Short-Term Neck Pain After Posterior Foraminotomy Compared with Anterior Discectomy with Fusion for Cervical Foraminal Radiculopathy

Background: Short-term neck pain after posterior cervical foraminotomy (posterior surgery) compared with anterior cervical discectomy with fusion (anterior surgery) treating cervical radiculopathy has only been assessed once, retrospectively, to our knowledge. The aim of this study was to prospectively evaluate the course of neck pain for 6 weeks after both treatments. Methods: This is a secondary analysis of the multicenter Foraminotomy ACDF Cost-Effectiveness Trial (FACET), conducted from January 2016 to May 2020. Of 389 patients who had single-level, 1-sided cervical radiculopathy and were screened for eligibility, 265 were randomly assigned to undergo posterior surgery (n = 132) or anterior surgery (n = 133). The primary outcome of the present analysis was neck pain, assessed weekly for 6 weeks using the visual analog scale (VAS), on a scale of 0 to 100. The secondary outcomes were arm pain, neck disability, work ability, quality of life, treatment satisfaction, motor and sensory changes, and hospital length of stay. Data were analyzed with mixed model analysis in intention-to-treat samples using 2-sided 95% confidence intervals (CIs). Results: In the first postoperative week, the mean VAS for neck pain was 56.2 mm (95% CI, 51.7 to 60.8 mm) after posterior surgery and 46.7 mm (95% CI, 42.2 to 51.2 mm) after anterior surgery. The mean between-group difference was 9.5 mm (95% CI, 3.3 to 15.7 mm), which gradually decreased to 2.3 mm (95% CI, −3.6 to 8.1 mm) at postoperative week 6. As of postoperative week 5, there was no significant difference between groups. Responder analyses confirmed this result. Secondary outcomes showed small differences between groups. Conclusions: Insight into the course of neck pain during the first 6 weeks after posterior compared with anterior surgery is provided. Despite initially more neck pain after posterior surgery, patients swiftly improved and, as of postoperative week 5, results similar to those after anterior surgery were observed. Our findings should enable improved patient counseling and enhanced shared decision-making between physicians and patients with cervical radiculopathy, where more neck pain in the first postoperative weeks should be balanced against the benefits of posterior surgery. Level of Evidence: Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.

of posterior compared with anterior surgery in patients with cervical radiculopathy 4 . At the 1-year follow-up, posterior surgery was noninferior to anterior surgery with regard to clinical outcome, with similar results for secondary outcomes including neck pain 5 . This result is consistent with current literature, mainly consisting of retrospective studies, reporting posterior surgery as a viable effective alternative to anterior surgery [6][7][8] .
However, a preference among surgeons for anterior surgery has been observed 1,9 . One of the reasons for this could be the belief that patients experience more short-term neck pain after posterior surgery 1,10 , although literature with regard to this matter has been limited. One retrospective study compared neck pain in the first 5 to 7 postoperative days, reporting a significant difference in neck pain favoring anterior surgery over posterior surgery 10 . However, no confidence intervals (CIs) or prolonged follow-up data were reported. A more thorough analysis would provide better insight into the course of neck pain in the first weeks after the surgical procedure.
Therefore, the primary aim of this study was to analyze the visual analog scale (VAS) for neck pain in the first 6 weeks after posterior compared with anterior surgery in patients treated for cervical radiculopathy included in the FACET, as well as to evaluate additional short-term clinical outcomes.

Materials and Methods
Trial Design T he FACET was an investigator-blinded, multicenter, randomized clinical trial. The protocol was approved by the Research Ethical Board of the University Medical Center Groningen, the Netherlands, and is registered with the Dutch Trial Register (NTR5536). A detailed description of the study design and 1-year clinical results have been published 4,5 . This analysis reports short-term outcomes of the FACET. The Consolidated Standards of Reporting Trials (CONSORT) guidelines were followed 11 .

Participants
Dutch-speaking patients (18 to 80 years of age) with an indication for a surgical procedure for single-level, 1-sided cervical radiculopathy attributable to foraminal intervertebral disc herniation and/or osteophyte formation were eligible. Patients were enrolled from 9 Dutch hospitals from January 2016 to May 2020. After obtaining written informed consent, patients underwent randomization. The full eligibility criteria are detailed in the protocol 4 .

