Introduction

Improving clinical outcomes and reducing excessive costs are especially relevant in providing value-based care for complex adult spinal deformity (ASD) patients. According to current literature, complications after ASD surgery are frequent in the early (24–36%) and late (11–15%) postoperative settings [1]. These include medical and surgery-related complications (pseudarthrosis, proximal junctional kyphosis (PJK), wound complications, DVT/PE, infection, pneumonia, ileus) often resulting in costly readmissions which may be avoidable [2,3,4,5,6,7]. This has led to a number of studies investigating risk factors for unplanned readmissions, but much of this data is limited by a lack of spine-specific factors and rigorous complication data, single-surgeon data, and short-term follow-up periods [8,9,10,11,12].

In recent decades, there has been a growing focus on patient-reported outcomes (PROs) in ASD patients [13]. A multicenter database study found that ASD operative candidates scored lower in every domain of the Short Form-36 Health-Survey compared to other chronic conditions (arthritis, chronic lung disease, diabetes, and congestive heart failure) [14]. Although it is well known that surgery can significantly reduce disability, pain, and improve overall quality of life compared to nonoperative management, the impact of unplanned readmissions and reoperations on PROs in the complex ASD population remains unclear [15,16,17,18,19,20,21,22].

Compared to prior literature, our study focuses on complex ASD surgical patients and evaluates the clinical data and PROs with a minimum 2-year postoperative follow-up. The purpose of this study was to assess the impact of readmissions and reoperations on PROs. In addition, we sought to provide an in-depth analysis on underlying reasons and risk factors for readmissions occurring anytime in the first 2 years following surgery.

Materials and methods

In this study, we retrospectively reviewed a prospectively accrued data set of a consecutive group of complex ASD (≥ 18 years) surgeries performed between 2015 and 2018. These patients were treated by three experienced spinal deformity surgeons at a single-institution. “Complex ASD” cases had a primary or revision diagnosis of ASD and underwent ≥ 6-level fusion surgery with either posterior column osteotomy (PCO), pedicle subtraction osteotomy (PSO), vertebral column resection (VCR), pelvic fixation, and/or interbody fusion (e.g., TLIF). Common diagnoses included adult idiopathic scoliosis, degenerative lumbar scoliosis, congenital spine deformity, kyphoscoliosis, fixed sagittal imbalance, fixed coronal imbalance, flat back deformity, neuromuscular scoliosis, and Scheuermann’s kyphosis (Table 1). Patients were excluded if they had a diagnosis of spinal trauma, active infection, and spinal tumor. The minimum follow-up was 2 years after the index hospital discharge date. This study was approved by the institutional review board at Columbia University Medical Center.

Table 1 Common diagnoses for the primary adult spinal deformity population, N = 175
Full size table

Demographics included sex, age, American Society of Anesthesiologists grade (ASA), and Body Mass Index (BMI, kg/m2). Medical comorbidities included cardiac (myocardial infarction, hypertension, hyperlipidemia, coronary artery disease), pulmonary (asthma, chronic obstruction pulmonary disease, obstructive sleep apnea, prior pneumonia), gastrointestinal (GI) (gastroesophageal reflux disease/GERD, peptic ulcer disease, ulcerative colitis, celiac disease), osteoporosis/osteopenia, diabetes, hypothyroidism, anemia, depression, history of deep vein thrombosis/pulmonary embolism (DVT/PE), and history of cancer (non-spine tumor). Operative variables included total instrumented levels (TIL), pelvic fixation, osteotomy type (PCO, PSO, VCR), anterior lumbar interbody fusion (ALIF), oblique lumbar interbody fusion (OLIF), transforaminal lumbar interbody fusion (TLIF), estimated blood loss (EBL), operative/anesthesia durations, and prior spine surgery.

Intraoperative/postoperative complications, which occurred during the same inpatient stay, included intraoperative dural tear, intraoperative motor/sensory loss, GI (ileus, nausea/vomiting), cardiac (cardiac/respiratory arrest, lower extremity edema, hypovolemic shock, pericardial effusion), pulmonary (respiratory distress/failure, pneumonia, pleural effusion), DVT/PE, hyponatremia, neurologic (motor deficit), infection (skin, urinary tract infection), acute kidney injury, and postoperative red blood cell transfusion.

