This retrospective controlled cohort study included 102 consecutive adult patients who underwent single-level ELD at a tertiary orthopedic and spine center (University Orthopedic and Rehabilitation Hospital, Zakopane, Poland) between January 2021 and December 2023. The study followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for observational cohort research [21].

This study was approved by the Bioethics Committee of the Jagiellonian University Medical College (decision No. 1072.6120.321.2021, dated December 15, 2021) and conducted in accordance with the Declaration of Helsinki. All patients provided written informed consent for surgical treatment and use of anonymized data for research.

Inclusion criteria were: (1) age ≥18 years; (2) magnetic resonance imaging-confirmed symptomatic lumbar disc herniation at L4–5 or L5–S1; (3) radicular pain persisting despite a minimum of 4 weeks of conservative management; (4) symptom duration ≤3 months; (5) availability of complete preoperative and follow-up Oswestry Disability Index (ODI) and visual analogue scale (VAS) data.

Exclusion criteria included previous lumbar surgery at the index level, degenerative spondylolisthesis or relevant central canal stenosis, nondegenerative etiologies such as infection or tumor, traumatic conditions, and severe comorbidities precluding general anesthesia or participation in rehabilitation.

Patients were assigned to either the rehabilitation group or the control group based solely on the availability of inpatient rehabilitation beds and patient preference. Importantly, allocation was not influenced by any clinical or radiological parameters, including herniation morphology, neurological status, degree of neural compression, or surgeon recommendation. No preferential assignment based on anticipated prognosis or socioeconomic status was practiced.

Because this pragmatic allocation strategy may introduce selection bias, all baseline variables were compared statistically, and adjusted analyses were performed to account for the observed imbalance in operated level. This approach reflects real-world clinical practice while ensuring methodological transparency.

All procedures were performed under general anesthesia with the patient in the prone position by surgeons experienced in full-endoscopic lumbar discectomy. Depending on disc level, anatomical configuration, and surgeon preference, a standardized full-endoscopic technique was used, consistent with previously described interlaminar and transforaminal approaches [2,18]. A standardized, full-endoscopic technique (Vertebris, Germany) with continuous irrigation was applied. Adequate decompression was confirmed by dural pulsation and free mobilization of the traversing nerve root. The distribution of surgical approaches between groups was recorded and compared for potential influence on outcomes.

Patients in the control group received standard postoperative care consisting of multimodal analgesia (nonsteroidal anti-inflammatory drugs and paracetamol, with pregabalin if required), wound care instructions, and activity recommendations. Early mobilization was encouraged, with short ambulation allowed on the day of surgery and progressive daily activity thereafter. Patients were advised to avoid lifting loads >5 kg and to limit continuous sitting to 45–60 minutes for the first 2 weeks. From the second postoperative week, patients were instructed in basic home-based isometric core activation exercises. Return to work was generally recommended after 4 to 6 weeks, depending on occupational demands. No supervised rehabilitation was provided.

Patients allocated to the rehabilitation group were admitted to the inpatient rehabilitation ward approximately 4 weeks after surgery and underwent a structured, standardized 2-week program. Each day included 2 supervised sessions of approximately 90 minutes under the guidance of physiotherapists specializing in postoperative spine rehabilitation.

The first week emphasized neuromuscular activation, gentle mobilization, and restoration of trunk control. This phase included multifidus and abdominal activation techniques, pelvic tilting, assisted mobilization using therapy balls and elastic bands, stretching of the hamstrings, hip flexors, and lumbar extensors, and adjunctive physical modalities such as transcutaneous electrical nerve stimulation and laser therapy.

The second week focused on dynamic lumbar stabilization, progressive strengthening, proprioceptive and balance training using unstable platforms, aerobic conditioning (treadmill walking or cycling), and functional movement strategies with ergonomic instruction. Each patient received an individualized home exercise plan at discharge.

All patients in both groups were prescribed a soft elastic lumbar orthosis for approximately 2 weeks after surgery. Compliance exceeded 90% as documented in outpatient follow-up records. No differences existed in orthosis protocol between groups.

The primary outcome was the ODI [22] measured at baseline, 6 weeks, and 12 months. Secondary outcomes included back and leg pain on the VAS [23], achievement of MCID, defined as an ODI improvement ≥10 points [24], and the incidence of complications or symptomatic recurrent disc herniation.

