Higher mortality and distinct complications after anterior cervical discectomy and fusion in dialysis patients, despite improvements over prior decade: A national inpatient study Authors: Mitchell K. Ng, Leonidas E. Mastrokostas, Paul G. Mastrokostas, William A. Green, Joshua Mathew, Gregorio Baek, Yasmine Eichbaum, Yulia Lee, Morgan Hitchner, Jonathan Dalton, Adam Fano, Alec Giakas, Rajendra Singh, Brendan Reitman, Afshin E. Razi, Daniel R. Fassett, Alan S. Hilibrand, Alexander R. Vaccaro, Christopher K. Kepler, Gregory D. Schroeder, Andrew P. Alvarez Published: 2026 ABSTRACT ABSTRACTIntroduction:Dialysis patients undergoing anterior cervical discectomy and fusion (ACDF) face elevated perioperative risk due to complex comorbidities. This study aimed to (1) identify associations between dialysis status and perioperative complications after ACDF, (2) quantify differences in length of stay (LOS), hospitalization costs, and discharge disposition, and (3) provide a contemporary national benchmark for outcomes in this high-risk population.Materials and Methods:The 2016–2022 National Inpatient Sample was queried for adult elective ACDF cases. Dialysis dependence was identified through International Classification of Diseases-10 codes for end-stage renal disease with hemodialysis or peritoneal dialysis. Demographics, comorbidities, complications, LOS, costs, and discharge disposition were compared. Multivariate logistic regression was adjusted for demographic, hospital, and surgical factors.Results:Of 374,975 weighted ACDF admissions, 1155 involved dialysis patients. Dialysis was independently associated with increased odds of cardiovascular complications (odds ratio [OR]: 2.06, 95% confidence interval [CI]: 1.38–3.07), mechanical complications (OR: 2.67, 95% CI: 1.46–4.90), sepsis (OR: 6.46, 95% CI: 2.98–13.98), nonroutine discharge (OR: 2.02, 95% CI: 1.53–2.68), and inpatient mortality (OR: 8.18, 95% CI: 3.22–20.78). Although absolute inpatient mortality was low in both the groups, it was significantly higher in dialysis patients (P < 0.001). The mean LOS was >3 times longer, and hospitalization costs were nearly double (P < 0.001).Conclusion:Dialysis dependence is associated with distinct complications, substantially prolonged LOS, higher costs, and greater likelihood of nonhome discharge following ACDF. Inhospital mortality remains significantly higher in dialysis patients. Our findings support targeted perioperative optimization and early discharge planning to improve outcomes in this population. INTRODUCTION End-stage renal disease (ESRD) represents a growing health burden in the United States, with over 800,000 individuals affected and the majority reliant on chronic hemodialysis or peritoneal dialysis for survival.[12] The systemic effects of ESRD including but not limited to vascular calcification, chronic inflammation, anemia, disturbances in calcium–phosphate metabolism, and impaired immunity, are well-recognized contributors to perioperative morbidity.[345] In surgical populations, dialysis dependence has been linked to higher rates of infection, cardiovascular complications, and hardware-related problems, along with extended recovery times.[678] Anterior cervical discectomy and fusion (ACDF) is among the most frequently performed procedures for cervical radiculopathy and myelopathy, with a generally favorable risk profile in the broader population.[910] However, the interplay between the physiologic demands of ACDF and the medical complexity of dialysis patients remains incompletely defined. While prior large-database studies have reported higher complication rates and longer length of stay (LOS) in dialysis-dependent patients, most have analyzed composite complication measures, without detailing the specific types of events most strongly associated with dialysis.[1112] Furthermore, the economic implications and discharge patterns in this population have received little attention, and inhospital mortality has rarely been examined in the context of these other outcomes.