1. Introduction
Cervical disc replacement (CDR) is a recently emerging alternative to anterior cervical discectomy and fusion (ACDF) for symptomatic cervical spondylosis. First described in the 1960s, CDR was developed with the notion that replacement with an artificial disc after neural decompression could preserve motion of the cervical spine, minimize the biomechanical stress on adjacent levels, and decrease the risk of subsequent adjacent segment disease and reoperation.1,2 Indeed, several series and mid-term results from randomized-controlled trials have demonstrated that CDR may prove to be an excellent surgical option when careful patient selection and well-trained, meticulous technique are able to be achieved.3, 4, 5, 6 Contrary to these positive findings, however, some studies have shown CDR to result in unexpectedly high rates of heterotopic ossification and loss of mobility, especially in two-level procedures.7,8 Needless to say, continued identification of patient populations that may benefit most from CDR over ACDF and a comprehensive understanding of the relative risks and benefits of the two procedures are necessary prior to complete adoption of this promising treatment.9
Due to the similarities in approach, CDR and ACDF share a similar short-term postoperative complication profile, which includes dysphagia, infection, dural tear, major vessel injury, Horner's syndrome, and recurrent laryngeal nerve injury.10 While the complication rates for each approach are well-described in the literature, associated outcomes of the index hospital stay, such as cost and length of stay (LOS), and associated readmission rates in the immediate postoperative period have been poorly characterized. Herein, we illustrate a comparison between CDR and ACDF regarding post-operative complications, readmission rates, hospital cost, and length-of-stay while identifying risk factors associated with readmission using a large, central-registry database representative of the U.S. population.
2. Materials and methods
2.1. Database
Admissions data were obtained from the 2015 and 2016 Healthcare Cost and Utilization Project (HCUP) Nationwide Readmissions Database (NRD). The NRD is a large scale database documenting patient hospitalizations over a calendar year. Data include diagnoses, procedures, and hospital factors.
2.2. Patient sample
15,418 patients undergoing single level ACDF or CDR were identified by ICD-10 procedural codes 0RG10A0 and 0RR30JZ respectively. 13,516 underwent ACDF and 1902 underwent CDR. Exclusion criteria included non-elective cases, patients under 18, death at the end of the index admission, and 2+ level cervical fusions (all ICD-10 0RG2 codes). In addition, cases were excluded if both ACDF and CDR were performed at the same index admission. Cases were included from the last quarter of 2015 to the end of 2016. Only the last quarter of 2015 was used because of the transition to ICD-10 starting in October 2015. We extracted patients from 14,519,781 total admissions from the 2016 NRD and 1,295,810 total admissions from the last quarter of the 2015 NRD.
2.3. Hospitalization data
For each index admission, we obtained baseline hospital, demographic, and comorbidity data. Hospital data included length of stay, billed cost, and discharge disposition. Demographic data were age, sex, and insurance status. The Elixhauser comorbidity index was used to define comorbidities on the basis of ICD 10 code groupings. Complications of clinical interest were also defined using ICD 10 code groupings. Readmissions within 30 and 90 days were calculated from the time of discharge at each index admission. Values of 10 or fewer were omitted as per HCUP reporting requirement.
2.4. Propensity score matching
Propensity score matching was performed to create comparable groups between patients undergoing ACDF and CDR. ACDF patients were matched to CDR patients on the basis of age, sex, cardiac arrythmia, congestive heart failure, chronic pulmonary disease, depression, diabetes and hypertension both with and without complication, hypothyroidism, liver disease, obesity, rheumatoid arthritis/collaged vascular disase, and renal failure. Logistic regression was used to generate propensity scores. Matching was performed using the nearest neighbors algorithm with a caliper of 0.001.
2.5. Statistical analysis
Univariate analyses comparing ACDF and CDR were performed using chi-squared and t-tests for categorical and continuous data respectively. After matching, logistic regression was used to generate odds ratios and 95 % confidence intervals comparing ACDF (reference) and CDR with regards to select complications. In addition, a multivariate regression was performed separately for each matched CDR and ACDF cohort to assess risk factors associated with 90 day readmission. Covariates included clinically significant comorbidities and complications with greater than 10 occurrences.
2.6. Software
Analyses were done using Python 3 and its open source statistical software SciPy (1.1.0) and statsmodels (0.11.1). Propensity score matching was performed in R (4.0.2) using the MatchIt (4.0.0) package, and comorbidities were defined using the R Comorbidity package (0.5.3).
3. Results
3.1. Baseline characteristics before propensity score matching
15,418 patients were included for analysis, 1902 (12.3 %) of whom underwent CDR and 13,516 (87.7 %) of whom underwent single-level ACDF. Baseline demographic characteristics and comorbidities prior to propensity score matching are shown in Table 1, Table 2, respectively. CDR patients were significantly younger, with 62.7 % of the cohort under 50 years old. CDR patients were significantly more likely to have private insurance (65.9 % vs 41.3 %) and less likely to have Medicare (10.9 % vs 35.6 %) compared with ACDF patients. A comparison of baseline comorbidities between the two cohorts showed that CDR patients were generally healthier. A significantly smaller proportion of CDR patients had comorbidities such as carotid artery disease, congestive heart failure, chronic pulmonary disease, depression, diabetes, hypertension, hypothyroidism, obesity, renal failure, and rheumatoid arthritis, among others.
Table 1.