Outcome assessments were performed at baseline (preoperatively), at 6 weeks postoperatively, and at 12 months. The rehabilitation group underwent an additional prerehabilitation clinical evaluation at 4 weeks; however, because the control group did not receive a 4-week assessment, this time point was not used for between-group comparisons to avoid temporal bias.

Continuous variables were assessed for normality using the Shapiro-Wilk test and compared using the Student t-test or Mann-Whitney U-test as appropriate. Categorical variables were compared using the chi-square or Fisher exact test. Baseline characteristics were compared with p-values to detect potential group imbalances.

Given that operated level (L4–5 vs. L5–S1) differed significantly between groups, an analysis of covariance (ANCOVA) was performed with operated level as a covariate. Subgroup analyses stratified by level were conducted to evaluate whether treatment effects differed between L4–5 and L5–S1. Percentage changes in ODI and VAS were calculated to improve clinical interpretability.

Recurrence and complication rates were analyzed descriptively, with emphasis on timing relative to the rehabilitation program. A significance threshold of p<0.05 was applied. All analyses were conducted using IBM SPSS Statistics ver. 26.0 (IBM Co., USA).

A total of 102 patients met the inclusion criteria and were analyzed, of whom 49 were allocated to the rehabilitation group and 53 to the control group. Baseline demographic and clinical characteristics were generally similar between groups, except for a statistically significant difference in operated level distribution. Flow diagram illustrating patient eligibility, pragmatic allocation based on inpatient bed availability and patient preference (with no clinical allocation criteria), follow-up schedule, and inclusion in the outcome analysis. A total of 102 patients met inclusion criteria and were assigned to the rehabilitation group (n=49) or control group (n=53) (Figure 1). Early postoperative functional and pain recovery trajectories are presented in Figure 2.

The rehabilitation and control groups did not differ significantly in age, sex, baseline ODI, baseline VAS back pain, or baseline VAS leg pain. Operated level, however, differed between groups, with a higher proportion of L5–S1 cases in the control group (p=0.004). Because the early postoperative course may differ by level, adjusted and subgroup analyses were performed.

Given potential heterogeneity between interlaminar endoscopic lumbar discectomy (IELD) and transforaminal endoscopic lumbar discectomy (TELD) approaches, the distribution of surgical approach by group is provided in Table 1, and exploratory sensitivity analyses in approach-homogeneous subsets are presented in Supplementary Table 1. Surgical approach was closely linked to operated level in our cohort (all L5–S1 cases were IELD and the control group had no L4–5–IELD cases); therefore, results are presented as IELD-only (L5–S1) and TELD-only (L4–5) subsets, with the L4–5–IELD subgroup in the rehabilitation arm reported descriptively. The direction of the rehabilitation-associated benefit at 6 weeks was consistent within these approach-homogeneous subsets (Supplementary Table 1). Overall, the subgroup patterns were directionally consistent, with no evidence of effect modification by surgical approach.

The additional 4-week ODI and VAS assessment performed in the rehabilitation group before initiation of their inpatient program was not used for any between-group comparisons because the control group did not undergo this interim evaluation.

A substantially higher proportion of rehabilitation patients achieved the MCID (ODI improvement ≥10 points) at 6 weeks. MCID was achieved by 95.9% of rehabilitation patients and by 81.1% of controls, corresponding to an absolute difference of 14.8% (95% CI, 2.5–26.3) and a number needed to treat of approximately 7. By 12 months, both groups exceeded 95%, and no significant difference remained (Figure 3A).

Percentage of patients achieving the MCID (ODI improvement ≥10 points) at 6 weeks. The rehabilitation group demonstrated a markedly higher MCID rate. Values represent percentages (Figure 3B).

Comparison of symptomatic recurrent disc herniation between groups. No recurrences occurred during the inpatient rehabilitation period. Rates were low and not significantly different between groups. Values represent percentages.