[13] This study was undertaken to address these gaps using a contemporary, nationally representative inpatient dataset. Specifically, we aimed to (1) determine whether dialysis dependence is independently associated with specific perioperative complications after ACDF; (2) compare LOS, hospital costs, and discharge disposition between dialysis and nondialysis patients; and (3) provide national benchmarks for resource utilization and outcomes in this high-risk group. We hypothesized that dialysis patients would experience higher rates of cardiovascular, mechanical, and infectious complications, longer hospitalizations, greater costs, and more frequent discharge to nonhome settings, but that inhospital mortality would remain similar to nondialysis patients, potentially due to careful surgical selection of only dialysis patients best able to tolerate ACDF. MATERIALS AND METHODS Data source This retrospective observational study used the National Inpatient Sample (NIS) from 2016 through 2022, maintained by the Healthcare Cost and Utilization Project (HCUP). The NIS is the largest publicly available all-payer inpatient database in the United States, capturing a 20% stratified sample of discharges from U.S. community hospitals, excluding rehabilitation and long-term acute care facilities.[1] As the data are de-identified and publicly available, this study was exempt from institutional review board approval. Cohort identification and exclusion criteria Adult patients (≥18 years) undergoing ACDF were identified using International Classification of Diseases, 10th Edition, Procedure Coding System (ICD-10-PCS) codes: 0RG10A0 (single-level ACDF, anterior approach, interbody device, open) and 0RG20A0 (multilevel ACDF, anterior approach, interbody device, open). Encounters with these codes in any procedure field were retained. We excluded records with missing or invalid demographic, hospital, or cost data; negative or zero LOS; nonelective admissions; procedures not performed on the day of admission; and records with payer coded as “other” or missing. Patients with a discharge disposition of “other” were excluded, and only routine or nonroutine discharges were analyzed. Dialysis status and comorbidities Dialysis dependence was identified by the presence of IInternational Classification of Diseases, 10th Edition, Procedure Coding System (ICD-10-PCS) codes codes N18.6 (ESRD) or Z99.2 (dependence on renal dialysis) in any diagnosis field. Comorbidities were assigned using HCUP Elixhauser definitions mapped from ICD-10-CM diagnosis codes. An Elixhauser Comorbidity Index (ECI) was calculated as the row-wise sum of individual comorbidity indicators. Complication definitions Perioperative complications were defined using ICD-10-CM/PCS codes and grouped into cardiovascular, respiratory, and neurological categories. Individual complications – including dysphagia, mechanical complications, sepsis, and inpatient mortality – were also captured. A binary variable representing the presence of one or more complications was created and labeled adverse events. Discharge disposition was classified as routine (discharge to home) or nonroutine (transfer to another short-term hospital, skilled nursing facility, intermediate care facility, or home health care). Cost adjustment Total hospital charges were converted to costs using HCUP hospital- and year-specific cost-to-charge ratios, and then adjusted to 2022 U.S. dollars using Consumer Price Index multipliers. Statistical analysis Survey weighting was applied to all analyses to produce national estimates using HCUP-recommended design variables for clustering and stratification.[1415] Demographic, hospital, and clinical characteristics were compared between dialysis and nondialysis patients using survey-weighted Chi-square tests for categorical variables and t-tests for continuous variables. Logistic regression models, adjusted for age, sex, race, income quartile, primary payer, hospital region, bed size, teaching status, ECI, and ACDF level (single vs. multilevel), were used to evaluate the association between dialysis status and each complication and the nonroutine discharge outcome. