Impact of the COVID-19 Pandemic on Outcomes and Perioperative Factors Associated with Posterior Cervical Fusion

Abstract

Introduction

While there is anecdotal evidence that the coronavirus disease 2019 (COVID-19) pandemic altered perioperative decision-making in patients requiring posterior cervical fusion (PCF), a national-level analysis to examine the significance of this hypothesis has not yet been conducted. This study aimed to determine the potential differences in perioperative variables and surgical outcomes of PCF performed before vs. during the COVID-19 pandemic.

Methods

Adults who underwent PCF were identified in the 2019 (prepandemic) and 2020 (intrapandemic) NSQIP datasets. Differences in 30-day readmission, reoperation, and morbidity were evaluated using multivariate logistic regression. On the other hand, differences in operative time and relative value units (RVUs) were estimated using quantile regression. Furthermore, the odds ratios (OR) for length of stay (LOS) were estimated using negative binomial regression. Secondary outcomes included rates of nonhome discharge and outpatient surgery.

Results

A total of 3,444 patients were included in this study (50.7% from 2020). Readmission, reoperation, morbidity, operative time, and RVUs per minute were similar between cohorts (p>0.05). The LOS (OR 1.086, p<0.001) and RVUs-per-case (coefficient +0.360, p=0.037) were significantly greater in 2020 compared to 2019. Operation year 2020 was also associated with lower rates of nonhome discharge (22.3% vs. 25.8%, p=0.017) and higher rates of outpatient surgery (4.8% vs. 3.0%, p=0.006).

Conclusions

During the COVID-19 pandemic, a 28% decreased odds of nonhome discharge following PCF and a 72% increased odds of PCF being performed in an outpatient setting were observed. The readmission, reoperation, and morbidity rates remained unchanged during this period. This is notable given that patients in the 2020 group were more frail. This suggests that patients were shifted to outpatient centers possibly to make up for potentially reduced case volume, highlighting the potential to evaluate rehabilitation-discharge criteria. Further research should evaluate these findings in more detail and on a regional basis.

Introduction

Due to the coronavirus disease 2019 (COVID-19) pandemic, both hospitals, and physicians were forced to make adaptive changes in their surgical practice, such as limiting inpatient stays, shifting procedures to outpatient centers, or canceling surgeries altogether^1,2). In patients with operative degenerative cervical disk disease, cancelation of elective surgeries and outpatient clinics leads to a significant delay in treatment, which greatly affects their quality of life¹⁾. However, the cumulative impact of these changes in spine patients remains unclear^1,3).

With the growing elderly population, it is crucial to understand which combination of factors yields ideal postoperative outcomes for patients undergoing posterior cervical fusion (PCF) so that hospitals may optimize their policies to maximally benefit the patient. Posterior cervical decompression and fusion is an option for the treatment of cervical myelopathy or radiculopathy caused by degenerative disk disease or stenosis^4,5). While there is anecdotal evidence that the COVID-19 pandemic altered perioperative decision-making in patients requiring PCF, a national-level analysis to examine the significance of this hypothesis has not yet been conducted⁶⁾. A recent study using data from one tertiary care center found that in patients undergoing elective thoracolumbar adult spinal deformity surgery, the length of stay (LOS) decreased while the rates of home discharge increased²⁾. Interestingly, there appeared to be no adverse effect on complication or readmission rates.

The present study aimed to test the hypothesis that the pandemic led to a significant shift in PCF to outpatient procedures, decreased rates of discharge to nonhome postoperative care, and shortened LOS in patients undergoing PCF. We compared patients who underwent PCF during the pre- and intrapandemic periods to determine whether these potential changes led to any difference in 30-day surgical outcomes.

Materials and Methods

For this study, patient consent, and Institutional Review Board approval were not necessary because a publicly available, de-identified national surgical database was used, and there was no direct patient involvement. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

This retrospective cohort study used data obtained from the American College of Surgeons National Surgical Quality Improvement Program database (ACS-NSQIP). ACS-NSQIP has been shown to have excellent validity, reliability, and low rate of reporting error^7,8). Patients aged ≥18 years who underwent PCF in 2019 and 2020 were identified and included based on the Current Procedural Terminology (CPT) code 22600. Patients were excluded if they had a diagnosis other than myelopathy or radiculopathy. Furthermore, patients were excluded via the CPT code if they underwent anterior, thoracic, lumbar, deformity, revision, tumor, nonelective, or emergency surgery or had missing data (Appendix A). Patients who had preoperative sepsis, septic shock, disseminated cancer, unexpected weight loss, preoperative open wound, or wound class other than level 1 were excluded by their respective NSQIP variables. Patients with missing data were also excluded to prevent biases in the results.