Interventions and Randomization
Patients were randomly assigned to posterior or anterior surgery in a 1:1 ratio, stratified by hospital, using a webbased block randomization program by an independent institute (Trans European Network for Clinical Trials Services [TENALEA]). Patient or surgeon blinding was not possible due to the nature of the surgical procedures. Participating neurosurgeons were familiar with both surgical techniques.
Posterior surgery was performed with the patient in the prone position and the patient's head in a 3-point head holder. The correct level was determined with fluoroscopy; partial hemilaminectomy and foraminotomy were performed using magnification. Osteophytes and compressing disc material were removed if necessary. Anterior surgery was performed with a standard ventral approach 12 . Following discectomy, any disc fragments extruding into the neuroforamen or osteophytic components as part of the uncovertebral joint were reduced, and an intervertebral spacer was applied to the disc space. No postoperative neck brace or additional fixation was used in either technique.
Outcomes A complete description of the primary and secondary outcomes of the FACET has been previously described 4 . In the current analysis, the primary outcome was VAS neck pain, visualizing experienced neck pain on a horizontal 100-mm scale ranging from 0 (no pain) to 100 (worst pain imaginable) 13 . VAS neck pain was assessed every week until 6 weeks postoperatively. Additionally, although VAS arm pain data were collected as a primary outcome in our original trial, VAS arm pain was also evaluated weekly in the first 6 weeks postoperatively for the purpose of short-term analysis, in which VAS for arm pain functioned as a surrogate of treatment efficacy. Also, the Neck Disability Index (NDI) (10-item index ranging from 0 to 50; sum of scores doubled to yield a percentage), Work Ability Index Single-Item (WAI-SI, 0 to 10; higher scores indicating better work ability), EuroQol 5-Dimensions 5-Level questionnaire (EQ-5D-5L), and treatment satisfaction (1 to 7 rating scale) were assessed in postoperative week 6, hospital length of stay was documented, and a neurologic examination assessing sensibility and motor strength was performed at discharge and postoperative week 6. Validated Dutch-language versions of the questionnaires were used [14][15][16][17] (see the questionnaires in the Appendix).

Statistical Analysis
Details regarding the power calculation have been previously described 4,5 . In brief, power for the FACET was calculated on the basis of a hypothesized overall similar success rate after anterior and posterior surgery with an alpha of 0.05, a power of 80%, a noninferiority margin of 10% (based on the success rate), and an anticipated 10% loss of follow-up, leading to a targeted sample size of 308 patients. Unfortunately, the inclusion rate was lower than anticipated, partially related to COVID-19 and its associated cancellation of nonemergency health care. Therefore, an interim analysis and power calculation were performed by an independent statistician indicating that it was safe to end the trial at 86% of the predefined sample size with low risks of false-negatives 5 .
Data were analyzed on the intention-to-treat basis. All analyses were performed in R version 4.0.5 (R Foundation for Statistical Computing). Continuous outcomes were reported as means and standard deviations, or (if non-normally distributed) as medians and interquartile ranges. Between-group differences were estimated using 2-sided 95% CIs 4 .
Between-group differences in VAS neck pain, including all follow-up times (1 to 6 weeks), were analyzed using linear mixed modeling. Time, baseline score, treatment (posterior compared with anterior surgery), and time-treatment interaction were included as fixed effects. Changes over time in individual participants were included as random effects (that is, random intercept and slope). Subsequently, additional covariates were included as fixed effects and tested for improved model fit using forward selection (likelihood ratio test, alpha = 0.05, maximum likelihood estimation). For an overview of all covariates tested and included, see the Appendix, Table S1. If covariates other than time, treatment, or the baseline value were included, a partial R 2 was calculated to quantify the relative importance of the included variable.
With regard to secondary outcomes, between-group differences in VAS arm pain (measured weekly) were assessed using linear mixed modeling with the same abovementioned fixed and random effects and model selection. Finally, differences in the length of stay, sensibility or motor strength, and outcomes assessed at postoperative week 6 were tested using 2-sample t tests (for continuous outcomes) or 2-proportion z-tests (for categorical outcomes).
Sensitivity analyses were performed for the primary outcome of VAS neck pain, assessing the robustness of the findings. First, responder analyses were conducted; the patients were dichotomized into responders and nonresponders according to their VAS neck pain scores. Patients were classified as responders when reporting a decrease in VAS neck pain compared with baseline reaching or exceeding the predefined threshold of the minimal clinically important difference (MCID) (26 mm for VAS neck pain 13 ) or reporting a reduction of ‡30% in VAS neck pain compared with baseline 18 . Generalized linear mixed-effects modeling for binary outcomes with the same abovementioned fixed and random effects and model selection were applied for the responder analyses. Additionally, secondary analyses of the primary VAS neck pain outcome were conducted using only complete cases, calculated change scores from baseline, logtransformed VAS neck pain scores, models with an alternative random effect structure, and multiple imputation by chained equations (MICE) to generate 20 different imputations for missing VAS scores 19 .