Baseline characteristics of unmatched CDR versus ACDF cohorts.
| Characteristics | Total | CDR | ACDF | p |
|---|---|---|---|---|
| Age, n (%) | <0.001 | |||
| 18–49 years, n (%) | 5770 (37.4) | 1193 (62.7) | 4577 (33.9) | |
| 50–59 years, n (%) | 4590 (29.8) | 499 (26.2) | 4091 (30.3) | |
| 60–69 years, n (%) | 3271 (21.2) | 181 (9.5) | 3090 (22.9) | |
| 70–79 years, n (%) | 1499 (9.7) | 29 (1.5) | 1470 (10.9) | |
| 80 + years, n (%) | 288 (1.9) | a | 288 (2.1) | |
| Sex | 0.238 | |||
| Male, n (%) | 7356 (47.7) | 932 (49) | 6424 (47.5) | |
| Female, n (%) | 8062 (52.3) | 970 (51) | 7092 (52.5) | |
| Payer Status | <0.001 | |||
| Medicaid, n (%) | 1557 (10.1) | 142 (7.5) | 1415 (10.5) | |
| Medicare, n (%) | 5016 (32.5) | 207 (10.9) | 4809 (35.6) | |
| Private Insurance, n (%) | 6840 (44.4) | 1254 (65.9) | 5586 (41.3) | |
| Self-Pay, n (%) | 180 (1.2) | 29 (1.5) | 151 (1.1) | |
| Other/Unknown, n (%) | 1809 (11.7) | 269 (14.1) | 1540 (11.4) | |
| No Charge, n (%) | 16 (0.1) | a | 15 (0.1) | |
| Discharge Disposition | <0.001 | |||
| Routine, n (%) | 14,044 (91.1) | 1816 (95.5) | 12,228 (90.5) | |
| Home Health Care, n (%) | 1035 (6.7) | 80 (4.2) | 955 (7.1) | |
| Transfer to Skilled Nursing Facility, n (%) | 307 (2.0) | a | 303 (2.2) | |
| Transfer to Short-Term Hospital, n (%) | 12 (0.1) | a | 12 (0.1) | |
| Against Medical Advice, n (%) | 20 (0.1) | a | 18 (0.1) |
Value is below NRD reporting minimum of 10 or fewer cases.
Table 2.
Baseline comorbidities of unmatched CDR versus ACDF cohorts.
| Comorbidities | CDR | ACDF | p |
|---|---|---|---|
| AIDS/HIV, n (%) | a | a | 0.896 |
| Alcohol Abuse, n (%) | 17 (0.9) | 142 (1.1) | 0.608 |
| Carotid Artery Disease, n (%) | 35 (1.8) | 571 (4.2) | <0.001 |
| Congestive Heart Failure, n (%) | a | 245 (1.8) | <0.001 |
| Coagulopathy, n (%) | 11 (0.6) | 140 (1.0) | 0.076 |
| Chronic Pulmonary Disease, n (%) | 231 (12.1) | 2322 (17.2) | <0.001 |
| Deficiency Anemia, n (%) | a | 71 (0.5) | 0.900 |
| Depression, n (%) | 239 (12.6) | 2096 (15.5) | 0.001 |
| Diabetes (complicated), n (%) | 28 (1.5) | 512 (3.8) | <0.001 |
| Diabetes (uncomplicated), n (%) | 147 (7.7) | 2032 (15.0) | <0.001 |
| Drug Abuse, n (%) | 27 (1.4) | 240 (1.8) | 0.307 |
| Fluid and Electrolyte Disorders, n (%) | 11 (0.6) | 252 (1.9) | <0.001 |
| Hypertension (complicated), n (%) | 16 (0.8) | 368 (2.7) | <0.001 |
| Hypertension (uncomplicated), n (%) | 474 (24.9) | 5708 (42.2) | <0.001 |
| Hypothyroidism, n (%) | 121 (6.4) | 1345 (10.0) | <0.001 |
| Liver Disease, n (%) | 17 (0.9) | 170 (1.3) | 0.213 |
| Obesity, n (%) | 216 (11.4) | 2195 (16.2) | <0.001 |
| Other Neurologic Disorder, n (%) | 34 (1.8) | 399 (3.0) | 0.005 |
| Paralysis, n (%) | a | 115 (0.9) | 0.019 |
| Pulmonary Circulation Disorders, n (%) | a | 42 (0.3) | 0.598 |
| Peripheral Vascular Disorders, n (%) | a | 238 (1.8) | <0.001 |
| Renal Failure, n (%) | 17 (0.9) | 366 (2.7) | <0.001 |
| Rheumatoid Arthritis, n (%) | 36 (1.9) | 432 (3.2) | 0.002 |
| Valvular Disease, n (%) | 15 (0.8) | 221 (1.6) | 0.007 |
| Elixhauser Comorbidity Index | |||
| Elixhauser Score, mean (SD) | 0.91 (1.12) | 1.51 (1.45) | <0.001 |
Value is below NRD reporting minimum of 10 or fewer cases.
3.2. Baseline characteristics after propensity score matching
According to the 1:1 propensity score matching performed in this analysis, 1844 patients were matched in the CDR and ACDF cohorts (Table 3). Matching eliminated significant differences that existed between the two cohorts regarding age and most all comorbidities (Table 4). However, payer status and comorbidities such as drug abuse, psychoses, and weight loss remained significantly different.
Table 3.
Baseline characteristics of matched CDR versus ACDF cohorts.
