Effect of Surgical Setting on Cost and Hospital Reported Outcomes for Single-Level Anterior Cervical Discectomy and Fusion

INTRODUCTION

Health care systems are shifting to valued-based models that require cost reduction while maintaining or improving patient-centered outcomes.^1–3 Value has been defined as patient health outcomes obtained relative to the cost.⁴ In an effort to shift to quality-based bundled payment models, the US Centers for Medicare and Medicaid Services (CMS) changed to a merit-based incentive payment system.³ Furthermore, health care spending is evaluated in the context of patient outcomes and patients' care-experience surveys.⁵

Hospitals and payers are using data-driven methods to identify and measure the extent of wasteful spending.⁶ To improve value and reduce costs, quantifiable metrics such as patient length of stay (LOS), in-hospital complications, and 30-day all-cause readmissions rates are being targeted by the hospitals as well as CMS.⁷ The Medicare Hospital Readmissions Reduction Program penalizes hospitals for having higher than expected 30-day readmission rates.^7,8 Because of the pressure to increase value while decreasing costs, institutions are interested in evaluating the factors that drive costs, complications, and readmissions.

Variations in costs and outcomes after spine surgery are shown to be associated with different treating subspecialties, hospital settings, geographic locations, and time of year.^9–14 Feinberg et al¹⁴ compared outcomes between teaching and nonteaching hospitals and found that teaching hospitals had longer hospital stays and increased costs. Other studies have identified differences in management and patient outcomes among neurosurgeons and orthopaedic surgeons.^10,11,15 Standardization of costs and outcomes across surgical subspecialties and hospital settings will improve the overall value of care provided. To date, the combined impact of hospital setting, surgical subspecialty, and operative characteristics on the cost and outcomes of anterior cervical discectomy and fusion (ACDF) has not been investigated. Our objective was to evaluate hospital-reported outcomes measures between tertiary hospitals (TH) and community hospitals (CH) within the same hospital system. We aimed to determine differences in LOS, complications, readmissions, and costs for elective single-level ACDF procedures performed at tertiary versus community hospitals.

METHODS

RESULTS

A total of 301 patients (60 TH, 241 CH) were included in the study. Table 1 details differences in baseline patient characteristics based on hospital settings. The patient populations treated at the 3 different hospital settings were similar with respect to age, sex, race, comorbidities, and insurance status. Patients at CHs were more likely to be obese than patients at the TH (24% versus 10%; P = .01). Relative to neurosurgeons, orthopaedic surgeons were more likely to be the operating surgeons at the TH (31% versus 5%; P < .0001). Instrumentation plate was more commonly utilized at the TH (88% versus 57%; P < .0001) whereas cages were more common at CHs (68% versus 25%; P < .0001) (Table 2).

The CHs had significantly longer LOS compared to the TH (1.25 ± 0.50 versus 1.08 ± 0.28 days, P = .01). There were no differences in complications and readmissions rates between hospital settings (Table 3). There were no complications reported during hospital stays. One person was readmitted at a CH due to acute postoperative neck pain. All of the patients from the TH were discharged home whereas 14% of patients from CHs were discharged to a rehabilitation facility (P = .0004). Multivariate regression for LOS showed that CHs, orthopaedic subspecialty, female sex, and myelopathy were predictors for longer LOS (Table 4).

Overall, TH total costs were significantly higher when compared to CHs ($17 171 versus $11 737; Δ$ = $5434 ± 3996; P < .0001). For the CHs, the drug, patient room, supply, and therapy costs were significantly higher (Table 5). The largest net difference between the TH and CHs was observed for operating room costs ($6987 ± 2103 versus $3121 ± 799; Δ$ = $3866 ± 1942; P < .0001), followed by other costs ($1819 ± 1029 versus $295 ± 433; Δ$ = $1524 ± 854; P < .0001). The higher operating room costs were largely due to higher fixed costs at the TH ($4667 ± 1646 versus $2158 ± 609; Δ$ = $2509 ± 1356; P < .0001). The other costs were also largely due to higher fixed costs ($1544 ± 773 versus $193 ± 262; Δ$ = $1351 ± 682; P < .0001), but also due to higher IONM costs ($1421 ± 314 versus $96 ± 296; Δ$ = $1325 ± 609; P < .0001) at the TH. The laboratory costs were also significantly greater at the TH ($1361 ± 1378 versus $129 ± 164; Δ$ = $1231 ± 798; P < .0001), despite the TH ordering 1.4 fewer laboratory tests than the CHs (mean, 4.2 ± 4.3 versus 5.6 ± 5.5 tests; P = .032). The fixed costs were responsible for 49.5%–88.6% of the cost differences across cost categories and accounted for 72.6% of the difference in total costs between the TH and CHs. The multivariate regression model for the cost of performing an ACDF showed the TH, orthopaedic surgeon specialty, presence of myelopathy, and a longer LOS were associated with higher costs. The use of an instrumentation plate and discharge to home were associated with lower costs (Table 6).