Source of Funding
The FACET was funded by the Netherlands Organisation for Health Research and Development (ZonMw, 843002604). The funder had no role in design, data collection, analysis, interpretation, or writing of the report.

Participant Characteristics
F rom January 2016 to May 2020, 265 participants were assigned to undergo posterior surgery (n = 132) or anterior surgery (n = 133) (Fig. 1). The mean age (and standard deviation) was 52.1 ± 8.3 years, with a male:female ratio of 133:132, as previously described 5 . Twenty-five participants (15 in the posterior surgery group and 10 in the anterior surgery group) did not receive the allocated intervention; the most common reason was improvement of symptoms without undergoing a surgical procedure (5 patients in the posterior surgery group and 6 patients in the anterior surgery group) (Fig. 1). Intention-to-treat analysis was performed for all randomized participants with available data at each time point, with 104 patients (87%) in the posterior surgery group and 108 patients (87%) in the anterior surgery group analyzed at postoperative week 6. There were no significant betweengroup differences in baseline characteristics 5 (Table I). Figure 2 shows the improvement in mean VAS neck pain after both posterior and anterior surgery as well as between-group differences in the primary mixed model analysis. This model fit adequately to the data, and the inclusion of additional covariates other than time, treatment, and baseline scores did not improve the model fit (see the Appendix, Tables S1 and S3 and Figure S1).

Primary Outcome
In the first week after the surgical procedure, the modelbased mean VAS neck pain was 56.2 mm (95% CI, 51.7 to 60.8 mm) in the posterior surgery group and 46.7 mm (95% CI, 42.2 to 51.2 mm) in the anterior surgery group. Thus, at postoperative week 1, mean VAS neck pain was 9.5 mm (95% CI, 3.3 to 15.7 mm) higher after posterior compared with anterior surgery. This difference between groups gradually decreased and became nonsignificant at week 5. At week 6, the mean between-group difference was 2.3 mm (95% CI, 23.6 to 8.1 mm) (Fig. 2, Table II). Responder analyses confirmed this, with proportions of responders for VAS neck pain increasing over time and the 95% CI including 0 beginning at postoperative week 5 (Table II). Using the MCID threshold, the proportions of responders were 0.58 in the posterior surgery group and 0.60 in the anterior surgery group at postoperative week 6 (difference, 20.02 [95% CI, 20.05 to 0.06]). Using the 30% improvement threshold, these proportions were 0.75 in the posterior surgery group and 0.78 in the anterior surgery group, with a difference of 20.03 (95% CI, 20.05 to 0.03) (Table III; see also  the Appendix, Tables S4 and S5). The sensitivity analyses, including those using complete cases, change scores, logtransformed VAS neck pain, alternative random effect structures, and multiple imputation, did not alter the findings of the main analysis (see the Appendix, Tables S7 through S11 and Figure S2).