| Characteristics | Total | CDR | ACDF | p |
|---|---|---|---|---|
| Age, n (%) | 0.874 | |||
| 18–49 years, n (%) | 2262 (61.3) | 1136 (61.6) | 1126 (61.1) | |
| 50–59 years, n (%) | 999 (27.1) | 498 (27.0) | 501 (27.2) | |
| 60–69 years, n (%) | 365 (9.9) | 181 (9.8) | 184 (10.0) | |
| 70–79 years, n (%) | 61 (1.7) | 29 (1.6) | 32 (1.7) | |
| 80 + years, n (%) | a | a | a | |
| Sex | 0.843 | |||
| Male, n (%) | 1797 (48.7) | 902 (48.9) | 895 (48.5) | |
| Female, n (%) | 1891 (51.3) | 942 (51.1) | 949 (51.5) | |
| Payer Status | <0.001 | |||
| Medicaid, n (%) | 405 (11.0) | 137 (7.4) | 268 (14.5) | |
| Medicare, n (%) | 533 (14.5) | 203 (11.0) | 330 (17.9) | |
| Private Insurance, n (%) | 2165 (58.7) | 1215 (65.9) | 950 (51.5) | |
| Self-Pay, n (%) | 63 (1.7) | 29 (1.6) | 34 (1.8) | |
| Other/Unknown, n (%) | 518 (14.0) | 259 (14.1) | 259 (14.1) | |
| No Charge, n (%) | a | a | a | |
| Discharge Disposition | 0.003 | |||
| Routine, n (%) | 3499 (94.9) | 1758 (95.3) | 1741 (94.4) | |
| Home Health Care, n (%) | 154 (4.2) | 80 (4.3) | 74 (4.0) | |
| Transfer to Skilled Nursing Facility, n (%) | 26 (0.7) | a | 22 (1.2) | |
| Transfer to Short-Term Hospital, n (%) | a | a | a | |
| Against Medical Advice, n (%) | a | a | a |
Value is below NRD reporting minimum of 10 or fewer cases.
Table 4.
Baseline comorbidities of matched CDR versus ACDF cohorts.
| Comorbidities | CDR | ACDF | p |
|---|---|---|---|
| AIDS/HIV, n (%) | a | a | 1.00 |
| Alcohol Abuse, n (%) | 14 (0.8) | 21 (1.1) | 0.308 |
| Carotid Artery Disease, n (%) | 29 (1.6) | 25 (1.4) | 0.681 |
| Congestive Heart Failure, n (%) | a | a | 0.627 |
| Coagulopathy, n (%) | a | 11 (0.6) | 0.823 |
| Chronic Pulmonary Disease, n (%) | 210 (11.4) | 218 (11.8) | 0.719 |
| Deficiency Anemia, n (%) | a | a | 0.422 |
| Depression, n (%) | 224 (12.1) | 227 (12.3) | 0.920 |
| Diabetes (complicated), n (%) | 25 (1.4) | 20 (1.1) | 0.549 |
| Diabetes (uncomplicated), n (%) | 140 (7.6) | 127 (6.9) | 0.446 |
| Drug Abuse, n (%) | 26 (1.4) | 55 (3.0) | 0.002 |
| Fluid and Electrolyte Disorders, n (%) | 11 (0.6) | 21 (1.4) | 0.110 |
| Hypertension (complicated), n (%) | 16 (0.9) | a | 0.054 |
| Hypertension (uncomplicated), n (%) | 458 (24.8) | 460 (24.9) | 0.970 |
| Hypothyroidism, n (%) | 113 (6.1) | 107 (5.8) | 0.728 |
| Liver Disease, n (%) | 16 (0.9) | a | 0.054 |
| Obesity, n (%) | 199 (10.8) | 187 (10.1) | 0.554 |
| Other Neurologic Disorder, n (%) | 34 (1.8) | 46 (2.5) | 0.214 |
| Paralysis, n (%) | a | 11 (0.6) | 0.210 |
| Pulmonary Circulation Disorders, n (%) | a | a | 1.00 |
| Peripheral Vascular Disorders, n (%) | a | 16 (0.9) | 0.094 |
| Renal Failure, n (%) | 16 (0.9) | a | 0.152 |
| Rheumatoid Arthritis, n (%) | 27 (1.5) | 20 (1.1) | 0.378 |
| Valvular Disease, n (%) | 14 (0.8) | a | 0.403 |
| Elixhauser Comorbidity Index | |||
| Elixhauser Score, mean (SD) | 0.88 (1.11) | 0.89 (1.08) | 0.845 |
Value is below NRD reporting minimum of 10 or fewer cases.
3.3. Comparison of outcomes between CDR and ACDF
An initial examination of intraoperative and postoperative complications prior to matching showed that CDR patients suffered overall fewer complications. Rates of dural tear, dysphagia, genitourinary complications, postoperative infection, and respiratory complications were significantly lower in the CDR cohort (Table 5). However, these observations were not replicated after propensity score matching, which indicates that the fewer complications in the CDR cohort are perhaps attributable to the lower proportion of patients suffering the comorbidities described in Table 2. In addition, a greater proportion of CDR patients had routine discharges as compared to ACDF patients. This was true both before and after matching.
Table 5.
Intraoperative and postoperative complications of unmatched CDR versus ACDF cohorts.
| Complications | Before matching |
After matching |
||||
|---|---|---|---|---|---|---|
| CDR | ACDF | p | CDR | ACDF | p | |
| Dehiscence, n (%) | 20 (1.1) | a | 1.00 | a | a | 1.00 |
| Dural Tear, n (%) | a | 71 (0.5) | 0.046 | a | a | 0.148 |
| DVT, n (%) | a | a | 0.766 | a | a | 1.00 |
| Dysphagia, n (%) | 64 (3.4) | 599 (4.4) | 0.037 | 64 (3.4) | 53 (2.9) | 0.396 |
| Esophageal Injury, n (%) | a | a | 0.766 | a | a | 1.00 |
| GI Complications, n (%) | a | 23 (0.2) | 0.861 | a | a | 1.00 |
| GU Complications, n (%) | 33 (1.7) | 345 (2.6) | 0.038 | 32 (1.7) | 22 (1.2) | 0.217 |
| Hematoma, n (%) | a | 44 (0.3) | 0.813 | a | a | 1.00 |
| Intraoperative Implant Related, n (%) | a | 53 (0.4) | 0.508 | a | a | 0.578 |
| Neurologic Complications, n (%) | 36 (1.9) | 317 (2.3) | 0.249 | 35 (1.9) | 31 (1.7) | 0.709 |
| Postoperative Infection, n (%) | a | 35 (0.3) | 0.049 | a | a | 1.00 |
| Pulmonary Embolism, n (%) | a | a | 0.601 | a | a | 1.00 |
| Respiratory Complications, n (%) | a | 194 (1.4) | 0.002 | a | 14 (0.8) | 0.539 |
| Vascular Injury, n (%) | a | a | 0.586 | a | a | 1.00 |
| Vocal Cord Paralysis, n (%) | a | 17 (0.1) | 0.239 | a | a | 1.00 |
Value is below NRD reporting minimum of 10 or fewer cases.
