbidity variables included in both the
e2
www.SCIENCEDIRECT.com
P Value
0.2564
2013
125,484 50.60
6760
49.06
2014
122,515 49.40
7020
50.94 <0.0001
Elixhauser Comorbidity Index <0
54,975 22.17
2505
18.18
0
143,095 57.70
3620
26.27
1e4
28,710 11.58
2060
14.95
5
21,220
5595
40.60
8.56
<0.0001
Charlson Comorbidity Index 0
155,790 62.82
5160
37.45
1e2
81,240 32.76
5685
41.26
3e4
8655
3.49
2090
15.17
5
2315
0.93
845
6.13
SD, standard deviation.
Elixhauser Index and CCI can be found in their respective original references. Complications coded with ICD-9 codes were as follows: airway complications requiring reintubation or tracheostomy (procedure codes 03.11, 03.12, 03.129, 960.4, 960.5, 0BH17EZ,
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 2. Univariate Regression of Charlson and Elixhauser Comorbidity Index Components of Patients Undergoing Anterior Cervical Discectomy and Fusion Who Experienced No Complications versus 1 Complication (N ¼ 261,780) ‡1 Complication
No Complication Category
N
%
N
%
P Value
11,490
4.63
2785
20.21
<0.0001
4375
1.76
470
3.41
<0.0001
Elixhauser comorbidities Cardiac arrhythmia Valvular disease
655
0.26
325
1.71
<0.0001
115,460
46.56
8015
58.16
<0.0001
6160
2.48
1390
10.09
<0.0001
39,705
16.01
2965
21.52
<0.0001
3790
1.53
950
6.89
<0.0001
23,835
9.61
1610
11.68
0.0004
Renal failure
5345
2.16
1485
10.78
<0.0001
Liver disease
3155
1.27
440
3.19
<0.0001
Pulmonary circulation disorder Hypertension Neurological disease Diabetes uncomplicated Diabetes complicated Hypothyroidism
Lymphoma
425
0.17
85
0.62
<0.0001
Solid tumor
1045
0.42
260
1.89
<0.0001
Rheumatoid arthritis/collagen vascular disease
7475
3.01
715
5.19
<0.0001
Coagulopathy Obesity
1795
0.72
795
5.77
<0.0001
29,490
11.89
2045
14.84
<0.0001
Weight loss
575
0.23
1275
9.25
<0.0001
Fluid or electrolyte abnormality
85
0.03
90
0.65
<0.0001
Blood loss anemia
290
0.12
100
0.73
<0.0001
Deficiency anemia
0
0.00
0
0.00
Not available
Alcohol use disorder
4220
1.70
1105
8.02
<0.0001
Drug use disorder
3845
1.55
675
4.90
<0.0001
Psychoses
815
0.33
265
1.92
<0.0001
Depression
40,850
16.47
2445
17.74
0.0796
Congestive heart failure
3805
1.53
1110
8.06
<0.0001
Peripheral vascular disease
4510
1.82
665
4.83
<0.0001
Chronic pulmonary disease
Shared comorbidities
40,975
16.52
2790
20.25
<0.0001
Paralysis
3380
1.36
1675
12.16
<0.0001
Human immunodeficiency virus/AIDS
220
0.09
20
0.15
0.3413
Peptic ulcer disease
840
0.34
125
0.91
<0.0001
Metastatic cancer
725
0.29
225
1.63
<0.0001
Myocardial infarction
7145
2.88
705
5.12
<0.0001
Cerebrovascular disease
2725
1.10
790
5.73
<0.0001
Dementia
225
0.09
100
0.73
<0.0001
Rheumatic disease
6840
2.76
570
4.14
<0.0001
Charlson comorbidities
Continues
WORLD NEUROSURGERY -: e1-e10, - 2019
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e3
Q6
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 2. Continued ‡1 Complication
No Complication Category
N
%
N
%
P Value
Mild liver disease
3120
1.26
430
3.12
<0.0001
Severe liver disease
120
0.05
50
0.36
<0.0001
40,240
16.23
3255
23.62
<0.0001
Diabetes complicated
3400
1.37
725
5.26
<0.0001
Renal disease
5355
2.16
1485
10.78
<0.0001
Any malignancy
1735
0.70
360
2.61
<0.0001
Peptic ulcer disease
840
0.34
140
1.02
<0.0001
Diabetes uncomplicated
0B110F4, and 0B113F4), postoperative hemorrhagic anemia (285.1), myocardial infarction (MI, 997.1), cardiac arrest (427.5, 997.1), cerebrovascular accident (CVA, 997.02), deep vein thrombosis (DVT, 997.2), pneumonia (997.39), pulmonary embolism (PE, 415.11, 415.19), acute kidney injury (AKI, 997.5, 584.9, 584.6, 584.7, 584.8, 584.9), wound dehiscence (998.32), superficial surgical site infection (SSI, 998.59, 958.3, 998.51), sepsis (995.91), septic shock (998.02), and urinary tract infection (UTI, 599.0). Complications were categorized as major or minor based on the classification of Glassman et al.,37 with major complications including airway complications, MI, cardiac arrest, CVA, DVT, pneumonia, PE, sepsis, septic shock, and death, and minor complications including hemorrhagic anemia, AKI, dehiscence, SSI, and UTI. In addition, a dichotomous variable representing the occurrence of any of the ICD-9 coded complications listed earlier was created. Hospitalizations resulting in death were identified as having a discharge disposition as “Expired.” To assess differences in those who experienced a complication during their hospitalization, descriptive statistics of patient demographics were calculated using c2 for categorical variables and a Student t test for continuous variables. Furthermore, similar analyses were run on rates of Elixhauser Index and CCI comorbidities. To assess the predictive ability of the indices, the area under the receiver operating characteristic curve (AUC) was used as a measure