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs). Survey-weighted means and t-tests were used to compare adjusted costs and LOS between the groups. All analyses were conducted using R (version 4.4.0; R Project for Statistical Computing, Vienna, Austria). Statistical significance was set at P < 0.05. RESULTS Patient demographics A total of 376,130 weighted ACDF admissions were identified, including 1155 (0.31%) patients on dialysis. Dialysis patients were significantly older than nondialysis patients (62.47 vs. 58.39 years, P < 0.001). The proportion of single-level versus multilevel fusion did not differ significantly between the groups (P = 0.378). Dialysis patients were more frequently male (62.8% vs. 47.6%, P < 0.001) and more often identified as black (34.6% vs. 10.4%, P < 0.001), while nondialysis patients were predominantly White (78.1% vs. 46.3%, P < 0.001). Socioeconomic differences were also observed: dialysis patients were more likely to reside in the lowest income quartile (37.2% vs. 25.1%, P < 0.001). Most ACDF procedures in dialysis patients were performed at urban teaching hospitals (83.1% vs. 73.6%, P = 0.003) and large hospitals (63.2% vs. 48.5%, P < 0.001). Regional variation was noted, with a higher proportion of dialysis cases in the South (53.7% vs. 44.4%, P = 0.013). Medicare was the primary payer for most dialysis patients (84.0% vs. 37.9%, P < 0.001). The mean ECI was more than twice as high in dialysis patients (4.13 vs. 1.83, P < 0.001) [Table 1]. Table 1Demographic information for patients undergoing anterior cervical discectomy and fusionDemographicsNo dialysis, n (%)Dialysis, n (%) P Weighted (n)374,9751155Average age (years)58.3962.47 <0.001 Levels fused (%) Single-level108,605 (29.0)365 (31.6)0.378 Multilevel266,370 (71.0)790 (68.4)Sex (%) Male178,320 (47.6)725 (62.8) <0.001 Female196,655 (52.4)430 (37.2)Race (%) White293,040 (78.1)535 (46.3) <0.001 Black38,960 (10.4)400 (34.6) Hispanic25,680 (6.8)140 (12.1) Asian or Pacific5540 (1.5)n≤10* Native American2015 (0.5)n≤10* Other9740 (2.6)n≤10*Income quartile 0–25th percentile94,010 (25.1)430 (37.2) <0.001 26th–50th percentile99,830 (26.6)265 (22.9) 51st–75th percentile95,670 (25.5)265 (22.9) 76th–100th percentile85,465 (22.8)195 (16.9)Hospital teaching status (%) Rural12,680 (3.4)n≤10* 0.003 Urban nonteaching86,460 (23.1)180 (15.6) Urban teaching275,835 (73.6)960 (83.1)Hospital size (%) Small92,900 (24.8)130 (11.3) <0.001 Medium100,215 (26.7)295 (25.5) Large181,860 (48.5)730 (63.2)Hospital region (%) Northeast64,595 (17.2)120 (10.4) 0.013 Midwest71,715 (19.1)195 (16.9) South166,460 (44.4)620 (53.7) West72,205 (19.3)220 (19.0)Payer status (%) Medicare141,995 (37.9)970 (84.0) <0.001 Medicaid35,900 (9.6)n ≤10* Private insurance159,755 (42.6)130 (11.3) Self-pay4140 (1.1)n ≤10* Other33,185 (8.8)n ≤10*Mean Elixhauser Index1.83 (1.52)4.13 (1.68) <0.001 *Signifies that the count number is <10 and cannot be reported. Bold values are statistically significant. n – Number Demographic information for patients undergoing anterior cervical discectomy and fusion *Signifies that the count number is <10 and cannot be reported. Bold values are statistically significant. n – Number Prevalence of comorbidities Dialysis patients had markedly higher rates of congestive heart failure (26.0% vs. 2.4%, P < 0.001), mild liver disease (9.1% vs. 2.2%, P < 0.001), and neurological disorders (9.1% vs. 1.6%, P < 0.001). Severe renal failure was present in all dialysis patients by definition (100% vs. 0.4%, P < 0.001). No significant differences were observed for chronic lung disease, obesity, or depression [Table 2]. Table 2Prevalence of comorbiditiesComorbiditiesNo dialysis (n=374,975), n (%)Dialysis (n=1155), n (%) P Congestive heart failure8885 (2.4)300 (26.0) <0.001 Chronic lung disease72,425 (19.3)265 (22.9)0.154Mild liver disease8355 (2.2)105 (9.1) <0.001 Neurological disorders5960 (1.6)105 (9.1) <0.001 Severe renal failure1400 (0.4)1155 (100.0) <0.001 Obesity74,820 (20.0)220 (19.0)0.728Depression64,810 (17.3)150 (13.0)0.080Bold values are statistically significant. n - Number Prevalence of comorbidities Bold values are statistically significant. n - Number Complications and discharge disposition In unadjusted comparisons, dialysis patients experienced higher rates of cardiovascular complications (14.7% vs. 3.4%, P < 0.001), respiratory complications (P < 0.001), neurological complications (8.7% vs. 1.9%, P < 0.001), mechanical complications (4.8% vs. 1.4%, P < 0.001), and sepsis (4.8% vs. 0.2%, P < 0.001). The proportion experiencing at least one complication was elevated in dialysis patients (49.4% vs. 38.7%, P < 0.001). Nonroutine discharge was substantially more common among dialysis patients (51.9% vs. 17.0%, P < 0.001). Dysphagia rates were similar between the