CDR patients had significantly lower 30-day and 90-day readmission rates compared to ACDF patients: 1.6 % vs 2.2 %; 2.8 % vs 4.0 %, respectively (Table 6), but these differences were not significant after matching. CDR patients had significantly higher cost of inpatient stay compared to ACDF patients with mean differences of $17277.33 and $18,823.75 in pre- and post-matched comparisons respectively. However, CDR patients generally had a shorter length of stay in both unmatched (0.91 days ± 1.12 vs 1.51 days ± 1.45, p < 0.001) and matched (1.19 days ± 0.76 vs 1.36 days ± 2.78, p < 0.014) analyses. Finally, a matched logistic regression analysis to determine the odds of complications in CDR over ACDF showed that CDR did not have increased odds of any complication (Table 7).
Table 6.
Outcomes of matched CDR versus ACDF cohorts.
| Outcomes | Total | CDR | ACDF | p |
|---|---|---|---|---|
| Readmission – Before matching | ||||
| 30-day Readmission, n (%) | 406 (2.6) | 29 (1.5) | 377 (2.8) | 0.002 |
| 90-day Readmission, n (%) | 751 (4.9) | 52 (2.7) | 699 (5.2) | <0.001 |
| Readmission – After matching | ||||
| 30-day Readmission, n (%) | 71 (1.9) | 29 (1.6) | 42 (2.2) | 0.150 |
| 90-day Readmission, n (%) | 126 (3.4) | 52 (2.8) | 74 (4.0) | 0.057 |
| Cost - Before matching | ||||
| Hospital Cost (dollars), mean (SD) | 77702.45 (46127.27) | 60425.12 (46410.32) | <0.001 | |
| Cost - After matching | ||||
| Hospital Cost (dollars), mean (SD) | 77,698.54 (46252.88) | 58,875.79 (45822.83) | <0.001 | |
| Length-of-Stay - Before matching | ||||
| LOS (days), mean (SD) | 1.20 (0.82) | 1.60 (2.67) | <0.001 | |
| Length-of-Stay - After matching | ||||
| LOS (days), mean (SD) | 1.19 (0.76) | 1.36 (2.78) | 0.014 | |
Table 7.
Odds of complications in matched CDR vs ACDF cohorts.
| Complications | OR | 97.5 % CI | p |
|---|---|---|---|
| Dural Tear | 0.33 | 0.09–1.23 | 0.099 |
| Dysphagia | 1.20 | 0.83–1.73 | 0.346 |
| GI Complications | 1.50 | 0.25–8.99 | 0.657 |
| GU Complications | 1.46 | 0.85–2.53 | 0.173 |
| Hematoma | 1.00 | 0.29–3.46 | 1.00 |
| Intraoperative Implant-Related | 0.62 | 0.20–1.91 | 0.409 |
| Neurologic Complications | 1.13 | 0.70–1.84 | 0.620 |
| Respiratory Complications | 0.71 | 0.32–1.61 | 0.415 |
3.4. Multivariate analysis for risk factors associated with 90-day readmission
Table 8 describes the results of a matched multivariate regression analysis to identify risk factors associated with 90-day readmission in both the CDR and ACDF cohorts. Complicated diabetes was significantly associated with an increased odds of readmission in both the CDR (OR 4.80, 97.5 % CI 1.30–17.77, p = 0.019) and ACDF cohorts (OR 4.76, 97.5 % CI 1.29–17.61, p = 0.019). Rheumatoid arthritis was associated with an increased odds of readmission in only the CDR cohort (OR 4.54, 97.5 % CI 1.27–16.17, p = 0.02).
Table 8.
Risk factors associated with readmission in matched CDR vs ACDF cohorts.
| Risk Factors | OR | 97.5 % CI | p |
|---|---|---|---|
| CDR | |||
| Chronic Pulmonary Disease | 1.35 | 0.62–2.94 | 0.453 |
| Depression | 1.45 | 0.69–3.10 | 0.332 |
| Diabetes (complicated) | 4.80 | 1.30–17.77 | 0.019 |
| Diabetes (uncomplicated) | 1.48 | 0.58–13.77 | 0.407 |
| Dysphagia | 0.53 | 0.07–4.01 | 0.541 |
| GU Complications | 0.94 | 0.11–7.55 | 0.953 |
| Hypothyroidism | 0.48 | 0.11–2.06 | 0.326 |
| Hypertension (uncomplicated) | 1.23 | 0.65–2.32 | 0.523 |
| Neurologic Complications | 0.94 | 0.12–7.31 | 0.950 |
| Obesity | 1.11 | 0.48–2.58 | 0.806 |
| Rheumatoid Arthritis | 4.54 | 1.27–16.17 | 0.020 |
| ACDF | |||
| Chronic Pulmonary Disease | 0.59 | 0.25–1.38 | 0.221 |
| Depression | 1.36 | 0.71–2.61 | 0.353 |
| Diabetes (complicated) | 4.76 | 1.29–17.61 | 0.019 |
| Diabetes (uncomplicated) | 1.27 | 0.54–2.98 | 0.582 |
| Dysphagia | 1.04 | 0.25–4.40 | 0.960 |
| GU Complications | 2.59 | 0.57–11.74 | 0.218 |
| Hypothyroidism | 1.80 | 0.80–4.05 | 0.156 |
| Hypertension (uncomplicated) | 1.44 | 0.85–2.45 | 0.180 |
| Neurologic Complications | 0.54 | 0.07–4.36 | 0.561 |
| Obesity | 0.80 | 0.36–1.81 | 0.595 |
| Rheumatoid Arthritis | 1.25 | 0.16–9.70 | 0.829 |
4. Discussion
While utilization of CDR continues to increase, ACDF remains the most common procedure for cervical disc disease.11,12 Controversy still exists as to which is the ideal surgical method for treatment of cervical degenerative disc disease.13, 14, 15 Overall, CDR has been established as noninferior to ACDF.15, 16, 17, 18, 19 More recently, CDR has been found to offer greater outcomes in terms of neurological success, pain reduction, range of motion, and complication rates.12,13,16,20, 21, 22
The present study was a retrospective analysis using the NRD to compare the associated risks of single level ACDF and CDR procedures. Our results corroborate other recent literature which suggests CDR when indicated may be preferable to ACDF for single level procedures. Previous studies have also found CDR to be noninferior to ACDF, but confounding variables persist, such as those due to surgeon selection bias.15, 16, 17, 18, 19 Here, propensity score matching affirmed previous findings of CDR noninferiority.