DISCUSSION

Optimizing the value of health care requires institutions to constantly evaluate resource utilization and clinical outcomes. Better clinical outcomes with efficient resource utilization lead to higher value of care. To improve value, hospital systems are merging with or acquiring other hospitals to build larger hospital systems that span both urban and suburban geographies. The larger hospital systems aim to utilize economies of scale to decrease cost and improve clinical outcomes.¹⁹ Our study investigated the costs and outcomes of elective single-level ACDFs in a large hospital system across 2 different types of hospital settings: tertiary and community hospitals.

Comparing the 2 hospital settings highlighted key differences that can be targeted for further optimization of care and resource utilization. While similar patient populations were treated in both settings, patients in CHs had a longer hospital stay and were more likely to get discharged to a rehabilitation facility. The cost analysis showed that single-level ACDFs performed at the TH were significantly more costly than those performed at the CHs. Differences in operating room, other (including IONM), and laboratory costs were the primary drivers of the difference observed.

The complexity of patients can often dictate the clinical outcomes. Our analysis did not find any differences in patient characteristics. While the TH is located in an urban setting, the percentage of patients with Medicare/Medicaid or workers compensation was not different between the 2 hospital settings. These findings contrast with reports showing that THs treat more complex patients with lower access to healthcare.^20–22 The populations in our 2 hospital settings were uneven (241 CH versus 60 patients in TH), which could have masked potential differences. Additionally, the analysis only included elective single-level ACDF patients. The surgeons at our hospital system have established guidelines for ideal surgical candidates. Therefore, the similar patient cohorts could be a function of pre-established guidelines for surgical candidates within the hospital system. The similarity of the patient groups, however, is valuable to the outcomes analysis. The outcomes—LOS, complications, readmissions, and costs—can be directly compared against each other without concern for confounding due to differences in patient characteristics.

Hospital systems and governing bodies such as CMS evaluate complications associated with a procedure to derive the quality and value of services provided. LOS, complications, and readmissions are 3 metrics critical to the value-based health care equation. CHs had significantly longer hospital stays and discharged more patients to rehabilitation facilities relative to the TH. Multivariate analysis showed that orthopaedic subspecialty, CHs, not using an instrumentation plate, and myelopathy were independent risk factors for a longer LOS. While each of these variables was significantly associated with a longer LOS, the mean difference of a 0.17-days (4.1-hour) longer LOS at the CHs and beta weights ranging from 0.10 to 0.13 days (2.4 to 3.1 hours) are unlikely to be clinically important given they are less than 1 day. Myelopathy has not been reported in earlier studies as an independent risk factor for prolonged LOS.^23–26 Given the potential of neurological injury associated with myelopathy, providers likely observe patients longer to ensure that the neurologic exam is stable. Arnold et al²⁶ determined old age (>50 years), female sex, and cardiac, urinary, and pulmonary complications were predictors for prolonged LOS after ACDF. Other factors such as perioperative nutrition status have also shown to be associated with a prolonged LOS.²⁷ While a greater proportion of CH patients were obese, obesity was not a predictor of LOS in the multivariate analysis, consistent with prior research.²⁸ Finally, the use of a plate was associated with a 0.13-day (3.1-hour) shorter LOS. This is consistent with prior research showing a marginally shorter (1.4-hour) LOS in patients undergoing plate fixation for single-level ACDFs.²⁹ The multivariate regression in our study only accounted for 24% of the variability in LOS. Therefore, other factors not directly evaluated in the study could have also contributed to longer LOS at CHs. In addition to the quantifiable factors, individual physician choices, ancillary staff training, and patient characteristics could all impact patient LOS.