Any readmission/reoperation within 2 years after the index hospital date of discharge was assessed. Reasons for readmission were reviewed carefully in each chart review by an orthopedic surgeon independent of the primary treatment team. χ2/Fisher’s exact test and t-tests/ANOVA were used for categorical and continuous variables, respectively. To determine the independent predictors for the outcomes of interest, stepwise multivariate logistic regression analysis was used. Statistical significance was defined as p value < 0.05. The C-statistic and Hosmer–Lemeshow (HL) value were used to measure concordance and goodness-of-fit for the final models. SAS Studio Version 3.4 (SAS Institute Inc, Cary, NC) was used for all statistical analyses.

PROs that included both the Scoliosis Research Society-22R (SRS-22R) and the Oswestry Disability Index (ODI) were recorded in both the preoperative and up to the 2-year postoperative period. For those with readmissions, the most recent PRO data were taken after the readmission discharge date. Several prior studies have demonstrated that both the SRS-22R and the ODI are reliable, valid, and responsive to change in patients undergoing adult spinal deformity surgery [23,24,25]. There are various proposed methods to calculate the minimum clinically important difference (MCID) for PRO’s. In this study, we used a difference of 0.5 times the standard deviation, which has been used in prior literature and known to be equivalent to 1-standard error of measurement for a reliability of 0.75 [26, 27].

Source of funding

No source of funding was provided for this study.

Results

A total 175 consecutive patients met inclusion criteria. The mean follow-up ± standard deviation was 2.5 ± 0.5 years (range 2–3.8 years). Percent follow-up for the 2-year postoperative period was 71.3%. The mean age was 52.6 ± 16.4 years and 72% were female. The two most common preoperative diagnoses included adult idiopathic scoliosis (51.4%) and fixed sagittal imbalance (42.3%) (Table 1). 40.6% had a prior spine surgery. The mean TIL was 13.3 ± 4.1 (range 6–25), with the majority of patients having pelvic fixation (76.6%), PCOs (84.6%, mean 5.4/patient), and TLIF (69.1%, mean 1.4/patient). A smaller number of patients had a 3-column osteotomy (3CO) [13.7% VCR (8.0%; mean 1.5/patient), PSO (5.7%; mean 1.0/patient)] and ALIF/OLIF (2.3%; mean 1.5/patient). Overall mean operative time was 473 ± 137 min with mean EBL of 1324 (24.6%) ± 822 mL (16.4%) (Table 2).

Table 2 Operative characteristics of ASD patients
Full size table

The 2-year readmission and reoperation rates were 16.6% and 12.0%, respectively (Table 3). The median number of days after index discharge date resulting in a readmission and reoperation were 173 days (range 6–716) and 227 days (range 16–716), respectively. Eight patients who were readmitted did not require a reoperation. These readmissions without reoperation were due to headache from a dural tear (1.7%), muscle spasms (0.6%), abdominal pain (0.6%), and poorly controlled pain (1.7%). Reoperations were due to pseudarthrosis (5.1%), PJK (4.0%), wound complication (2.3%), upper and/or lower extremity weakness (2.3%), prominence over incision (1.1%), postoperative fall (1.1%), radiculopathy (0.6%) (Table 4). For the two patients with prominence over incision, the wound closure was well healed; however, both had prominent spinous processes causing localized pain requiring excision.

Table 3 Breakdown of readmissions and follow up after ASD surgery
Full size table
Table 4 Top complications requiring readmission (patients had ≥ 1 complication)
Full size table

In the multivariate analysis for 2-year readmissions, GI comorbidity (OR: 12.6, 95%CI 3.1–50.7), TLIF (OR: 10.1, 95%CI 2.0–50.6), VCR (OR: 8.8, 95%CI 1.7–44.6), pulmonary comorbidity (OR: 5.4, 95%CI 1.7–17.1), depression (OR: 4.4, 95%CI 1.4–13.8), and cardiac comorbidity (OR: 3.3, 95%CI 1.2–9.2) were independent risk factors and demonstrated good model performance (C-statistic = 0.88; HL = 0.1). Of note, 82.4% of the GI complications were in patients with a history of GERD, and 85.3% of the pulmonary comorbidities were in patients with asthma/COPD. Prior spinal surgery increased readmission risk by 2.5-fold, but was not statistically significant (p = 0.097). Although age, PSO, operative time, postoperative transfusion, and postoperative inpatient neurologic complications were significant in the bivariate analysis, they were not statistically significant factors in the multivariate analysis (Tables 5, 6).