4.1. Baseline characteristics
Surgeons take multiple factors into account when deciding to operate, including age and preexisting comorbidities. Prior to matching, several baseline characteristics were significantly different between ACDF and CDR cohorts. In agreement with previous studies, ACDF patients tended to be older with higher comorbidity burden.23,24 CDR tends to be performed on younger patients with decreased number of comorbidities,25, 26, 27 corroborated in our study. However, CDR utilization for older patients continues to increase as providers become more comfortable with the procedure.12 After matching, the prevalence of drug abuse, psychoses and weight loss were higher in the ACDF cohort. These may be key factors to consider in the management of ACDF vs CDR patients. Privately insured patients are more likely to be in the upper income and social classes. The impact of income and social strata may result in lower rates of these comorbid conditions. Furthermore, our results are in agreement with previous studies finding that patients with Medicaid are more likely to receive ACDF, while privately insured patients are more likely to receive CDR.14,25
4.2. Perioperative outcomes
LOS was longer for ACDF patients in both unmatched and matched analyses. Previous retrospective studies have shown decreased LOS associated with CDR.14,25,28,29 However recent meta-analyses have found no significant difference in hospital stay between procedures.13,20 Our results find a difference but one that is negligible to the patient and care facility. Hypertension, BMI, and diabetes have been associated with an increased LOS in CDR patients.30 Furthermore, our results found an increased proportion of unfavorable discharges among ACDF patients. This corroborates previous analyses25,28 and has been observed after controlling for patient baseline characteristics.11 We also found an increased cost of index procedure for CDR patients. This is in stark contrast to other studies in which CDR has repeatedly been found to be less costly for the index procedure and for all related costs well-past 90 days postoperatively.14,29,31 Several Markov analyses have found CDA to be more cost-effective in the long term.32, 33, 34 Differing results in our study may be due to the use of billed cost in this study.
4.3. Adverse outcomes
Prior to matching, our results found significantly higher rates of 30-day and 90-day readmission in ACDF relative to CDR patients. However, no significant differences were found after matching. This may be attributed to higher rates of comorbidity and older age in the baseline ACDF cohort. There is a paucity of literature that compares short term readmission rates between CDR and ACDF.29 However, having one or more comorbidities has been found to increase a patient's short term risk of readmission for either procedure.29 In this study, diabetes status was found to be a risk factor for 90-day readmission after either procedure, while rheumatoid arthritis was a significant factor for CDR only. Diabetes has been frequently associated with adverse outcomes, such as readmission, reoperation, infection, or prolonged LOS following ACDF and CDR.30,35, 36, 37 Our study is in agreement with prior literature. However, a recent study found no significant difference between diabetics and nondiabetics with regards to reoperation, 30- and 90-day readmission, and complication rates following ACDF.38 Thus, the association between diabetes and adverse outcomes may require further investigation. Rheumatoid arthritis was found to be a significant predictor of 90-day readmission in CDR but not ACDF patients. This was also corroborated by a 2013 NRD study by Rumalla et al. However, the authors analyzed ACDF and CDR in tandem,37 while in this study the two were analyzed separately after matching. Rheumatoid arthritis commonly affects the cervical spine and has been implicated as a contraindication to CDR.39, 40, 41 This may be a key factor to consider when deciding between ACDF or CDR. Overall, readmission rates at 30 and 90 days were relatively low, suggesting that either procedure when indicated based on patient risk factors may be suitable.
Additionally, there was no difference in postoperative or intraoperative complication risk between the procedures in our matched cohort analysis. The most common adverse event for both procedures was dysphagia. While some meta-analyses find a higher rate of dysphagia with ACDF vs CDR,16,22,42,43 other studies similar to our own do not find the difference reaching statistical significance.44,45 It is not yet clear whether the introduction of zero-profile fusion devices may reduce these rates46, 47, 48 in ACDF. Respiratory insufficiency, hematoma, vascular injury, surgical site infection, and dural tear are serious but uncommonly documented complications of ACDF.49,50
This study has several limitations to consider. First, we included data from the last quarters of 2015 and 2016. Consequently, not all readmissions may have been captured, as the NRD does not allow patients to be tracked past the calendar year. Second, a substantial portion of hospital admissions across the United States are not documented in the NRD, which may limit generalizability. Third, some admissions were missing from our original sample of the 2015 and 2016 NRDs due to technical processing issues, which may impact results.
5. Conclusions
This propensity score matched comparison between ACDF and CDR lends support to prior findings of CDR noninferiority. While ACDF patients tended to be sicker and older, similar rates of complications and 90-day readmission were found after propensity score matching. CDR patients also had shorter lengths of stay and a higher frequency of routine discharge, but greater index cost.