groups (P = 0.529) [Table 3]. Inpatient mortality in the dialysis cohort was rare but significantly different from the nondialysis group (P < 0.001). Table 3Complications in patients undergoing anterior cervical discectomy and fusionComplicationsNo dialysis (n=374,975), n (%)Dialysis (n=1155), n (%) P Cardiovascular12,815 (3.4)170 (14.7) <0.001 Respiratory1435 (0.4)n ≤10* <0.001 Neurological7280 (1.9)100 (8.7) <0.001 Dysphagia31,430 (8.4)110 (9.5)0.529Mechanical complications5420 (1.4)55 (4.8) <0.001 Sepsis710 (0.2)55 (4.8) <0.001 Adverse events145,130 (38.7)570 (49.4) 0.001 Discharge disposition63,650 (17.0)600 (51.9) <0.001 Inpatient mortality440 (0.1)n ≤10* <0.001 *Signifies that the count number is <10 and cannot be reported. Bold values are statistically significant. Cardiovascular complications=Transfusion, venous thromboembolism, acute posthemorrhagic anemia, and myocardial infarction; Respiratory complications=Pneumonia and pulmonary embolism; Neurological complications=Cerebrovascular accident, nervous system complication and CSF leak/dural tear; Adverse events=1 or more complications. n - Number; CSF - Cerebrospinal fluid Complications in patients undergoing anterior cervical discectomy and fusion *Signifies that the count number is <10 and cannot be reported. Bold values are statistically significant. Cardiovascular complications=Transfusion, venous thromboembolism, acute posthemorrhagic anemia, and myocardial infarction; Respiratory complications=Pneumonia and pulmonary embolism; Neurological complications=Cerebrovascular accident, nervous system complication and CSF leak/dural tear; Adverse events=1 or more complications. n - Number; CSF - Cerebrospinal fluid Multivariable regression analysis After adjusting for demographic, hospital, and surgical factors, dialysis remained independently associated with increased odds of cardiovascular complications (OR: 2.06, 95% CI: 1.38–3.07, P < 0.001), mechanical complications (OR: 2.67, 95% CI: 1.46–4.90, P = 0.001), sepsis (OR: 6.46, 95% CI: 2.98–13.98, P < 0.001), nonroutine discharge (OR: 2.02, 95% CI: 1.53–2.68, P < 0.001), and inpatient mortality (OR: 8.18, 95% CI: 3.22–20.78, P < 0.001). Dysphagia was less likely after adjustment (OR: 0.62, 95% CI: 0.40–0.96, P = 0.033). No significant association was observed with respiratory complications, neurological complications, or adverse events [Table 4]. Table 4Regression modeling predicting the influence of dialysis on inpatient complicationsComplicationsOR (95% CI) P Cardiovascular2.06 (1.38–3.07) <0.001 Respiratory1.77 (0.68–4.59)0.240Neurological1.52 (0.92–2.52)0.188Dysphagia0.62 (0.40–0.96) 0.033 Mechanical complications2.67 (1.46–4.90) 0.001 Sepsis6.46 (2.98–13.98) <0.001 Adverse events1.12 (0.86–1.46)0.396Discharge disposition2.02 (1.53–2.68) <0.001 Inpatient mortality8.18 (3.22–20.78) <0.001 Bold values are statistically significant. Cardiovascular complications=Transfusion, venous thromboembolism, acute posthemorrhagic anemia, and myocardial infarction; Respiratory complications=Pneumonia and pulmonary embolism; Neurological complications=Cerebrovascular accident, nervous system complication, and CSF leak/dural tear; Adverse events=1 or more complications. CI - Confidence interval; OR - Odds ratio; CSF - Cerebrospinal fluid Regression modeling predicting the influence of dialysis on inpatient complications Bold values are statistically significant. Cardiovascular complications=Transfusion, venous thromboembolism, acute posthemorrhagic anemia, and myocardial infarction; Respiratory complications=Pneumonia and pulmonary embolism; Neurological complications=Cerebrovascular accident, nervous system complication, and CSF leak/dural tear; Adverse events=1 or more complications. CI - Confidence interval; OR - Odds ratio; CSF - Cerebrospinal fluid Costs and length of stay Mean adjusted hospital costs were significantly higher for dialysis patients ($45,962 vs. $25,075, P < 0.001). LOS was more than triple in dialysis patients (7.04 vs. 2.06 days, P < 0.001) [Table 5]. Table 5Costs and length of stay for patients undergoing anterior cervical discectomy and fusionOutcomesNo dialysis (n=374,975)Dialysis (n=1155) P Costs, mean ($)25,075.3045,962.00<0.001LOS, mean (days)2.067.04<0.001n - Number; LOS - Length of stay Costs and length of stay for patients undergoing anterior