of predictive ability for each individual complication, major complications, minor complications, any complication, and mortality in a logistic regression model. The receiver operating characteristic curve provides a graphic representation of sensitivity and specificity across a wide spectrum of values. Assessing the area under this curve allows for a simple measure of the predictive ability of the logistic regression model. A baseline model taking into account age, sex, race, and payer was used as the control, and separately the Elixhauser Index and CCI were added to this baseline model to assess their additive ability in terms of predictive power. Furthermore, the predictive ability of the Elixhauser Index and CCI in the presence of the baseline model was compared by calculating the relative increase in the AUC of the more predictive model compared with the less predictive model. For example, a difference in the AUC between the CCI and Elixhauser Comorbidity Index scores of 0.7 and 0.8, when the base model AUC is 0.65, corresponds to a 67% relative increase in AUC:
e4
www.SCIENCEDIRECT.com
(0.80e0.65) e (0.70e0.65)/(0.80e0.65) ¼ 0.67.38 The ability of each index to assign a high probability of complication occurrence to those patients who experienced that complication was categorized. Models that produced AUC values <0.7 were categorized as “poor,” values between 0.7 and 0.8 were categorized as “acceptable,” values between 0.8 and 0.9 were categorized as “excellent,” and values >0.9 were categorized as “outstanding.”39 All analyses were performed on Statistical Analysis Software version 9.4 (SAS Inc., Cary, North Carolina, USA) using specific procedures designed to account for the complex survey design of the NIS (e.g., proc surveyfreq and surveylogistic). More information on these procedures can be found on the HCUP distributor Web site.40 RESULTS An estimated 261,780 hospitalizations in which ACDF was performed between 2013 and 2014 were retrieved from the NIS database. Those who experienced 1 complication (N ¼ 13,780, 5.3%) were statistically different from the no-complication cohort (N ¼ 248,000, 94.7%) in that they were predominantly male (53.30% vs. 48.49%; P < 0.0001), had a higher proportion of nonwhite patients (29.06% vs. 19.68%; P < 0.0001), had a higher proportion of both Medicare (45.79% vs. 31.84%; P < 0.0001) and Medicaid (11.54% vs. 7.88%; P < 0.0001) patients, and were on average >4 years older (59.33 years vs. 55.01 years; P < 0.0001). In addition, the complication cohort had a greater percentage of patients with Elixhauser Index 1e4 and 5 as well as CCI score of 1e2, 3e4, and 5, indicating a higher overall comorbidity burden by both measures. The average length of hospital stay for the complication cohort was 11.2 days, whereas for the no-complication cohort it was only 1.9 days (Table 1). Of the comorbidity components used in the Elixhauser Comorbidity Index, the 4 most common conditions in the complication cohort were hypertension (58.2%), uncomplicated diabetes (21.5%), chronic pulmonary disease (20.2%), and cardiac arrhythmia (20.2%). For the no-complication cohort, the 4 most common comorbidities were hypertension (46.6%), depression (16.5%), chronic pulmonary disease (16.5%), and uncomplicated diabetes (16.0%). When the comorbidity components used in the CCI score were examined, the 4 most common conditions in the
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 3. Postoperative Complication Rates and Complication Severity For All Patients Undergoing Anterior Cervical Discectomy and Fusion (N ¼ 261,780) Category
N
%
Severity
Airway
3265
1.25
Major
Hemorrhagic anemia
4705
1.80
Minor
Acute kidney injury
2865
1.09
Minor
Myocardial infarction
650
0.25
Major
Cardiac arrest
1050
0.40
Major
Cerebrovascular accident
55
0.02
Major
Deep vein thrombosis
145
0.06
Major
Pneumonia
535
0.20
Major
Pulmonary embolism
395
0.15
Major
Dehiscence
45
0.02
Major
Superficial surgical site infection
205
0.08
Minor
Sepsis
1110
0.42
Major
Septic shock
350
0.13
Major
Urinary tract infection
3750
1.43
Minor
Death
650
0.25
Major
1 minor complications
10,360
4.00
—
1 major complications
5500
2.10
—
13,780
5.26
—
1 complications