Secondary Outcomes
In the first week after the surgical procedure, the model-based mean VAS arm pain was 25.0 mm (95% CI, 20.2 to 29.7 mm) in the posterior surgery group and 18.6 mm (95% CI, 14.1 to 23.1 mm) in the anterior surgery group. The between-group difference was 6.3 mm (95% CI, 2.2 to 10.5 mm), indicating significantly higher arm pain after posterior compared with anterior surgery; this difference did not change significantly over time. Comparable length of stay, sensory and motor changes, and outcomes assessed at postoperative week 6 using the NDI,   WAI-SI, EQ-5D-5L, and treatment satisfaction were observed between the groups (Table IV; see also the Appendix, Tables S2 and S6).

Discussion
T his is the first multicenter, randomized clinical trial of which we are aware to analyze neck pain in the first weeks after posterior compared with anterior surgery for the treatment of cervical radiculopathy. There was more neck pain in the first 4 weeks after posterior compared with anterior surgery, but VAS neck pain trajectories became similar, as of postoperative week 5, with £3.7 mm of difference, which was confirmed by responder analyses. Secondary outcomes were comparable between groups.
We are aware of only 1 previous study that assessed shortterm neck pain by retrospectively scanning postoperative charts in the first 5 to 7 days. Those authors reported significantly less neck pain after anterior compared with posterior surgery 10 . In our study, mean VAS neck pain was initially 9.5 mm (95% CI, 3.3 to 15.7 mm) higher after posterior compared with anterior surgery. This was expected and is likely attributable to the nature of the posterior approach, as the muscles in the neck are retracted. The between-group difference in neck pain, comparing posterior surgery with anterior surgery, gradually decreased and became nonsignificant at postoperative week 5, with a difference of only 2.3 mm (95%  †ASA = American Society of Anesthesiologists, and NSAID = nonsteroidal anti-inflammatory drug. ‡The values are given as the mean and the standard deviation. §The values are given as the number of patients, with the percentage in parentheses; the percentages may not total 100% because of rounding. #The values are given as the median, with the interquartile range in parentheses. **The ASA classification system ranges from I to VI, where higher classes indicate a greater risk. No patients had an ASA IV, V, or VI classification. † †Data were missing for 3 patients in the posterior surgery group and 2 patients in the anterior surgery group. ‡ ‡Data were missing for 1 patient in the posterior surgery group and 2 patients in the anterior surgery group. § §Data were missing for 2 patients in the anterior surgery group. ##The specification of comorbidities is on an event level (not a patient level); therefore, no percentages are given, because several patients had multiple comorbidities. The estimated mean VAS neck pain after posterior surgery and anterior surgery. The shading represents the 95% CIs. Fig. 2-A The estimated mean scores for VAS neck pain are depicted for both the posterior and anterior surgery groups. Fig. 2-B The estimated mean between-group difference (posterior compared with anterior surgery) is depicted. It can be noted that, starting at postoperative week 5, the 95% CIs of the between-group difference included 0.  CI, 23.6 to 8.1 mm) at postoperative week 6. Although our responder analyses confirmed this, it remains challenging to determine whether significant between-group differences in the first 4 weeks postoperatively were also clinically relevant. The current literature does not provide a threshold for clinically relevant neck pain differences between groups, only within groups 20 . Because posterior surgery avoids fusion-related complications, maintains postoperative range of motion, and has potential economic benefits 7,21,22 , it remains up to the physician and the patient to balance the 5.2 to 9.5-mm higher VAS for neck *Responder analyses were performed using both the predefined threshold of the MCID for VAS neck pain (26 mm) as well as a reported reduction of ‡30% in VAS for neck pain compared with baseline. It can be noted that the MCID threshold responder analysis was more conservative than the ‡30% threshold analysis. †The values are given as the difference, with the 95% CI in parentheses; the 95% CIs for the responder analyses were obtained with bootstrapping. pain in the first 4 postoperative weeks against the benefits of posterior surgery. Moreover, despite the majority of patients having reached or exceeded the predefined MCID threshold for improvement in VAS neck pain at postoperative week 6 (0.58 percentage points in the posterior surgery group and 0.60 percentage points in the anterior surgery group), one might argue that relatively large proportions of patients did not reach this level or exceeded it. One retrospective study showed even lower rates of achieving the same MCID value after a 2-year follow-up (50% in the posterior surgery group and 51% in the anterior surgery group), although it did not report measures