cervical discectomy and fusion n - Number; LOS - Length of stay DISCUSSION This study aimed to assess the impact of dialysis dependence status on the perioperative course of patients undergoing ACDF. Using a large, national cohort, we found that dialysis patients were more likely to develop cardiovascular, mechanical, and infectious complications. These complications were accompanied by markedly longer hospitalizations, substantially higher inpatient costs, and a greater need for discharge to facilities rather than home. The demographic profile of our dialysis cohort differed in ways that likely contribute to these results. As expected, patients receiving dialysis were, on average, older and had a greater burden of comorbid conditions, particularly diabetes, congestive heart failure, and peripheral vascular disease. This is in line with current literature. A single-institution retrospective review of 1605 dialysis-dependent patients likewise found dialysis patients were older (61.2 vs. 52.7 years, P < 0.001) and increased Charlson Comorbidity Index (3.55 vs. 0.41, P < 0.001).[11] Insurance coverage also differed, with dialysis patients more often insured through Medicare, and they were overrepresented in the lowest income quartiles. These factors, combined with the higher prevalence of black patients in the dialysis group, reflect national ESRD epidemiology and point toward persistent socioeconomic and racial disparities.[16] Such socioeconomic differences may potentially influence postoperative recovery trajectories, access to rehabilitation services, and the likelihood of requiring nonhome discharge. Our complication profile analysis provides a more comprehensive understanding of risk than prior work. A retrospective study of 1317 patients with ESRD and/or dialysis-dependent undergoing elective ACDF revealed that dialysis patients had higher overall complication rates, longer hospital stays, and increased mortality.[12] In contrast, our study separates complications into cardiovascular, mechanical, and infectious categories, providing a more actionable framework for targeted perioperative strategies. Specifically, in our study, by isolating cardiovascular, mechanical, and infectious complications, our results align with the known pathophysiology of ESRD: vascular calcification and hemodynamic instability may predispose to cardiovascular events; renal osteodystrophy and reduced bone quality may contribute to hardware failure; and immune dysfunction increases infection susceptibility.[171819] When taken together, our findings suggest that dialysis patients undergoing ACDF may potentially benefit from targeted perioperative strategies such as preoperative cardiac optimization, meticulous intraoperative technique to reduce construct-related complications, and heightened infection surveillance. The financial and systems-level implications of our data are also notable. Average hospitalization costs for dialysis patients were almost twice those of their nondialysis counterparts ($46,000 vs. $25,100, P < 0.001), driven largely by longer LOS (7.0 vs. 2.1 days) and higher rates of postacute care facility use. These are also consistent with the literature. A database study by De la Garza Ramos et al. found a similar three-fold increase in LOS for patients with ESRD relative to those without kidney disease (4.7 vs. 1.5 days, P < 0.001).[12] Another PearlDiver database study of 1750 patients with ESRD undergoing ACDF found that such patients were at significantly higher odds of intensive care unit admission (12 vs. 1%, P < 0.001) and readmissions (38% vs. 3%).[13] In an era of bundled payments and value-based care, these differences represent a tangible challenge for hospitals and payers. Early rehabilitation planning and proactive coordination with skilled nursing or rehabilitation facilities may help reduce downstream costs without compromising outcomes. The increased nonhome discharge rate of dialysis patients relative to nondialysis patients likewise highlights the importance of considering functional outcomes alongside surgical safety. Prior work has noted prolonged recovery in this population, but discharge disposition has received limited attention.