complication cohort were uncomplicated diabetes (23.6%), chronic pulmonary disease (20.2%), paralysis (12.2%), and chronic kidney disease (10.8%), whereas for the no-complication cohort, they were chronic pulmonary disease (16.5%), uncomplicated diabetes (16.2%), previous MI (2.9%), and connective tissue disease (2.8%) (Table 2). The cohort of hospitalizations during which 1 postoperative complication was reported had statistically significantly higher rates of every comorbidity in both measures except for depression (P ¼ 0.0796), human immunodeficiency virus/AIDS (P ¼ 0.3413), and deficiency anemia (unable to assess P value because of incidence of 0.0%). The most common complications in this study were postoperative hemorrhagic anemia (1.8%), UTI (1.4%), airway complications (1.2%), and AKI (1.1%). In addition, 5.3% of records were found to have experienced any complication, whereas 1.9% and 4.0% of records were found to have experienced any minor complication and any major complication, respectively (Table 3). On construction of the base, CCI, and Elixhauser models and analysis of the produced AUC values, the Elixhauser model was found to be statistically superior to the CCI model in the prediction of airway complications (AUC, 0.81 vs. 0.75; P < 0.0001; 45% superior), hemorrhagic anemia (AUC, 0.67 vs. 0.63; P ¼ 0.0015; 41% superior), PE (AUC, 0.91 vs. 0.77; P < 0.0001; 60% superior), wound dehiscence (AUC, 0.80 vs. 0.55; P ¼ 0.0080; 106% superior), sepsis (AUC, 0.87 vs. 0.82; P ¼ 0.0001; 39% superior), and septic shock (AUC, 0.94 vs. 0.83; P < 0.0001; 64% superior) as
WORLD NEUROSURGERY -: e1-e10, - 2019
well as any minor complication (AUC, 0.66 vs. 0.63; P ¼ 0.0016; 41% superior) and any major complication (AUC, 0.74 vs. 0.73; P ¼ 0.0487; 17% superior) (Figure 1). The CCI model was not found to be statistically superior to the Elixhauser model in the prediction of any individual postoperative complication, or in the prediction of any minor complication, any major complication, or any complication (Table 4). Neither the CCI nor the Elixhauser model proved to be statistically superior in the prediction of postoperative AKI (AUC, 0.80 vs. 0.81; P ¼ 0.1896), MI (AUC, 0.69 vs. 0.69; P ¼ 0.8815), cardiac arrest (AUC, 0.70 vs. 0.72; P ¼ 0.2900), CVA (AUC, 0.92 vs. 0.92; P ¼ 0.9726), DVT (AUC, 0.78 vs. 0.82; P ¼ 0.3375), pneumonia (AUC, 0.65 vs. 0.64; P ¼ 0.7182), SSI (AUC, 0.71 vs. 0.72; P ¼ 0.7798), and UTI (AUC, 0.73 vs. 0.73; P ¼ 0.5619). In addition, neither the CCI nor the Elixhauser model proved superior in the prediction of mortality (AUC, 0.87 vs. 0.90; P ¼ 0.1568) or any complication (AUC, 0.70 vs. 0.70; P ¼ 0.1077) after elective ACDF. The ability of the base model and CCI and Elixhauser models to accurately predict the occurrence of any given complication in a patient who subsequently experienced that complication is represented in Table 5. The Elixhauser model proved to be an excellent or outstanding predictor of airway complications, AKI, CVA, DVT, PE, sepsis, septic shock, and death, whereas it was a poor predictor of postoperative hemorrhagic anemia, MI, pneumonia, and any minor complication. The CCI model proved to be an excellent or outstanding predictor of AKI, CVA, PE, sepsis, septic shock, and death, whereas it was a poor predictor of postoperative hemorrhagic anemia, MI, pneumonia, wound dehiscence, any minor complication, and any complication overall (Table 5). DISCUSSION With an ever-increasing focus on surgical quality improvement, the preoperative risk evaluation of surgical candidates continues to be of the utmost importance. One of the most important factors to consider in this evaluation is the patient’s specific comorbidity burden. Although the presence of individual comorbidities has been linked with increased risk of various postoperative adverse events, a patient’s comorbidity burden is often assessed in aggregate, most commonly with the CCI or the Elixhauser Comorbidity Index in orthopedic surgery. Despite the frequency with which they are used, to the best of our knowledge, no literature exists investigating the predictive capabilities of these 2 measures with respect to specific postoperative complications in spine procedures, and more specifically in ACDF. This study aimed to fill this gap in the literature and found the Elixhauser model to be either an excellent or an outstanding predictor of 8 of the 15 complications investigated and to significantly outperform the CCI in the prediction of 6 of these complications. Similar to the results of previous studies on cervical spine surgical patients by Chitale et al.41 and patients with cervical spine fracture by Menendez et Al.,19 our investigation showed the CCI model to be a poor predictor of any complication after ACDF. In addition, the CCI model was found to be a poor predictor of any minor complication and an acceptable predictor of any major complication. The Elixhauser model performed slightly
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e5
523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Figure 1. Receiver operating characteristic (ROC) curves of the Elixhauser Index, Charlson Index, and base model for the prediction of (A) death, (B)
e6
www.SCIENCEDIRECT.com
minor complications, (C) major complications, and (D) any complication.