CRediT authorship contribution statement
Ravi S. Nunna: Writing - review & editing, Writing - original draft, Validation, Project administration, Methodology, Investigation, Conceptualization. James S. Ryoo: Writing - original draft, Formal analysis, Data curation. Philip B. Ostrov: Writing - review & editing, Writing - original draft. Saavan Patel: Writing - review & editing, Writing - original draft. Periklis Godolias: Visualization, Validation. Zeyad Daher: Writing - review & editing, Visualization, Validation. Richard Price: Writing - review & editing, Visualization. Jens R. Chapman: Writing - review & editing. Rod J. Oskouian: Writing - review & editing, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
- 1.Fernström U. Arthroplasty with intercorporal endoprothesis in herniated disc and in painful disc. Acta Chir Scand Suppl. 1966;357:154–159. [PubMed] [Google Scholar]
- 2.Puttlitz C.M., Rousseau M.A., Xu Z., Hu S., Tay B.K.-B., Lotz J.C. Intervertebral disc replacement maintains cervical spine kinetics. Spine. 2004;29:2809–2814. doi: 10.1097/01.brs.0000147739.42354.a9. [DOI] [PubMed] [Google Scholar]
- 3.Davis R.J., Nunley P.D., Kim K.D., et al. Two-level total disc replacement with Mobi-C cervical artificial disc versus anterior discectomy and fusion: a prospective, randomized, controlled multicenter clinical trial with 4-year follow-up results. J Neurosurg Spine. 2015;22:15–25. doi: 10.3171/2014.7.SPINE13953. [DOI] [PubMed] [Google Scholar]
- 4.Delamarter R.B., Zigler J. Five-year reoperation rates, cervical total disc replacement versus fusion, results of a prospective randomized clinical trial. Spine. 2013;38:711–717. doi: 10.1097/BRS.0b013e3182797592. [DOI] [PubMed] [Google Scholar]
- 5.Luo J., Huang S., Gong M., et al. Comparison of artificial cervical arthroplasty versus anterior cervical discectomy and fusion for one-level cervical degenerative disc disease: a meta-analysis of randomized controlled trials. Eur J Orthop Surg Traumatol. 2015;25(Suppl 1):S115–S125. doi: 10.1007/s00590-014-1510-4. [DOI] [PubMed] [Google Scholar]
- 6.Radcliff K., Coric D., Albert T. Five-year clinical results of cervical total disc replacement compared with anterior discectomy and fusion for treatment of 2-level symptomatic degenerative disc disease: a prospective, randomized, controlled, multicenter investigational device exemption clinical trial. J Neurosurg Spine. 2016;25:213–224. doi: 10.3171/2015.12.SPINE15824. [DOI] [PubMed] [Google Scholar]
- 7.Nunley P.D., Jawahar A., Kerr E.J., 3rd, et al. Factors affecting the incidence of symptomatic adjacent-level disease in cervical spine after total disc arthroplasty: 2- to 4-year follow-up of 3 prospective randomized trials. Spine. 2012;37:445–451. doi: 10.1097/BRS.0b013e31822174b3. [DOI] [PubMed] [Google Scholar]
- 8.Wu J.-C., Huang W.-C., Tsai H.-W., et al. Differences between 1- and 2-level cervical arthroplasty: more heterotopic ossification in 2-level disc replacement: clinical article. J Neurosurg Spine. 2012;16:594–600. doi: 10.3171/2012.2.SPINE111066. [DOI] [PubMed] [Google Scholar]
- 9.Ding D., Shaffrey M.E. Cervical disk arthroplasty: patient selection. Clin Neurosurg. 2012;59:91–97. doi: 10.1227/NEU.0b013e31826b6fbe. [DOI] [PubMed] [Google Scholar]
- 10.Leven D., Meaike J., Radcliff K., Qureshi S. Cervical disc replacement surgery: indications, technique, and technical pearls. Curr Rev Musculoskelet Med. 2017;10:160–169. doi: 10.1007/s12178-017-9398-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Saifi C., Fein A.W., Cazzulino A., et al. Trends in resource utilization and rate of cervical disc arthroplasty and anterior cervical discectomy and fusion throughout the United States from 2006 to 2013. Spine J. 2018;18:1022–1029. doi: 10.1016/j.spinee.2017.10.072. [DOI] [PubMed] [Google Scholar]
- 12.Niedzielak T.R., Ameri B.J., Emerson B., Vakharia R.M., Roche M.W., Malloy J.P., 4th Trends in cervical disc arthroplasty and revisions in the Medicare database. J Spine Surg. 2018;4:522–528. doi: 10.21037/jss.2018.09.04. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Xie L., Liu M., Ding F., Li P., Ma D. Cervical disc arthroplasty (CDA) versus anterior cervical discectomy and fusion (ACDF) in symptomatic cervical degenerative disc diseases (CDDDs): an updated meta-analysis of prospective randomized controlled trials (RCTs) SpringerPlus. 2016;5:1188. doi: 10.1186/s40064-016-2851-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kumar C., Dietz N., Sharma M., Wang D., Ugiliweneza B., Boakye M. Long-term comparison of health care utilization and reoperation rates in patients undergoing cervical disc arthroplasty and anterior cervical discectomy and fusion for cervical degenerative disc disease. World Neurosurg. 2020;134:e855–e865. doi: 10.1016/j.wneu.2019.11.012. [DOI] [PubMed] [Google Scholar]
- 15.Davis R.J., Kim K.D., Hisey M.S., et al. Cervical total disc replacement with the Mobi-C cervical artificial disc compared with anterior discectomy and fusion for treatment of 2-level symptomatic degenerative disc disease: a prospective, randomized, controlled multicenter clinical trial: clinical article. J Neurosurg Spine. 2013;19:532–545. doi: 10.3171/2013.6.SPINE12527. [DOI] [PubMed] [Google Scholar]