The primary outcomes were 30-day readmission, reoperation, and morbidity, LOS, relative value units (RVUs)-per-minute and per-case, and operative time. The secondary outcomes were the rates of nonhome discharge and outpatient surgery. A separate subgroup analysis was conducted to compare outpatient-only outcomes between 2020 and 2019 to further characterize our results. Readmission includes any inpatient stay in the same or another hospital related to the surgical procedure. Reoperation includes all major surgical procedures requiring return to the operating room for any intervention type. Morbidity includes the occurrence of one or more complications reported in the NSQIP dataset, such as infectious, cardiopulmonary, renal, neurological, hematologic, and thromboembolic complications.

All statistical analyses were conducted using the SPSS software (version 28, IBM, Armonk, New York). Demographic, comorbidity, laboratory, and procedural factors were individually analyzed for baseline differences between patients in the 2019 and 2020 datasets using Student's t-test, Kruskal-Wallis H test, chi-squared test, or Fisher's exact test as appropriate for univariate analysis. Furthermore, the admission quarter was separately analyzed for association with primary outcomes to evaluate for potential temporal effects related to COVID-19-prompted lockdowns in 2020. Baseline variables that significantly (p<0.05) differed between 2019 and 2020 were included and adjusted for in multivariate regression, in which variables with the highest p-values were sequentially eliminated until only the statistically significant variables (p<0.05) remained. Multivariate analysis of readmission, reoperation, and morbidity was conducted using logistic regression. The assumption of normality for LOS, RVUs-per-case, RVUs-per-minute, and operative time was assessed using the Kolmogorov-Smirnov test and was not met. Therefore, regression coefficients for RVUs-per-case, RVUs-per-minute, and operative time were estimated via quantile (median) regression, and odds ratios (OR) for LOS were estimated via negative binomial regression. Goodness of fit was evaluated using the Hosmer-Lemeshow test, Akaike Information Criteria, or Bayesian Information Criteria as appropriate.

Results

A total of 3,444 patients (1,699 from 2019; 1,745 from 2020) met the inclusion criteria. Univariate analysis of baseline differences revealed that there were slight differences between patients in 2019 and 2020. In 2020, there were fewer females (39.1 vs. 43.3%, p=0.013), more patients of nonwhite race (22.9 vs. 19.1%, p=0.012), and more patients with a 5-item modified frailty index score (mFI5)≥2 (29.2 vs. 23.7%, p<0.001) (Table 1). Furthermore, there were lower rates of nonhome discharge (22.3 vs. 25.8%, p=0.017) and higher rates of outpatient surgery (4.8 vs. 3.0%, p=0.006).

Table 1.

Baseline Patient Characteristics.

	2019 N (%)	2020 N (%)	p-value
Demographics	N=1,699	N=1,745
Age (years), mean±SD	63.9 (10.1)	64.2 (10.7)	0.175
Female sex	735 (43.3%)	682 (39.1%)	0.013
Nonwhite race	279 (19.1%)	344 (22.9%)	0.012
Hispanic ethnicity	114 (7.7%)	133 (8.5%)	0.385
Comorbidities
5-item modified frailty index
mFI-5=0	538 (31.7%)	513 (29.5%)	0.159
mFI-5=1	759 (44.7%)	722 (41.4%)	0.051
mFI-5≥2	402 (23.7%)	509 (29.2%)	<0.001
BMI	30.1 (6.4)	30.0 (6.5)	0.399
Diabetes mellitus	393 (23.1%)	460 (26.4%)	0.028
Dyspnea	125 (7.4%)	125 (7.2%)	0.826
Functional dependence	77 (4.6%)	82 (4.7%)	0.836
Smoker	356 (21.0%)	402 (23.0%)	0.140
COPD	101 (5.9%)	124 (7.1%)	0.168
Congestive heart failure	9 (0.5%)	8 (0.5%)	0.765
Hypertension requiring medication	1,034 (60.9%)	1,132 (64.9%)	0.015
Chronic steroid use	93 (5.5%)	75 (4.3%)	0.109
Bleeding disorder	19 (1.1%)	27 (1.5%)	0.273
ASA class ≥3	1,147 (67.5%)	1,221 (70.0%)	0.119
Laboratory values
Creatinine	0.96 (0.54)	1.01 (0.73)	0.056
White cell count	7.4 (2.5)	7.4 (2.5)	0.989
Hematocrit	41.5 (4.5)	41.3 (4.8)	0.175
Platelet count	247 (73)	245 (71)	0.357
Procedural factors
Nonhome discharge	438 (25.8%)	389 (22.3%)	0.017
Outpatient surgery	51 (3.0%)	84 (4.8%)	0.006
4+ levels fused	260 (15.3%)	270 (15.5%)	0.890