The single most important factor that could influence peri-operative and postoperative outcomes for patients is the surgeon performing the surgery. Our study evaluated ACDFs performed by both orthopaedic surgeons and neurosurgeons. We found that ACDFs performed by an orthopaedic surgeon were associated with longer LOS; however, this finding does not necessarily implicate orthopaedic surgeons as the cause of longer LOS. Additionally, the 0.11-day (2.6-hour) LOS difference for orthopaedic surgeons is unlikely clinically important. Hijji et al¹⁵ similarly showed no significant differences in LOS for patients undergoing single-level ACDFs by orthopaedic and neurosurgeons. The same disease can be managed very differently by different subspecialties.^9–12 Studies have compared the clinical decision making of orthopaedic surgeons and neurosurgeons and found that factors such as surgeon age, geographic location, and experience can all influence clinical decision making.^9,10,12 Therefore, the impact of surgical subspecialty on LOS is variable and it is one of the many factors that influences LOS.

In addition to LOS, complications and readmissions are also evaluated by CMS to determine the value of health care. No complications were reported and only 1 readmission was reported for the patients in our cohort. The complication and readmission rates are lower than those reported in prior literature.^30–32 The low complication and readmission rates in our study are largely due to coding according to the CMS provider-preventable conditions and 30-day all-cause, all-payer readmissions to the same hospital. Since we aimed to determine the effect of hospital setting on hospital-reported outcomes that affect hospital reimbursement, our complication and readmission rates may not be directly compared to studies with different criteria for identifying complication and readmission rates. For example, dysphagia is a potential complication of ACDF and would be reportable in a study evaluating ACDF complications. CMS, however, categorizes those complications that require a return to the operating room during the same hospital stay (a provider-preventable condition) or those that would require a readmission and possible return to the operating room within 30 days (a readmission). Both events would significantly affect the cost of the perioperative course, and are reportable to CMS, affecting hospital reimbursement. Therefore, we expect lower complication and readmission rates in our study.

Standardization across hospital systems improves system-level outcomes and decrease costs. The costs of performing the same procedure in 2 different hospital setting were significantly different. Although there were no differences in complications and readmissions, the cost of an ACDF at the TH was nearly 1.5 times ($5434) higher than at the CHs. Operating room, other (including IONM), and laboratory costs were the biggest contributors to the cost differentials between the TH and CHs. The higher costs were largely due to the fixed costs associated with running a large tertiary referral center, where 49.5%–88.6% of the cost difference was attributed to higher fixed costs. The difference in fixed costs accounted for 64.9% of the operating room and 84.1% of the laboratory cost variability. Despite ordering 1.4 fewer tests at the TH, the laboratory costs were still significantly higher due to the overhead costs of running a large tertiary facility. The significantly higher other costs were due to higher fixed costs, but also due to greater use of IONM at the TH. The use of IONM is a surgeon preference and is likely used more in the TH where residents and fellows play a greater role in the surgery. Prior studies, however, have not shown a reduction in neurologic injury rates with routine use of IONM in patients undergoing ACDF.³³ Interestingly, the supply costs were not significantly different between hospital settings, nor were the implant costs, likely due to similar contracts with medical device companies within the single hospital system. Multivariate analysis showed the older age, TH, orthopaedic specialty, myelopathy, and LOS were associated with higher costs, while use of an instrumentation plate and home discharge were associated with lower costs. To reduce costs while maintaining quality, hospitals systems will need to focus on modifiable and nonmodifiable factors. This analysis highlights discharge destination, LOS, differences in subspecialty, and hospital settings as potential modifiable factors.

CONCLUSION

In summary, we compare hospital-reported outcomes and costs at a TH and CHs within a single medical system for patients undergoing primary, elective, single-level ACDF. Despite the different geographical locations of the 2 hospital settings (suburban versus urban), the patients had similar baseline demographics. Complication and readmission rates were similar, but LOS was significantly longer at CHs. The significantly longer LOS, however, is unlikely clinically important given it was less than 1 day. The cost analysis showed the cost of performing an ACDF was significantly higher at the TH, primarily due to differences in fixed costs associated with running the hospital. We identify modifiable factors such as surgical subspecialty, hospital setting, and discharge destination as potential targets to reduce hospital costs across the system.