- 16.McAfee P.C., Reah C., Gilder K., Eisermann L., Cunningham B. A meta-analysis of comparative outcomes following cervical arthroplasty or anterior cervical fusion: results from 4 prospective multicenter randomized clinical trials and up to 1226 patients. Spine. 2012;37:943–952. doi: 10.1097/BRS.0b013e31823da169. [DOI] [PubMed] [Google Scholar]
- 17.Zigler J.E., Delamarter R., Murrey D., Spivak J., Janssen M. ProDisc-C and anterior cervical discectomy and fusion as surgical treatment for single-level cervical symptomatic degenerative disc disease: five-year results of a Food and Drug Administration study. Spine. 2013;38:203–209. doi: 10.1097/BRS.0b013e318278eb38. [DOI] [PubMed] [Google Scholar]
- 18.Boselie T.F.M., Willems P.C., van Mameren H., de Bie R.A., Benzel E.C., van Santbrink H. Arthroplasty versus fusion in single-level cervical degenerative disc disease: a Cochrane review. Spine. 2013;38:E1096–E1107. doi: 10.1097/BRS.0b013e3182994a32. [DOI] [PubMed] [Google Scholar]
- 19.Coric D., Nunley P.D., Guyer R.D., et al. Prospective, randomized, multicenter study of cervical arthroplasty: 269 patients from the Kineflex|C artificial disc investigational device exemption study with a minimum 2-year follow-up: clinical article. J Neurosurg Spine. 2011;15:348–358. doi: 10.3171/2011.5.SPINE10769. [DOI] [PubMed] [Google Scholar]
- 20.Gao F., Mao T., Sun W., et al. An updated meta-analysis comparing artificial cervical disc arthroplasty (CDA) versus anterior cervical discectomy and fusion (ACDF) for the treatment of cervical degenerative disc disease (CDDD) Spine. 2015;40:1816–1823. doi: 10.1097/BRS.0000000000001138. [DOI] [PubMed] [Google Scholar]
- 21.Rao M.-J., Nie S.-P., Xiao B.-W., Zhang G.-H., Gan X.-R., Cao S.-S. Cervical disc arthroplasty versus anterior cervical discectomy and fusion for treatment of symptomatic cervical disc disease: a meta-analysis of randomized controlled trials. Arch Orthop Trauma Surg. 2015;135:19–28. doi: 10.1007/s00402-014-2122-5. [DOI] [PubMed] [Google Scholar]
- 22.Findlay C., Ayis S., Demetriades A.K. Total disc replacement versus anterior cervical discectomy and fusion: a systematic review with meta-analysis of data from a total of 3160 patients across 14 randomized controlled trials with both short- and medium- to long-term outcomes. Bone Joint Lett J. 2018;100-B:991–1001. doi: 10.1302/0301-620X.100B8.BJJ-2018-0120.R1. [DOI] [PubMed] [Google Scholar]
- 23.Bhashyam N., De la Garza Ramos R., Nakhla J., et al. Thirty-day readmission and reoperation rates after single-level anterior cervical discectomy and fusion versus those after cervical disc replacement. Neurosurg Focus. 2017;42:E6. doi: 10.3171/2016.11.FOCUS16407. [DOI] [PubMed] [Google Scholar]
- 24.Qureshi S.A., Koehler S.M., Lu Y., Cho S., Hecht A.C. Utilization trends of cervical artificial disc replacement during the FDA investigational device exemption clinical trials compared to anterior cervical fusion. J Clin Neurosci. 2013;20:1723–1726. doi: 10.1016/j.jocn.2013.03.002. [DOI] [PubMed] [Google Scholar]
- 25.Lu Y., McAnany S.J., Hecht A.C., Cho S.K., Qureshi S.A. Utilization trends of cervical artificial disc replacement after FDA approval compared with anterior cervical fusion: adoption of new technology. Spine. 2014;39:249–255. doi: 10.1097/BRS.0000000000000113. [DOI] [PubMed] [Google Scholar]
- 26.Qureshi S.A., McAnany S., Goz V., Koehler S.M., Hecht A.C. Cost-effectiveness analysis: comparing single-level cervical disc replacement and single-level anterior cervical discectomy and fusion: clinical article. J Neurosurg Spine. 2013;19:546–554. doi: 10.3171/2013.8.SPINE12623. [DOI] [PubMed] [Google Scholar]
- 27.Nesterenko S.O., Riley L.H., 3rd, Skolasky R.L. Anterior cervical discectomy and fusion versus cervical disc arthroplasty: current state and trends in treatment for cervical disc pathology. Spine. 2012;37:1470–1474. doi: 10.1097/BRS.0b013e31824ee623. [DOI] [PubMed] [Google Scholar]
- 28.Upadhyayula P.S., Yue J.K., Curtis E.I., Hoshide R., Ciacci J.D. A matched cohort comparison of cervical disc arthroplasty versus anterior cervical discectomy and fusion: evaluating perioperative outcomes. J Clin Neurosci. 2017;43:235–239. doi: 10.1016/j.jocn.2017.04.027. [DOI] [PubMed] [Google Scholar]
- 29.Radcliff K., Zigler J., Zigler J. Costs of cervical disc replacement versus anterior cervical discectomy and fusion for treatment of single-level cervical disc disease: an analysis of the Blue Health Intelligence database for acute and long-term costs and complications. Spine. 2015;40:521–529. doi: 10.1097/BRS.0000000000000822. [DOI] [PubMed] [Google Scholar]
- 30.Zeidan M., Goz V., Lakomkin N., Spina N., Brodke D.S., Spiker W.R. Predictors of readmission and prolonged length of stay after cervical disc arthroplasty. Spine. 2021;46:487–491. doi: 10.1097/BRS.0000000000003839. [DOI] [PubMed] [Google Scholar]
- 31.Radcliff K., Lerner J., Yang C., Bernard T., Zigler J.E. Seven-year cost-effectiveness of ProDisc-C total disc replacement: results from investigational device exemption and post-approval studies. J Neurosurg Spine. 2016;24:760–768. doi: 10.3171/2015.10.SPINE15505. [DOI] [PubMed] [Google Scholar]