Bold indicates statistical significance (p<0.05). SD, standard deviation; COPD, chronic obstructive pulmonary disease; ASA, American Society of Anesthesiologists; mFI-5, 5-item modified frailty index

In the univariate analysis, LOS was significantly greater in 2020 than in 2019 (4.2 vs. 4.0 days, p=0.015). The readmission (7.1 vs. 7.2%), reoperation (3.6 vs. 4.1%), and morbidity (10.3 vs. 9.8%) rates, operative time (159 vs. 155 min), and RVUs-per-minute (0.36 vs. 0.36) were statistically similar (p>0.05) between 2020 and 2019, respectively (Table 2). Separate analysis of the association between admission quarter and primary outcomes revealed no significant differences.

Table 2.

Unadjusted 30-day Outcomes of Posterior Cervical Fusions Performed in 2019 vs. 2020.

	2019 N (%)	2020 N (%)	p-value
	N=1,699	N=1,745
30-day outcomes
Readmission	122 (7.2%)	124 (7.1%)	0.932
Reoperation	70 (4.1%)	63 (3.6%)	0.438
Morbidity	167 (9.8%)	179 (10.3%)	0.676
Complications
Superficial SSI	28 (1.6%)	23 (1.3%)	0.423
Deep SSI	7 (0.4%)	12 (0.7%)	0.275
Organ/space SSI	13 (0.8%)	11 (0.6%)	0.635
Wound disruption	12 (0.7%)	14 (0.8%)	0.745
Pneumonia	13 (0.8%)	17 (1.0%)	0.509
Unplanned intubation	11 (0.6%)	7 (0.4%)	0.316
Pulmonary embolism	13 (0.8%)	14 (0.8%)	0.902
Ventilator >48 hours	4 (0.2%)	4 (0.2%)	1.000
Renal insufficiency	4 (0.2%)	1 (0.1%)	0.212
Acute renal failure	2 (0.1%)	1 (0.1%)	0.620
Urinary tract infection	24 (1.4%)	28 (1.6%)	0.644
Stroke	3 (0.2%)	4 (0.2%)	1.000
Cardiac arrest requiring CPR	3 (0.2%)	4 (0.2%)	1.000
Myocardial infarction	8 (0.5%)	10 (0.6%)	0.678
Transfusion	49 (2.9%)	53 (3.0%)	0.791
Deep venous thrombosis	18 (1.1%)	11 (0.6%)	0.168
Sepsis	19 (1.1%)	10 (0.6%)	0.080
Perioperative outcomes
Operative time (min; median, IQR)	155 (117-208)	159 (118-207)	0.723
RVUs-per-minute (median, IQR)	0.36 (0.24-0.49)	0.36 (0.26-0.49)	0.558
RVUs-per-case (median, IQR)	57.8 (48.8-65.3)	57.8 (51.3-65.1)	0.127
Length of stay (days; mean±SD)	4.0 (3.2)	4.2 (3.5)	0.015

Bold values indicate statistical significance (p<0.05). SSI, surgical site infection; CPR, cardiopulmonary resuscitation; IQR, interquartile range

In the multivariate analysis, the 2020 operative year independently predicted increased LOS (OR=1.086; p<0.001; CI⁹⁵: 1.037, 1.136) and greater RVUs-per-case (+0.360; p=0.037; CI⁹⁵: 0.021, 0.699) (Table 3). In addition, 2020 predicted lower rates of nonhome discharge (22 vs. 26%; OR=0.719; p=0.001; CI⁹⁵: 0.597, 0.867) and greater rates of outpatient surgery (4.8 vs. 3.0%; OR=1.716; p=0.007; CI⁹⁵: 1.163, 2.533). Subgroup analysis of outpatient-only cases revealed similar median RVUs-per-case and 30-day readmission, reoperation, and morbidity rates between 2020 and 2019; however, outpatient-only operative time (p=0.062) tended to be significantly greater in 2020 than in 2019 (Appendix B).

Table 3.

Multivariate Analysis of Impact of Operation Year 2020 on Postoperative Outcomes.