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 4. Predictive Ability and Comparison of Base Model, Weighted Elixhauser Comorbidity Index Model, and Weighted Charlson Comorbidity Index Model with Respect to Perioperative Complications for All Patients Undergoing Anterior Cervical Discectomy and Fusion (N ¼ 261,780) Q7
Base Model AUC*
Base Model 95% CI*
Elixhauser AUC
Elixhauser 95% CI
Charlson AUC
Charlson 95% CI
Elixhauser vs. Charlson P Value
Superiority
Percent Superior
Airway
0.68
0.67e0.70
0.81
0.79e0.82
0.75
0.74e0.76
<0.0001
Elixhauser
45.06
Hemorrhagic anemia
0.59
0.57e0.60
0.67
0.65e0.68
0.63
0.62e0.65
0.0015
Elixhauser
41.44
Acute kidney injury
0.73
0.72e0.75
0.81
0.80e0.83
0.80
0.79e0.82
0.1896
—
—
Myocardial infarction
0.68
0.65e0.72
0.69
0.65e0.72
0.69
0.66e0.73
0.8815
—
—
Cardiac arrest
0.69
0.66e0.71
0.72
0.70e0.75
0.70
0.68e0.73
0.2900
—
—
Cerebrovascular accident
0.90
0.85e0.95
0.92
0.89e0.95
0.92
0.87e0.97
0.9726
—
—
Deep vein thrombosis
0.75
0.69e0.81
0.82
0.76e0.87
0.78
0.72e0.84
0.3375
—
—
Pneumonia
0.64
0.60e0.67
0.64
0.60e0.68
0.65
0.62e0.68
0.7182
—
—
Pulmonary embolism
0.68
0.64e0.72
0.91
0.89e0.94
0.77
0.74e0.81
<0.0001
Elixhauser
59.90
Dehiscence
0.57
0.44e0.70
0.80
0.69e0.90
0.55
0.41e0.70
0.0080
Elixhauser
105.77
Superficial surgical site infection
0.67
0.60e0.74
0.72
0.66e0.79
0.71
0.64e0.78
0.7798
—
—
Sepsis
0.73
0.71e0.75
0.87
0.85e0.89
0.82
0.80e0.84
0.0001
Elixhauser
38.51
Septic shock
0.77
0.73e0.81
0.94
0.92e0.95
0.83
0.80e0.86
<0.0001
Elixhauser
63.50
Urinary tract infection
0.68
0.67e0.70
0.73
0.72e0.75
0.73
0.71e0.74
0.5619
—
—
Category
Death
0.82
0.80e0.85
0.90
0.87e0.92
0.87
0.85e0.90
0.1568
—
—
Any minor complication
0.64
0.63e0.65
0.70
0.69e0.71
0.68
0.67e0.69
0.0108
Elixhauser
37.06
Any major complication
0.67
0.65e0.68
0.77
0.76e0.79
0.73
0.72e0.74
<0.0001
Elixhauser
76.92
Any complication
0.64
0.63e0.65
0.71
0.70e0.71
0.70
0.69e0.70
0.1077
—
—
AUC, area under the receiver operating characteristic curve; CI, confidence interval. *Base model: age, sex, race, and primary payer.