[121318] Incorporating prehabilitation programs and multidisciplinary perioperative pathways could improve functional recovery and increase the likelihood of returning home after surgery. Perhaps the most striking finding was the low rate of absolute inhospital mortality despite the higher complication burden. This contrasts with other surgical specialties, where dialysis dependence often predicts higher perioperative mortality. A 2016 retrospective study of patients with ESRD undergoing inpatient ACDF found a significant increase in inpatient mortality between patients undergoing ACDF.[12] Our study provides novel 10-year updated information that this difference may be diminishing. The decreased absolute inpatient mortality in our analysis relative to previous studies likely reflects several changes in practice over the past decade. Perioperative management for dialysis patients has improved, with greater attention to preoperative optimization, improved control of volume status, and closer coordination of dialysis sessions around surgery.[2021] Surgeons may also be more selective, offering ACDF only to those dialysis patients with adequate cardiovascular reserve and manageable comorbidities.[2223] Furthermore, advances in anesthetic care and surgical technique, including shorter operative times and lower blood loss, may further reduce perioperative stress.[242526] Together, these developments may explain why our cohort demonstrated higher complication rates but low rates of inpatient mortality. Of note, when taking a conservative approach and assuming only 10 inpatient deaths among dialysis patients, the odds of mortality remained significantly higher than for nondialysis patients. This suggests that although the absolute rate of death has declined compared with earlier studies, dialysis dependence still carries a meaningful increase in perioperative risk. This study has several limitations inherent to its design. Use of the NIS relies on administrative coding, which is subject to miscoding and may not capture all relevant clinical details.[2728] The database does not provide granular operative information such as surgical level, fusion technique, graft type, or intraoperative complications, nor does it track important dialysis-specific variables such as timing and adequacy of treatments, nutritional status, or laboratory parameters. In addition, outcomes are limited to the index hospitalization and inpatient stay, so we were unable to assess long-term survival, readmission, or delayed complications. The number of inpatient deaths among dialysis patients was small, limiting the precision of mortality estimates. To avoid overinterpretation, we used a conservative assumption of 10 deaths in this group when assessing relative risk, as the NIS database only reports deaths as 10 or less but does not specify the exact figure. Residual confounding is possible despite multivariable adjustment, as certain comorbidities and social factors cannot be fully accounted for. Finally, while the large, nationally representative sample strengthens generalizability, our findings may not reflect outcomes at individual high-volume spine centers with specialized/unique perioperative protocols for dialysis patients. Nevertheless, despite these limitations, this study provides a 5+ year update and provides a more detailed complication profile and resource utilization assessment than previously available. CONCLUSION In a large, national, inpatient cohort, dialysis dependence was associated with a distinct pattern of complications after ACDF (cardiovascular, mechanical, and infectious) along with substantially longer hospital stays, higher inpatient costs, and a greater likelihood of nonhome discharge. Despite this increased morbidity and resource use, inhospital mortality was significantly different although still below 1%, suggesting that with careful patient selection and modern perioperative care, elective ACDF can be performed safely in selected dialysis patients. By identifying the specific complications most strongly linked to dialysis dependence, our findings expand upon prior work and offer practical targets for preoperative optimization, surgical planning, and discharge coordination. Efforts to reduce complication rates and streamline postacute care may improve both patient outcomes and the value of care delivered to this vulnerable population. Conflicts of interest There are no conflicts of interest.