30-day outcomes	OR (95% CI)	p-value
Readmission	0.968 (0.736, 1.271)	0.813
Reoperation	0.801 (0.548, 1.170)	0.251
Morbidity	0.915 (0.711, 1.177)	0.489
Perioperative variables	OR (95% CI)	p-value
Length of stay (days)	1.086 (1.037, 1.136)	<0.001
Perioperative variables	Coefficient (95% CI)	p-value
RVUs-per-minute	0.004 (−0.011, 0.020)	0.589
RVUs-per-case	0.360 (0.021, 0.699)	0.037
Operative time (min)	1.481 (−4.292, 7.254)	0.615

Bold values indicate statistical significance (p<0.05). OR, odds ratio; CI, confidence interval

There were 246 readmissions in 3,444 patients, and the overall readmission rate was 7.1%. In the multivariate analysis, mFI5≥2 (OR=1.548; p=0.018; CI⁹⁵: 1.078, 2.224) and operative time (hour; OR=1.170; p=0.002; CI⁹⁵: 1.058, 1.293) predicted readmission. Furthermore, there were 133 reoperations, and the overall reoperation rate was 3.9%. In the multivariate analysis, diabetes (OR=1.556; p=0.035; CI⁹⁵: 1.032, 2.348) and LOS (days; OR=1.081; p<0.001; CI⁹⁵: 1.037, 1.126) predicted reoperation. There were 346 patients who experienced morbidity, and the overall morbidity rate was 10.0%. Nonhome discharge (OR=1.558; p=0.002; CI⁹⁵: 1.181, 2.055), mFI5≥2 (OR=1.835; p<0.001; CI⁹⁵: 1.308, 2.576), LOS (days; OR=1.108; p<0.001; CI⁹⁵: 1.071, 1.147), and operative time (hour; OR=1.304; p<0.001; CI⁹⁵: 1.197, 1.420) predicted morbidity. Of note, an mFI5 score of 1 predicted neither readmission, reoperation, nor morbidity.

Nonwhite race (OR=1.112; p<0.001; CI⁹⁵: 1.053, 1.174), mFI5=1 (OR=1.084; p=0.004; CI⁹⁵: 1.026, 1.147), mFI5≥2 (OR=1.147; p<0.001; CI⁹⁵: 1.078, 1.221), nonhome discharge (OR=1.664; p<0.001; CI⁹⁵: 1.582, 1.750), and operative time (hour; OR=0.673; p<0.001; CI⁹⁵: 0.585, 0.774) predicted a 1-day increase in LOS. Outpatient surgery was protective against increased LOS (OR=0.673; p<0.001; CI⁹⁵: 0.585, 0.774).

Nonhome discharge (+7.94 min; p=0.032; CI⁹⁵: 0.672, 15.2), LOS (days; +4.33 min; p<0.001; CI⁹⁵: 3.37, 5.31), and RVUs-per-case (+0.886 min; p<0.001; CI⁹⁵: 0.734, 1.04) predicted increased median operative time. Moreover, female sex (−11.9 min; p<0.001; CI⁹⁵: −17.7, −6.05) and outpatient surgery (−24.7 min; p=0.001; CI⁹⁵: −39.5, −9.93) predicted decreased median operative time.

Female sex (+0.024 RVUs-per-minute; p=0.003; CI⁹⁵: 0.008, 0.040) and outpatient surgery (+0.050 RVUs-per-minute; p=0.013; CI⁹⁵: 0.011, 0.090) predicted increased RVUs-per-minute. LOS (day; −0.006 RVUs-per-minute; p<0.001; CI⁹⁵: −0.009, −0.003) predicted decreased RVUs-per-minute. Operative time (hour; +0.911 RVUs-per-case; p<0.001; CI⁹⁵: 0.777, 1.045) predicted increased RVUs-per-case. Outpatient surgery (−10.03 RVUs-per-case; p<0.001; CI⁹⁵: −10.9, −9.17) predicted decreased RVUs-per-case.

Discussion

Upon analyzing the 2019 and 2020 NSQIP datasets for national-level changes occurring during the COVID-19 pandemic, multivariate analysis revealed that there was a 28% decreased odds of nonhome discharge after PCF and a 72% increased odds of PCF being performed in an outpatient surgery setting. This shifting trend was likely influenced by the desire to reduce infection risk associated with inpatient stays within hospitals or rehabilitation centers. Traditionally, the barriers to outpatient cervical spine surgery included risk of airway compromise and neck hematoma, though a growing evidence showed its safety^9-11).

Our study also found that the rates of readmission, reoperation, and morbidity remained statistically similar in 2020 compared with 2019. Previous studies have suggested that spine surgeries performed during the pandemic are not associated with worse outcomes. Riley et al. reported that spine surgical procedures performed in 2020 were associated with similar 30-day mortality and complication rates to those in 2019¹²⁾. Louie et al. reported statistically similar 1-month postoperative complication rates between patients who underwent spine surgery during vs. before the pandemic¹³⁾. Wang et al. demonstrated that the readmission and complication rates did not change during the pandemic among patients undergoing adult spinal deformity surgery²⁾.