better, producing acceptable predictions of any complication and any major complication and poor prediction of any minor complication. These results seem to suggest that both models are of relatively little value in predicting postoperative complications. However, when postoperative complications are analyzed individually rather than in aggregate, the data suggest a different interpretation. This study found the Elixhauser Comorbidity Index to be statistically superior to the CCI in predicting airway complications, bleeding, PE, wound dehiscence, sepsis, and septic shock during the inpatient postoperative period and not to be statistically inferior in the prediction of any individual complications. Furthermore, the Elixhauser model proved to be an excellent or outstanding predictor of airway complications, AKI, CVA, DVT, PE, sepsis, septic shock, and death, which collectively comprise 44.68% of all complications recorded in this study. Despite its strong performance in the prediction of various individual postoperative adverse events, the Elixhauser model performed poorly in its predictive capabilities of any minor postoperative complication, with a calculated area under the curve of 0.66. This relatively small value is most likely the result of poor predictive capabilities of superficial SSI, bleeding anemia, and
WORLD NEUROSURGERY -: e1-e10, - 2019
UTI, given that they collectively comprise 74.8% of all minor complications recorded in this study. Existing studies of multiinstitutional data on patients undergoing a variety of orthopedic procedures have identified multiple patient characteristics and comorbidities independently predictive of each of these complications individually.42-47 Although the Elixhauser model used in this study accounts for many of the risk factors identified in the literature, it does not take into account chronic preoperative steroid use. In 2 studies from 2016, Alvarez et al.47 found chronic preoperative steroid use to double the risk of postoperative UTI and De la Garza-Ramos et al.44 reported that it greatly increased the risk of SSI, with an odds ratio of 3.66 in similar patients. In addition, a recent study of 22,903 patients undergoing posterior lumbar fusion surgery48 found chronic preoperative steroid use to be predictive of increased risk of SSI, wound dehiscence, UTI, and PE. Collectively, these studies suggest that the addition of chronic steroid use to the Elixhauser model may improve its ability to predict minor postoperative complications. With the recent trends in health care reform aimed at containing costs in orthopedic surgery (i.e., the proposed implementation of the Surgical Hip and Femur Fracture Treatment bundled payment plan), the results of this study are also
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e7
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Table 5. Categorization of Predictive Capability of Base Model, Elixhauser Comorbidity Model, and Charlson Comorbidity Index Model in Predicting Inpatient Postoperative Complications Base Model*
Elixhauser Comorbidity Model
Charlson Comorbidity Model
Airway
Poor
Excellent
Acceptable
Hemorrhagic anemia
Poor
Poor
Poor
Acute kidney injury
Complication
Acceptable
Excellent
Excellent
Myocardial infarction
Poor
Poor
Poor
Cardiac arrest
Poor
Acceptable
Acceptable
Cerebrovascular accident
Outstanding
Outstanding
Outstanding
Deep vein thrombosis
Acceptable
Excellent
Acceptable
Pneumonia
Poor
Poor
Poor
Pulmonary embolism
Poor
Outstanding
Excellent
Dehiscence
Poor
Acceptable
Poor
Superficial surgical site infection
Poor
Acceptable
Acceptable
Sepsis
Acceptable
Excellent
Excellent
Septic shock
Acceptable
Outstanding
Excellent
Urinary tract infection
Poor
Acceptable
Acceptable
Excellent
Excellent
Excellent
Any minor complication
Poor
Acceptable
Poor
Any major complication
Poor
Acceptable
Acceptable
Any complication
Poor
Acceptable
Poor
Death
*Poor, c < 0.70; acceptable, c ¼ 0.70e0.80; excellent, c ¼ 0.80e0.90; outstanding, c > 0.90.
applicable to future reimbursement policy. As is well established, bundled payment plans without risk adjustment measures may be inequitable to providers and may restrict access to care for certain patients. To combat this situation, the Surgical Hip and Femur Fracture Treatment model plans to adjust for comorbidities by separating patients into 3 cohorts based on diagnosis-related group (DRG) codes: DRG 482 (no complication or comorbidity), DRG 481 (complication or comorbidity), and DRG 480 (major complication or comorbidity).49 The list of comorbidities and complications that qualify a patient for inclusion in DRG 480 and 481 cohorts is extensive and can be found on the Centers for Medicare and Medicaid Web site.50 However, based on a study of current reimbursement data after surgical repair of hip and femur fractures by Cairns et al.,51 the CCI had a significantly stronger association with the level of reimbursement compared with using solely the DRG to
e8
www.SCIENCEDIRECT.com
determine reimbursement. Because this study shows a significant association between patient comorbidities and total hospital cost, the literature clearly shows a strong association between inpatient complications and total hospital cost, and the current shows that the Elixhauser model is superior to the CCI model in predicting inpatient complications, the Elixhauser model may be superior to both the DRG and CCI models for explaining reimbursement rates.52 Although more research is certainly needed to confirm this hypothesis and extend it to surgical procedures of the spine, Centers for Medicare and Medicaid should consider implementing the Elixhauser Comorbidity Index in place of either the DRG or CCI scores to optimize equitability in their reimbursements and avoid disincentivizing surgical treatment of patients with a higher comorbidity burden. In an era of ballooning costs, when U.S. health care expenditures accounted for nearly 18% of the annual GDP in 2017, this study highlights the strong association between inpatient complications and