Subgroup analyses stratified by L4–5 and L5–S1 demonstrated that the early benefit of rehabilitation remained consistent at both levels. Patients undergoing rehabilitation showed significantly greater reductions in ODI and VAS at 6 weeks in both subgroups (p<0.01 for L4–5 and p<0.05 for L5–S1). Interaction testing revealed no significant treatment-by-level interaction (p=0.41), indicating that the effect of rehabilitation was not dependent on the operated level.

Evidence on postoperative rehabilitation after lumbar disc surgery remains heterogeneous with respect to timing, duration, and delivery format [5-9]. Randomized trials have shown that structured, supervised physiotherapy can accelerate early functional improvement after lumbar decompression [5,8], although many studies have focused on microdiscectomy or mixed decompression cohorts rather than full-endoscopic techniques. In contrast, evidence specifically evaluating rehabilitation protocols after full-endoscopic lumbar discectomy is limited [16-18]. Large ELD series by Ruetten [18], Kim et al. (2020) [16] and Kim et al. (2007) [17] primarily emphasized surgical outcomes, recurrence, and technical factors, with little attention to standardized postoperative rehabilitation pathways.

Our findings extend the existing literature by suggesting that a consistent, short inpatient program after ELD may improve early outcomes without altering long-term recovery. A plausible explanation is that supervised rehabilitation provides structured exposure to progressive loading and movement re-education, which may improve neuromuscular control, reduce maladaptive movement patterns, accelerate proprioceptive recovery, and mitigate fear-avoidance behaviors—mechanisms repeatedly proposed as relevant during early postoperative recovery [10-14]. This may be particularly pertinent after ELD: while minimal soft-tissue disruption facilitates rapid mobilization, the absence of supervision during the early transition back to activity may increase the likelihood of suboptimal loading strategies.

The between-group differences observed at 6 weeks were modest in absolute ODI points but may still be clinically meaningful when considered against baseline disability. Because ODI values after ELD are typically already low at early follow-up, even a differential improvement of 5–6 points (-78.8% vs. -64.7%) can translate into noticeable differences in early function, comfort, and return to routine activities. The lack of a sustained difference at twelve months is consistent with prior rehabilitation studies in which supervised therapy accelerated early recovery but did not substantially change the final outcome trajectory at one year [5-9]. From a clinical perspective, the main benefit of inpatient rehabilitation after ELD may therefore be earlier recovery rather than superior ultimate results.

A frequent concern regarding early or intensive rehabilitation is that excessive loading could increase the risk of recurrence. In our cohort, no recurrence occurred during the supervised inpatient rehabilitation period; all recurrent herniations occurred after 6 weeks, and rates were low and comparable between groups. This pattern is consistent with endoscopic literature suggesting that recurrence is influenced predominantly by annular competence, herniation morphology, and patient factors rather than postoperative mobilization alone [15,18,25,26]. These observations support the feasibility of supervised, structured rehabilitation after ELD when initiated after the early healing phase, although prospective evaluation is warranted.

Although inpatient rehabilitation increases short-term direct costs, earlier functional gains may generate indirect benefits, including shorter disability periods, earlier return to work, and improved patient satisfaction. Prior economic analyses indicate that accelerated postoperative recovery can reduce broader societal costs [9]. While formal cost-effectiveness evaluation was outside the scope of this study, it remains an important direction for future work, particularly in health systems where rehabilitation resources and inpatient capacity are constrained.

Several limitations should be considered. First, the retrospective design and nonrandom allocation based on bed availability and patient preference introduce potential selection bias and residual confounding. Participants opting for inpatient rehabilitation may have differed in motivation, socioeconomic factors, occupational demands, or perceived need for structured recovery. Although baseline characteristics were largely comparable (aside from operated level) and we attempted to mitigate confounding using ANCOVA and level-stratified subgroup analyses, unmeasured confounders cannot be excluded. Second, an interim 4-week assessment was available only in the rehabilitation group and was not used for between-group comparisons; therefore, all primary analyses were restricted to matched time points (baseline, 6 weeks, and 12 months). Third, while the sample size was sufficient to detect early functional differences, the study was not designed to robustly assess low-frequency outcomes such as recurrence or complications. Finally, this was a single-center study, which may limit generalizability. Prospective randomized trials incorporating standardized rehabilitation protocols, broader outcome assessment, and cost-effectiveness analyses are warranted to better define the role of structured inpatient rehabilitation after ELD.