Within our study cohort, the 30-day postoperative outcomes remained statistically unchanged despite an apparent increase in outpatient PCF cases and a decrease in discharge to inpatient postacute care facilities. This is particularly notable given that patients in the 2020 operative group were more frail and that the numbers of cases between the two years were equal. During the initial peak of the pandemic, the operating room time in inpatient hospital settings was significantly limited and required careful rationing, typically being reserved for emergent and urgent procedures. However, freestanding ambulatory surgery centers often had high operating room availability^14,15). Therefore, the patients included in the present study may have been shifted to outpatient centers to make up for potentially reduced case volume as opposed to operating on a greater proportion of patients who would have been traditionally selected for an inpatient procedure.

Furthermore, separate subgroup analysis of outpatient-only cases in 2020 vs. 2019 revealed no significant differences in the 30-day readmission, reoperation, and morbidity rates. Interestingly, the median operative time tended to be significantly greater in 2020 than in 2019 (p=0.062), whereas the total RVUs-per-case remained statistically unchanged, and the RVUs-per-minute tended to be lower in 2020 than in 2019 (p=0.055). This may further suggest a transition of cases from the inpatient to outpatient setting that would have otherwise been performed as an inpatient. Taken together, this study suggests that it is safe to perform PCF in an outpatient setting and to avoid rehabilitation stays postoperatively.

Interestingly, 2020 predicted an 8.6% increased odds of a 1-day increase in LOS, suggesting that some patients spent more time admitted before being discharged to home rather than to a postacute care facility. This finding matched those within a UK healthcare system that demonstrated decreased nonhome discharge rates but increased LOS for elective orthopedic surgery¹⁶⁾. Multiple factors could have resulted in this finding, including staffing inefficiencies, logistic issues, and insufficient home support related to the COVID-19 pandemic resulting in a delay in discharge.

However, there is conflicting evidence in the current literature regarding the impact of the COVID-19 pandemic on LOS following spine surgery. Wang et al. demonstrated that among patients undergoing adult spinal deformity surgery, those who underwent surgery during the pandemic had significantly lower LOS²⁾. Louie et al. reported that LOS remained statistically similar between patients undergoing spine surgery at a single institution before and during the pandemic¹³⁾. Given this discrepancy present in the literature, the impact of the COVID-19 pandemic on perioperative variables such as LOS after spine surgery warrants further investigation. Further research should also evaluate these findings in more detail and on a regional basis.

This study had several limitations. The data were collected only from the ACS-NSQIP participating hospitals, and this may lead to sampling bias. The ACS-NSQIP database is comprised mainly of academic medical centers, which may limit the generalizability of the study findings. Furthermore, there is a lack of granularity of data inherent to large database studies, such as the inability to account for the wide variability in quality of care at the level of institutions or surgeons. Despite these limitations, this study provides valuable evidence for evaluating the impact of the COVID-19 pandemic on short-term outcomes after PCF surgery on a national level.

In conclusion, patients who underwent PCF during the COVID-19 pandemic were 28% less likely to be discharged to a nonhome setting and 72% more likely to undergo outpatient surgery. Despite an increasingly frail population, the most vulnerable during the COVID-19 pandemic, operative time, postoperative complications, and overall outcomes remained similar compared with patients operated in an inpatient setting in 2019.

Conflicts of Interest: The authors declare that there are no relevant conflicts of interest.

Sources of Funding: None

Author Contributions: Austen D. Katz: conception, data acquisition, data analysis, data interpretation, manuscript drafting, manuscript revision, supervision

Junho Song: data analysis, data interpretation, manuscript drafting, manuscript revision

Priya Duvvuri: data interpretation, manuscript drafting, manuscript revision

Alex Ngan: data interpretation, manuscript drafting, manuscript revision

Terence Ng: data interpretation, manuscript drafting, manuscript revision

Sayyida Hasan: data analysis, data interpretation, manuscript revision

Sohrab Virk: data interpretation, manuscript revision, supervision, administrative support

Jeff Silber: data interpretation, manuscript revision, supervision, administrative support

David Essig: data acquisition, data interpretation, manuscript revision, supervision, administrative support

Ethical Approval: Institutional Review Board approval was not required for this study as it was conducted using a publicly available, de-identified national surgical database.

Informed Consent: Informed consent was not required for this study as it was conducted using a publicly available, de-identified national surgical database.