drastically increased resource use. Specifically, this study shows that in patients who experienced no inpatient complications postoperatively, the average length of stay was 1.8 days whereas those who experienced 1 complication averaged >11 days in the hospital after the procedure. A recent study by Virk et al.53 found that each extra day in the hospital after ACDF accrued an additional cost of $652. Based on these numbers, the patients in our study who experienced 1 postoperative complication averaged a cost increase >$24,000, more than doubling the mean cost of all patients undergoing ACDF ($18,622), as reported by Kalakoti et al.14 Given that 5.3% of patients in this study experienced 1 complication, this finding represents a substantial use of limited health care resources and highlights the important implication of this study: if comorbidity indexes can accurately predict high-risk patients and their potential complications preoperatively, there is potential for implementation of preventive measures, conserving hospital resources, and optimizing patient outcomes. The results of this study should be interpreted after taking into account several potential limitations. First, chronic medical conditions tend to be underreported in large administrative databases even when the number of diagnoses is not limited.54-56 Second, when ICD-9 codes are used to determine patient comorbidities, as in this study, coding errors cannot be completely avoided.57 Third, the NIS reports information only with respect to the patient’s index surgery and subsequent inpatient stay, and thus all complications occurring after discharge from the hospital were not captured. In addition, this study was retrospective and thus subject to recall bias. For example, in patients who experienced an inpatient complication, providers may have been more likely to accurately record all predisposing comorbid risk factors, whereas patients who experienced no complications may have had their comorbidities underreported, resulting in an overestimation of the predictive capabilities of the CCI and Elixhauser models. Also, the CCI and Elixhauser models in this study are limited in their generalizability to populations with different characteristics. Another limitation is that ICD-9 codes may not be readily available or easy to use in typical outpatient clinic settings, thereby potentially limiting their clinical applicability. With respect to the proposal of adding preoperative steroid use to the list of
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
Elixhauser comorbidities as mentioned in the Discussion section, the appropriate weighting of this variable and its potential to either positively or negatively affect the performance of the model in predicting the various other postoperative complications is uncertain and requires further investigation. The Elixhauser Comorbidity Index proved to be either excellent or outstanding in predicting 8 of the 15 complications investigated and performed significantly better than the CCI in predicting 6 of these 15 complications. Because postoperative complication risk prediction and avoidance has the potential to
Q3
comorbidities on Medicare diagnosis-related group reimbursement for adult spinal deformity surgery. Neurosurg Focus. 2017;43:E11.
REFERENCES 1. Song KJ, Choi BY. Current concepts of anterior cervical discectomy and fusion: a review of literature. Asian Spine J. 2014;8:531-539. 2. Sugawara T. Anterior cervical spine surgery for degenerative disease: a review. Neurol Med Chir (Tokyo). 2015;55:540-546. 3. Di Capua J, Somani S, Kim JS, et al. Elderly age as a risk factor for 30-day postoperative outcomes following elective anterior cervical discectomy and fusion. Global Spine J. 2017;7:425-431. 4. Weiler C, Schietzsch M, Kirchner T, Nerlich AG, Boos N, Wuertz K. Age-related changes in human cervical, thoracal and lumbar intervertebral disc exhibit a strong intra-individual correlation. Eur Spine J. 2012;21(suppl 6):S810-S818. 5. Pine M, Jordan HS, Elixhauser A, et al. Enhancement of claims data to improve risk adjustment of hospital mortality. JAMA. 2007;297:71-76. 6. Phan K, Wang N, Kim JS, et al. Effect of preoperative anemia on the outcomes of anterior cervical discectomy and fusion. Global Spine J. 2017;7: 441-447. 7. Narain AS, Hijji FY, Haws BE, et al. Impact of body mass index on surgical outcomes, narcotics consumption, and hospital costs following anterior cervical discectomy and fusion. J Neurosurg Spine. 2018;28:160-166. 8. Chughtai M, Sultan AA, Padilla J, et al. Postoperative stroke after anterior cervical discectomy and fusion in patients with carotid artery stenosis: a statewide database analysis. Spine J. 2019;19: 597-601. 9. Kapadia BH, Decker SI, Boylan MR, Shah NV, Paulino CB. Risk factors for cerebrospinal fluid leak following anterior cervical discectomy and fusion. Clin Spine Surg. 2019;32:E86-E90. 10. Zaki O, Jain N, Yu E, Khan SN. 30- and 90-day unplanned readmission rates, causes, and risk factors after cervical fusion: a single institution analysis. Spine (Phila Pa 1976). 2019;44:762-769. 11. Bovonratwet P, Fu MC, Tyagi V, et al. Incidence, risk factors, and clinical implications of postoperative hematoma requiring reoperation following anterior cervical discectomy and fusion. Spine (Phila Pa 1976). 2019;44:543-549. 12. Nunley PD, Mundis GM, Fessler RG, et al. Impact of case type, length of stay, institution type, and
improve patient outcomes and optimize the use of scarce hospital resources, the use of the Elixhauser Comorbidity Index to predict specific inpatient complications after elective ACDF should be considered. Early identification of patients at high risk for inpatient complications may allow for the implementation of intervention strategies to optimize modifiable risk factors. Future research should assess the performance of the Elixhauser Index in predicting long-term complications and optimize the model to more accurately predict the various complications for which it proved to be inadequate.