Table 5 Patient clinical characteristics by readmission
Full size table
Table 6 Independent predictors for 2-year readmissions after complex ASD Surgery (C-stat = 0.876, HL = 0.1)
Full size table

Inpatient complications (intraoperative and postoperative) were assessed to determine potential differences in PROs in those with and without complications (Table 7). Patients with “any” inpatient complication had worse ODI scores at baseline and at 1-year postoperative. However, no significant differences were seen by 2 years postoperative. For SRS, no significant differences were seen at baseline, 1 year, and 2 years postoperative for either any inpatient complications, intraoperative complications, or inpatient-postoperative complications. 2 patients required revision surgery for lower extremity pain/weakness, but neither sustained permanent neurologic deficit. The 2-year MCID PROs (SRS and ODI) were not significantly different between those with and without inpatient complications (Table 8).

Table 7 Intraoperative and inpatient postoperative complications *patients had ≥ 1 complication
Full size table
Table 8 Patient reported outcomes by inpatient complications
Full size table

2 years after the index surgery, 86.4% and 68.2% of all patients reached MCID for SRS and ODI, respectively. At 1-year follow-up, patients who did not require a readmission had significantly better PROs compared to preoperative values (baseline PRO: SRS 59.9, ODI 33.4; 1-year PRO: SRS 85.9, ODI 18.0; p < 0.001) as well as those who required either a readmission (1-year SRS 78.5, p < 0.001; 1-year ODI 20.3; p < 0.001) or reoperation (1-year SRS 71; p < 0.001; 1-year ODI 23.8; p < 0.001) (Table 9). At 2 years postoperative, those who required a readmission had significant increases in SRS and reductions in ODI (2-year: SRS 90.6, ODI 15.5) compared to 1-year (SRS 78.5, ODI 20.3; p < 0.001) and preoperative values (SRS 60.1, ODI 44.9; p < 0.001). A similar significant difference was seen for revised patients as well (2-year Revised: SRS 90.9, ODI 13.4; 1-year Revised: SRS 71, ODI 23.8; p < 0.001; Baseline: SRS 56.6, ODI 45.6; p < 0.001). At 2 years postoperative, a substantial number of readmitted patients achieved MCID for SRS (100%) and ODI (83.3%). The same was seen for revised cases: MCID SRS 100%; ODI 85.7%. For those without readmission/reoperation, MCID was achieved in 83.4% for SRS and 64.9% for ODI. No statistically significant differences between those with and without readmission/reoperation were observed for 2-year PRO’s: SRS (readmission: 90.6 vs. non-readmission: 87.4; p = 0.586) or ODI (readmission: 15.5 vs. 16.9; p = 0.834) | SRS (reoperation: 90.9 vs. non-reoperation: 87.4; p = 0.585) or ODI (reoperation: 13.4 vs. non-reoperation: 16.9; p = 0.583).

Table 9 Patient reported outcomes by readmissions/reoperations
Full size table

Significant improvements occurred in each SRS domain score, but the largest changes occurred in the pain and self-image domains. Similarly, the largest significant changes for ODI occurred in pain, standing, sleeping, and traveling domains (Figs. 1, 2). By 2 years postoperative, patients who experienced a transient complication versus a permanent complication had a significantly higher total SRS score (89.2 vs. 67.7; p = 0.016), higher MCID percentage (85.7% vs. 67.7%; p < 0.001), and a higher ODI MCID (66.7% vs. 15.4%; p = 0.001) (Table 10).

Fig. 1
figure 1

A comparison of the mean SRS-22R domain scores by follow-up period (preoperative baseline, 1 year, 2 years) for patients who had a readmission

Full size image
Fig. 2
figure 2

A comparison of the mean ODI domain scores by follow-up period (preoperative baseline, 1 year, 2 years) for patients who had a readmission

Full size image
Table 10 Patient reported outcomes by complications
Full size table

The findings of this study may be best represented by patient-AB, a 69-year-old female with severe thoracolumbar posttraumatic osteoporotic kyphosis (> 100°) (Fig. 3). Medical comorbidities included depression, osteoporosis, and GERD. Preoperatively, she suffered from severe back pain and difficulty breathing and eating because of her severe kyphotic posture. Her preoperative SRS/ODI was 69/40. Surgery involved posterior spinal instrumented fusion from T1-sacrum/ilium, multiple PCOs (8), T11 VCR, and L5-S1 TLIF, with an EBL of 800 mL, and operative time of 545mins(Fig. 3). Her hospital course was complicated by pleural effusion and acute blood loss anemia requiring blood transfusion, and a 9-day LOS. Post-discharge, she otherwise did well, until 2 years postoperative when she complained of new onset lower back pain with X-rays demonstrating rod fractures at L4-5(Fig. 4). She was readmitted to undergo revision instrumentation from L1-pelvis with repair of the L4-5 pseudarthrosis (Fig. 4). At the most recent follow-up visit (2 years after revision surgery), she was doing very well without major issues. Her postoperative SRS/ODI scores after revision at latest follow-up were 100/18, respectively.