of uncertainty 23 . Among 3 studies analyzing MCID achievement after anterior surgery, of which 2 studies used a combined VAS for arm and neck pain, marked variation was demonstrated in MCID achievement rates ranging from 46% to 88% [24][25][26] . This variation is likely due to their use of different MCID threshold values. One of these studies used the same MCID threshold as the current study and found a comparable percentage of responders for VAS neck pain after anterior surgery (55%, with no stated measure of uncertainty) 26 , although we also observed higher proportions of responders using a reduction of ‡30% in VAS neck pain from baseline as the threshold to define responders. Therefore, we argue that it is important to further investigate the MCID threshold values to come to a universally accepted threshold for responder analyses, as these are increasingly utilized as a measure of overall performance after a surgical procedure 27 .
With regard to secondary outcomes, similar results between groups were found, except for a stable between-group difference of 6.3 mm (95% CI, 2.2 to 10.5 mm) in VAS arm pain. The current literature provides conflicting evidence regarding short-term postoperative arm pain. One retrospective study demonstrated significantly lower arm pain after anterior compared with posterior surgery in the first postoperative 5 to 7 days 10 . Conversely, among studies that analyzed VAS arm pain with longer follow-up, 1 study found a between-group difference of 6.8 mm after 5 years (0.68 on a 0-to-10 VAS scale, converted to a 0-to-100 scale for comparison), which is similar to the difference in our study. Those authors concluded that this between-group difference was not significant or clinically relevant 28 . Therefore, we do not regard the 6.3-mm between-group difference found in the current study as clinically relevant, especially as all other secondary outcomes at postoperative week 6 were similar between groups. Nevertheless, additional research should further elaborate on this.
The current study is an analysis of short-term outcomes of the FACET, a noninferiority randomized controlled trial. Unfortunately, no noninferiority margins for secondary outcomes, including VAS neck pain, were defined a priori. Because noninferiority margins are often chosen empirically because of the lack of sufficient data, our data could be used to define such margins for future research, as our trial had accurate controlled sets of short-term and long-term data.
The strengths of our study are the multicenter and randomized design and the inclusion of only patients with cervical foraminal radiculopathy without myelopathy. To our knowledge, this is the first study that longitudinally analyzed shortterm postoperative neck pain. We believe that our results are important, as they provide clarity regarding the course of neck pain after posterior compared with anterior surgery, which is relevant to all physicians involved in the treatment of patients with cervical radiculopathy as it allows them to provide unbiased information on either treatment to their patients.
The current study had some limitations. First, the power for the FACET was calculated on the basis of the success rate. The trial was not designed and powered for the VAS neck pain that was the primary outcome analyzed in the present study; therefore, our results should be considered exploratory. Second, we did not assess VAS neck pain in the first days after the surgical procedure or the EQ-5D-5L or treatment satisfaction in the first 4 weeks postoperatively. Those outcomes could have provided more information on whether neck pain was present immediately after the surgical procedure and whether more neck pain in the first 5 weeks after posterior surgery had an influence on quality of life or treatment satisfaction. Also, the MCID threshold was based on values reported in the literature, but, as we demonstrated, the responder analysis gave different results when alternative values ( ‡30% improvement) were used.
Our findings demonstrated the prevalence of more neck pain in the first 4 weeks after posterior compared with anterior surgery, whereafter patients in the posterior surgery group rapidly improved, with the remaining between-group difference becoming small and clinically unimportant at postoperative week 5. Length of stay, sensibility and motor changes, arm pain, neck disability, work ability, quality of life, and treatment satisfaction were similar between groups.
In conclusion, during the first 4 weeks after posterior surgery, more neck pain was observed than after anterior surgery. This gradually improved and, starting at postoperative week 5, results similar to those after anterior surgery were observed. Secondary outcomes were comparable between groups. Our results provide a better comprehension of the course of neck pain in the first weeks postoperatively and add to the decision-making process between physicians and patients with cervical radiculopathy, where neck pain in the first postoperative weeks and the benefits of posterior surgery should be weighed against one another.