13. Daley J, Henderson WG, Khuri SF. Risk-adjusted surgical outcomes. Annu Rev Med. 2001;52:275-287. 14. Kalakoti P, Gao Y, Hendrickson NR, Pugely AJ. Preparing for bundled payments in cervical spine surgery: do we understand the influence of patient, hospital, and procedural factors on the cost and length of stay? Spine. 2019;44:334-345. 15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383. 16. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8-27. 17. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45: 613-619. 18. Menendez ME, Neuhaus V, Ring D. Inpatient mortality after orthopaedic surgery. Int Orthop. 2015;39:1307-1314.
survival using administrative health data. Cancer. 2011;117:1957-1965. 25. Grendar J, Shaheen AA, Myers RP, et al. Predicting in-hospital mortality in patients undergoing complex gastrointestinal surgery: determining the optimal risk adjustment method. Arch Surg. 2012; 147:126-135. 26. Chu YT, Ng YY, Wu SC. Comparison of different comorbidity measures for use with administrative data in predicting short- and long-term mortality. BMC Health Serv Res. 2010;10:140. 27. Kim CY, Sivasundaram L, LaBelle MW, Trivedi NN, Liu RW, Gillespie RJ. Predicting adverse events, length of stay, and discharge disposition following shoulder arthroplasty: a comparison of the Elixhauser Comorbidity Measure and Charlson Comorbidity Index. J Shoulder Elbow Surg. 2018;27:1748-1755. 28. Sivaganesan A, Zuckerman S, Khan I, et al. Predictive model for medical and surgical readmissions following elective lumbar spine surgery: A national study of 33,674 patients. Spine (Phila Pa 1976). 2019;44:588-600. 29. Dobran M, Marini A, Nasi D, et al. Risk factors of surgical site infections in instrumented spine surgery. Surg Neurol Int. 2017;8:212.
19. Menendez ME, Ring D, Harris MB, Cha TD. Predicting in-hospital mortality in elderly patients with cervical spine fractures: a comparison of the Charlson and Elixhauser Comorbidity Measures. Spine. 2015;40:809-815.
30. Seicean A, Seicean S, Alan N, et al. Preoperative anemia and perioperative outcomes in patients who undergo elective spine surgery. Spine. 2013;38: 1331-1341.
20. Farley JF, Harley CR, Devine JW. A comparison of comorbidity measurements to predict healthcare expenditures. Am J Manag Care. 2006;12:110-119.
31. Bronheim RS, Oermann EK, Cho SK, Caridi JM. Coagulation profile as a risk factor for 30-day morbidity following cervical laminectomy and fusion. Spine. 2018;43:239-247.
21. Poses RM, McClish DK, Smith WR, Bekes C, Scott WE. Prediction of survival of critically ill patients by admission comorbidity. J Clin Epidemiol. 1996;49:743-747.
32. Goyal A, Elminawy M, Kerezoudis P, et al. Impact of obesity on outcomes following lumbar spine surgery: a systematic review and meta-analysis. Clin Neurol Neurosurg. 2018;177:27-36.
22. Menendez ME, Neuhaus V, van Dijk CN, Ring D. The Elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat Res. 2014;472:2878-2886.
33. Dias CGP, Roberto BB, Basaglia L, Lenza M, Nicolau RJ, Ferretti M. Interaction of demographic factors with the results of the surgery for degenerative disease of the cervical spine: a retrospective evaluation. Rev Bras Ortop. 2017;52:575-581.
23. Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data. Med Care. 2004;42:355-360.
34. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9CM and ICD-10 administrative data. Med Care. 2005;43:1130-1139.
24. Lieffers JR, Baracos VE, Winget M, Fassbender K. A comparison of Charlson and Elixhauser comorbidity measures to predict colorectal cancer
35. Moore BJ, White S, Washington R, Coenen N, Elixhauser A. Identifying increased risk of readmission and in-hospital mortality using hospital
WORLD NEUROSURGERY -: e1-e10, - 2019
www.journals.elsevier.com/world-neurosurgery
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
e9
987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
ORIGINAL ARTICLE WILLIAM A. RANSON ET AL.