Fig. 3
figure 3

The preoperative anterior–posterior (a) and lateral (c) radiographs show patient AB with severe thoracolumbar osteoporotic kyphosis (> 100°). Her surgery involved posterior spinal instrumented fusion from T1 to pelvis, multiple PCOs (8), T11 VCR, TLIF L5-S1 as shown in (b) and (d)

Full size image
Fig. 4
figure 4

Patient AB was readmitted for new onset lower back pain and found to have rod fractures at L4-5 (a, c). She underwent revision instrumentation from L1 to pelvis with repair of the pseudarthrosis at L4-5 (b, d)

Full size image

Discussion

ASD is an increasingly common condition that is known to cause substantial pain and disability. Corrective surgery can improve health-related quality of life (HRQoL) outcomes; however, complications still occur requiring unplanned readmissions. Several large multicenter studies have reported on ASD outcomes, but there remains a paucity of literature on the impact of postoperative complications on PROs [28,29,30]. Furthermore, ASD patients are often defined by > 2 spinal fusion levels and/or radiographic parameters [6, 8, 31,32,33]. Within these cohorts, patients may undergo a range of procedures with more complex ones [e.g., extended fusions (≥ 6), 3COs] at greater risk for significant complications. The purpose of this study was to examine the impact of reoperation/readmission on HRQoL for complex ASD patients defined as those who had a minimum 6-level spinal fusion. We hypothesized that patients who experienced treatable postoperative medical or surgical complications would show significant improvement in HRQoL by 2 years despite the need for readmission/reoperation.

Based on our single-institutional analysis, the overall 2-year readmission (16.6%) and reoperation (12%) rates were relatively low given the surgical complexity involved [34]. The majority of reoperations were for pseudarthrosis (5.1%) and PJK (4%). By 2 years postoperative, the majority of patients reached MCID for SRS (86.4%) and ODI (68.2%). Patients who did not require a readmission had significantly higher PROs compared to preoperative values and to those who required a readmission or reoperation at 1-year follow-up. Those who required a readmission had significant increases in SRS and reductions in ODI at 2 years compared to 1-year and preoperative baseline values. For readmitted patients, the main drivers for improvement in the SRS-22r from 1 to 2 years were the pain (+ 1.0; p < 0.001) and the self-image (+ 0.7; p < 0.001) domains. In regards to the ODI scores, the main drivers of change from 1 to 2 years were pain (− 0.5; p = − 0.022), standing (− 0.7; p = 001), and sleeping (− 0.5; p = 0.026). The magnitude of the change appears small, but a single point can mean the difference between someone who could stand “as long as I want with extra pain” versus “pain prevents me from standing for more than 1 h.” Finally, TLIF was also found to be associated with readmission in the multivariate analysis (OR: 10.1, 95%CI 2.0–50.6). This could potentially be explained by the significantly higher rate of intraoperative complications associated with TLIFs (TLIF: 35.5% vs. No TLIF: 20.4%, p = 0.045) and specifically intraoperative durotomies (TLIF: 28.9 vs. No TLIF: 11.1, p = 0.01). Although there were higher rates for pseudarthrosis (8.3 vs 7.4, p = 0.847), postop extremity weakness (2.5 vs. 1.9, p = 0.798), pain requiring readmission (4.1 vs. 3.7, p = 0.894), wound complications (2.5% vs. 1.9, p = 0.798), and implant failure (5.0 vs. 3.7, p = 0.714), these were not statistically significant. Nearly every TLIF was performed with pelvic fixation (94.2%). Those with TLIF and pelvic fixation had a lower rate of pseudarthrosis (8.8% vs. 15%; p = 0.385) and higher rate of PJK (6.1% vs. 0%; p = 0.499) than those with pelvic fixation alone. However, these differences were also not statistically significant.