- 32.Ghori A., Konopka J.F., Makanji H., Cha T.D., Bono C.M. Long term societal costs of anterior discectomy and fusion (ACDF) versus cervical disc arthroplasty (CDA) for treatment of cervical radiculopathy. Int J Spine Surg. 2016;10:1. doi: 10.14444/3001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.McAnany S.J., Merrill R.K., Overley S.C., Kim J.S., Brochin R.L., Qureshi S.A. Investigating the 7-year cost-effectiveness of single-level cervical disc replacement compared to anterior cervical discectomy and fusion. Global Spine J. 2018;8:32–39. doi: 10.1177/2192568217726283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.McAnany S.J., Overley S., Baird E.O., et al. The 5-year cost-effectiveness of anterior cervical discectomy and fusion and cervical disc replacement: a Markov analysis. Spine. 2014;39:1924–1933. doi: 10.1097/BRS.0000000000000562. [DOI] [PubMed] [Google Scholar]
- 35.Phan K., Kim J.S., Lee N., Kothari P., Cho S.K. Impact of insulin dependence on perioperative outcomes following anterior cervical discectomy and fusion. Spine. 2017;42:456–464. doi: 10.1097/BRS.0000000000001829. [DOI] [PubMed] [Google Scholar]
- 36.Di Capua J., Somani S., Kim J.S., et al. Predictors for patient discharge destination after elective anterior cervical discectomy and fusion. Spine. 2017;42:1538–1544. doi: 10.1097/BRS.0000000000002140. [DOI] [PubMed] [Google Scholar]
- 37.Rumalla K., Smith K.A., Arnold P.M. Cervical total disc replacement and anterior cervical discectomy and fusion: reoperation rates, complications, and hospital resource utilization in 72 688 patients in the United States. Neurosurgery. 2018;82:441–453. doi: 10.1093/neuros/nyx289. [DOI] [PubMed] [Google Scholar]
- 38.Shuman W.H., Neifert S.N., Gal J.S., et al. The impact of diabetes on outcomes and health care costs following anterior cervical discectomy and fusion. Global Spine J. 2020 doi: 10.1177/2192568220964053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Auerbach J.D., Jones K.J., Fras C.I., Balderston J.R., Rushton S.A., Chin K.R. The prevalence of indications and contraindications to cervical total disc replacement. Spine J. 2008;8:711–716. doi: 10.1016/j.spinee.2007.06.018. [DOI] [PubMed] [Google Scholar]
- 40.Moatz B., Tortolani P.J. Cervical disc arthroplasty: pros and cons. Surg Neurol Int. 2012;3:S216–S224. doi: 10.4103/2152-7806.98582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Gillick J.L., Wainwright J., Das K. Rheumatoid arthritis and the cervical spine: a review on the role of surgery. Int J Rheumatol. 2015;2015 doi: 10.1155/2015/252456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Zhong Z.-M., Li M., Han Z.-M., et al. Does cervical disc arthroplasty have lower incidence of dysphagia than anterior cervical discectomy and fusion? A meta-analysis. Clin Neurol Neurosurg. 2016;146:45–51. doi: 10.1016/j.clineuro.2016.04.002. [DOI] [PubMed] [Google Scholar]
- 43.Gendreau J.L., Kim L.H., Prins P.N., et al. Outcomes after cervical disc arthroplasty versus stand-alone anterior cervical discectomy and fusion: a meta-analysis. Global Spine J. 2020;10:1046–1056. doi: 10.1177/2192568219888448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Liu F.-Y., Yang D.-L., Huang W.-Z., et al. Risk factors for dysphagia after anterior cervical spine surgery: a meta-analysis. Medicine. 2017;96:e6267. doi: 10.1097/MD.0000000000006267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Shi S., Zheng S., Li X.-F., Yang L.-L., Liu Z.-D., Yuan W. Comparison of 2 zero-profile implants in the treatment of single-level cervical spondylotic myelopathy: a preliminary clinical study of cervical disc arthroplasty versus fusion. PLoS One. 2016;11 doi: 10.1371/journal.pone.0159761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Sun Z., Liu Z., Hu W., Yang Y., Xiao X., Wang X. Zero-profile versus cage and plate in anterior cervical discectomy and fusion with a minimum 2 Years of follow-up: a meta-analysis. World Neurosurg. 2018;120:e551–e561. doi: 10.1016/j.wneu.2018.08.128. [DOI] [PubMed] [Google Scholar]
- 47.Vaishnav A.S., Saville P., McAnany S., et al. Predictive factors of postoperative dysphagia in single-level anterior cervical discectomy and fusion. Spine. 2019;44:E400–E407. doi: 10.1097/BRS.0000000000002865. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Alimi M., Njoku I., Hofstetter C.P., et al. Anterior cervical discectomy and fusion (ACDF): comparison between zero profile implants and anterior cervical plate and spacer. Cureus. 2016;8:e573. doi: 10.7759/cureus.573. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Tracey R.W., Kang D.G., Cody J.P., Wagner S.C., Rosner M.K., Lehman R.A., Jr. Outcomes of single-level cervical disc arthroplasty versus anterior cervical discectomy and fusion. J Clin Neurosci. 2014;21:1905–1908. doi: 10.1016/j.jocn.2014.05.007. [DOI] [PubMed] [Google Scholar]
- 50.Epstein N.E. A review of complication rates for anterior cervical diskectomy and fusion (ACDF) Surg Neurol Int. 2019;10:100. doi: 10.25259/SNI-191-2019. [DOI] [PMC free article] [PubMed] [Google Scholar]