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
administrative data: the AHRQ Elixhauser Comorbidity Index. Med Care. 2017;55:698-705. 36. Available at: https://www.hcup-us.ahrq.gov/ toolssoftware/comorbidity/comindex2012-2015.txt. Accessed January 14, 2019. 37. Glassman SD, Hamill CL, Bridwell KH, Schwab FJ, Dimar JR, Lowe TG. The impact of perioperative complications on clinical outcome in adult deformity surgery. Spine. 2007;32:2764-2770. 38. Tseng VL, Yu F, Lum F, Coleman AL. Risk of fractures following cataract surgery in Medicare beneficiaries. JAMA. 2012;308:493-501. 39. Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64:749-759. 40. HCUP Methods Series Calculating National Inpatient Sample (NIS) Variances for Data Years 2012 and Later. Available at: https://www.hcup-us.ahrq.gov/ reports/methods/2015_09.jsp. Accessed December 31, 2018. Q4 Q5
41. Chitale R, Campbell PG, Yadla S, Whitmore RG, Maltenfort MG, Ratliff JK. International classification of disease clinical modification 9 modeling of a patient comorbidity score predicts incidence of perioperative complications in a nationwide inpatient sample assessment of complications in spine surgery. J Spinal Disord Tech. 2015;28:126-133. 42. Sayeed Z, Anoushiravani AA, Chambers MC, et al. Comparing in-hospital total joint arthroplasty outcomes and resource consumption among underweight and morbidly obese patients. J Arthroplasty. 2016;31:2085-2090. 43. Anoushiravani AA, Sayeed Z, Chambers MC, et al. Assessing in-hospital outcomes and resource utilization after primary total joint arthroplasty among underweight patients. J Arthroplasty. 2016; 31:1407-1412.
e10
www.SCIENCEDIRECT.com
ACDF: ELIXHAUSER VS. CHARLSON COMORBIDITY INDICES
44. De la Garza-Ramos R, Abt NB, Kerezoudis P, et al. Deep-wound and organ-space infection after surgery for degenerative spine disease: an analysis from 2006 to 2012. Neurol Res. 2016;38:117-123.
54. Powell H, Lim LL-Y, Heller RF. Accuracy of administrative data to assess comorbidity in patients with heart disease. J Clin Epidemiol. 2001;54: 687-693.
45. Echt M, De la Garza Ramos R, Nakhla J, et al. The effect of cigarette smoking on wound complications after single-level posterolateral and interbody fusion for spondylolisthesis. World Neurosurg. 2018;116:e824-e829.
55. Preen DB, Holman CDJ, Lawrence DM, Baynham NJ, Semmens JB. Hospital chart review provided more accurate comorbidity information than data from a general practitioner survey or an administrative database. J Clin Epidemiol. 2004;57: 1295-1304.
46. Nahhas CR, Hu K, Mehta AI. Incidence and risk factors of wound complications in long segment instrumented thoracolumbar spinal fusions: a retrospective study. J Spine Surg. 2018;4:233-240. 47. Alvarez AP, Demzik AL, Alvi HM, Hardt KD, Manning DW. Risk factors for postoperative urinary tract infections in patients undergoing total joint arthroplasty. Adv Orthop. 2016;2016:7268985.
56. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. Comorbidities, complications, and coding bias. Does the number of diagnosis codes matter in predicting in-hospital mortality? JAMA. 1992;267:2197-2203. 57. DeFrances CJ, Lucas CA, Buie VC, Golosinskiy A. 2006 national hospital discharge survey. Natl Health Stat Report. 2008;5:1-20.
48. Ranson WA, White SJW, Cheung ZB, et al. The effects of chronic preoperative steroid therapy on perioperative complications following elective posterior lumbar fusion. Global Spine J. 2018;8: 834-841. 49. Surgical Hip and Femur Fracture Treatment Model. 50. Draft ICD-10-CM/PCS MS-DRGv28 Definitions Manual. Available at: https://www.cms.gov/ icd10manual/fullcode_cms/P0370.html. Accessed May 4, 2019. 51. Cairns MA, Ostrum RF, Clement RC. Refining risk adjustment for the proposed CMS surgical hip and femur fracture treatment bundled payment program. J Bone Joint Surg Am. 2018;100:269-277.
Conflict of interest statement: J.M.C. has financial relationships with Zimmer Biomet (paid consultant). S.K.C. has financial relationships with Corentec (paid consultant), Globus (paid consultant), Medtronic (paid consultant), and Zimmer (paid consultant; research support). None of the remaining authors has any financial interests or affiliations with institutions, organizations, or companies relevant to the article. None of the authors received payment or support in kind for any aspect of the submitted work. Received 21 July 2019; accepted 17 October 2019 Citation: World Neurosurg. (2019). https://doi.org/10.1016/j.wneu.2019.10.102
52. Healy MA, Mullard AJ, Campbell DA, Dimick JB. Hospital and payer costs associated with surgical complications. JAMA Surg. 2016;151:823-830.
Journal homepage: www.journals.elsevier.com/worldneurosurgery
53. Virk SS, Phillips FM, Moffatt-Bruce SD, Khan SN. What factors influence reimbursement for 1 to 2 level anterior cervical discectomy and fusion procedures? Spine. 2019;44:E33-E38.
1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.
Available online: www.sciencedirect.com
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.10.102
FLA 5.6.0 DTD WNEU13573_proof 11 November 2019 2:53 pm ce
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160