Few prior studies directly examined the impact of complications on PROs in the complex ASD population, and the results are somewhat conflicting. In a single-center retrospective study, Riley et al. [35] examined HRQoL in complex ASD patients based on Scoli-RISK-1 (SR-1) criteria with minimum 2-year follow-up. Their ASD population was considerably more complex given that nearly 40% of patients underwent a 3CO and 23.4% of patients suffered a major complication. Nevertheless, significant improvements were observed in all SRS-22r domains, and more than 50% achieved MCID by 2 years postoperative. Similar to our study, the greatest improvements occurred in SRS pain and self-image domains. However, patients with postoperative neurological deficit or a major complication were unlikely to achieve MCID for the SRS function domain. It is possible that patients with permanent complications are less likely to experience substantial improvements in HRQoL. Auerbach et al. [36] studied outcomes in patients who underwent 3COs for ASD and showed that patients with permanent major complications seemed to have lower mean satisfaction rates, but the difference was not statistically different from those with transient complications. By 2 years, patients who experienced major complications were still able to achieve satisfactory clinical outcomes. In contrast, Glassman et al. demonstrated that major complications negatively impact PROs for ASD patients compared to those with only minor or no complications. This study was limited to a 1-year postoperative follow-up, which may not be sufficient time to account for potentially recoverable complications [37]. In our study, we found that those with “any inpatient” complication had worse ODI at 1 year, but the differences were not significant at 2 years. Furthermore, the 2-year MCID% was not significantly different between those with and without “any inpatient” complication, “intraoperative” complication, and “inpatient postoperative” complication for either SRS or ODI. These prior studies provide valuable information, but did not examine the impact of revision surgery or readmission on PROs.

In a prospective multicenter study, Passias et al. [34] reviewed the readmission and reoperation data in an ASD population defined by radiographic parameters. They reported a 22.8% readmission rate and 19.5% reoperation rate. Similar to our findings, a major reason for revision surgery was implant failure. HRQoL analysis revealed an overall improvement in the total population but less improvement in those who were readmitted. Undergoing reoperation following readmission did not have any impact on HRQoL. In our analysis, patients who required either a reoperation or readmission achieved significantly better PROs at 2 years compared to 1 year and baseline values. When comparing 2-year PROs for those readmitted versus not, there was no statistical difference for either SRS or ODI.

The fact that readmitted and revised patients were able to recover as well as those without readmission is likely attributed, at least in part, to the high-volume nature of our spine-focused hospital and aggressive attention for any complication requiring readmission and/or reoperation. Fluid communication between the surgeons and their staff, ancillary subspecialty teams, and the patient can lead to prompt and effective management of complications and overall improved patient outcomes.

A number of limitations must be acknowledged for this study. First, although this was a retrospective review, all of the data was prospectively entered into a standardized electronic database on a continual basis and all patients were consecutively enrolled. Unfortunately, nearly 30% of patients were lost to follow-up by 2 years postoperative. Those without follow-up had a similar preoperative comorbidity burden (ASA > 2: 29.1% [with follow-up] vs. 26.4% [without follow-up]; p = 0.663) and baseline operative characteristics (TIL: 13.3 [with follow-up] vs. 13.7 [without follow-up]; p = 0.418; operative time: 473 min [with follow-up] vs. 461 min [without follow-up]; p = 0.504; PSO: 5.7% [with follow-up] vs. 5.6% [without follow-up]; p = 0.961; VCR: 8% [with follow-up] vs. 9.7% [without follow-up]; p = 0.659; TLIF: 69.1% [with follow-up] vs. 66.7% [without follow-up]; p = 0.643). Nevertheless, it is possible with lack of follow-up that we underestimated the true readmission/reoperation rates and overestimated the benefit of surgery based on PRO’s. Next, given the single-center nature of our study, our findings may not be generalizable to other institutions who treat ASD patients. Finally, complications including pseudarthrosis and PJK are known to occur beyond the 2-year follow-up period. Studies with extended follow-up periods are needed to understand the full extent of long-term complications.

Conclusion

Our results demonstrate that the vast majority of patients can achieve clinically significant improvement in HRQoL after complex ASD surgery. Major improvements were observed in every domain of the SRS survey and several domains of the ODI survey (pain, standing, sleeping, social life, and traveling). Several predictors were identified for unplanned readmissions which may help surgeons with preoperative risk-stratification. Furthermore, and somewhat surprisingly, our findings suggest that readmissions and revision surgery do not adversely affect PRO’s in complex ASD patients by 2 years postoperative. Those who require an unplanned readmission or reoperation can significantly improve their HRQoL by 2-year follow-up compared to preoperative-baseline and 1-year follow-up as well as achieve similar improvements compared to those who did not require a readmission. These findings can provide valuable insight for patients and providers during the shared decision-making process for these complex surgical cases.