How to assess the long-term recovery outcomes of patients with cauda equina syndrome before surgery: a retrospective cohort study

Abstract

Background:

Factors influencing recovery after decompression surgery for cauda equina syndrome (CES) are not completely identified. The authors aimed to investigate the most valuable predictors (MVPs) of poor postoperative recovery (PPR) in patients with CES and construct a nomogram for discerning those who will experience PPR.

Methods:

Three hundred fifty-six patients with CES secondary to lumbar degenerative diseases treated at Xijing Hospital were randomly divided into training (N=238) and validation (N=118) cohorts at a 2:1 ratio. Moreover, 92 patients from the 970^th Hospital composed the testing cohort. Least Absolute Shrinkage and Selection Operator regression (LASSO) was used for selecting MVPs. The nomogram was developed by integrating coefficients of MVPs in the logistic regression, and its discrimination, calibration, and clinical utility were validated in all three cohorts.

Results:

After 3 to 5 years of follow-up, the residual rates of bladder dysfunction, bowel dysfunction, sexual dysfunction, and saddle anesthesia were 41.9, 44.1, 63.7, and 29.0%, respectively. MVPs included stress urinary incontinence, overactive bladder, low stream, difficult defecation, fecal incontinence, and saddle anesthesia in order. The discriminatory ability of the nomogram was up to 0.896, 0.919, and 0.848 in the training, validation, and testing cohorts, respectively. Besides, the nomogram showed good calibration and clinical utility in all cohorts. Furthermore, the optimal cutoff value of the nomogram score for distinguishing those who will experience PPR was 148.02, above which postoperative outcomes tend to be poor.

Conclusion:

The first pretreatment nomogram for discerning CES patients who will experience PPR was developed and validated, which will aid clinicians in clinical decision-making.

Introduction

Highlights

• There is a lack of pretreatment tools to predict long-term recovery outcomes after surgery for cauda equina syndrome (CES).

• We unveiled the most valuable predictors of poor postoperative recovery (PPR) in patients with CES.

• We created the first pretreatment nomogram for discerning patients who will experience PPR, which may aid in clinical decision-making.

Cauda equina syndrome (CES) is a rare condition affecting the nerve tracts at the end of the spinal cord, with a significant impact on patients’ quality of life^1,2. An annual incidence rate of up to 2.7 per 100 000 has been reported for CES³. Often representing a medical emergency, CES symptoms require immediate attention⁴. Studies with long-term follow-up after surgery for CES have shown that residual bowel, bladder, or sexual dysfunction occurred in at least 1/3 of patients⁵. Lumbar disc herniation is the most common cause of CES, accounting for 45% of the cases⁶. Even under mild mechanical compression, patients with pre-existing lumbar spinal stenosis often develop CES due to the reduced space in the lumbosacral canal⁷. However, due to prolonged symptoms in this cohort, the nerves are prone to tolerance⁸. Also, mild symptoms may result in delaying treatment, resulting in potentially avoidable long-term disabilities, causing irreversible psychological trauma to patients^2,9,10. Moreover, untimely decompression surgery often results in legal implications for clinicians⁷.

Despite conflicting evidence on the link between surgical timing and functional outcomes, there is strong consensus among spine surgeons who favor emergency decompression surgery for CES¹¹. However, predicting postoperative recovery in cauda equina nerve injury is largely based on clinical reports or the speculation of medical professionals; therefore, providing patients with an accurate prognosis is a challenge. Furthermore, it is difficult to establish postoperative factors influencing recovery and even more challenging to develop predictive models owing to studies with small sample sizes in the literature^12–14. The Xijing Hospital (Xi’an, China) serves a large number of CES cases and is considered a Class Three referral medical center for multiple northern provinces. Therefore, it is possible to establish prediction models for CES postoperative outcomes based on extensive clinical data.

Currently, obtaining valid and authentic information through telephone follow-ups has been deemed effective and scalable¹⁵. Patient self-assessment using validated scales for CES has become one of the primary methods of collecting postoperative information and has been widely applied in clinical research^16,17. Therefore, using patient self-assessment, our study aimed to analyze the residual rate of specific symptoms after decompression surgery and factors affecting poor postoperative recovery (PPR) in CES secondary to lumbar degenerative diseases (LDDs; lumbar disc herniation and lumbar spinal stenosis). More importantly, we aimed to construct a pretreatment nomogram to discern who will experience PPR to help clinical decision-making before surgery.

Materials and methods

Study design and population

Data of patients with CES secondary to LDDs diagnosed at Xijing Hospital (Xi’an, China) between January 2011 and November 2020 and those at the 970^th Hospital (Yantai, China) between January 2017 and November 2020 were retrospectively collected. This study was performed according to the Helsinki Declaration and was approved by The Ethics Committee of Xijing Hospital (20232023-C-1). In addition, this retrospective study was registered with ResearchRegistry.com (Unique identification Number: researchregistry9743). Data has been reported in line with strengthening the reporting of cohort, cross-sectional, and case–control studies in surgery (STROCSS) 2021 criteria¹⁸ (STROCSS Guideline Checklist, Supplemental Digital Content 1, http://links.lww.com/JS9/C196).

The inclusion criteria were (1) CES secondary to LDDs, including lumbar disc herniation and lumbar spinal stenosis; (2) lower limb neurogenic deficits (motor/sensory loss or reflex changes), as well as at least one of the following: bladder dysfunction, bowel dysfunction, sexual dysfunction, or saddle anesthesia symptoms; (3) has completed lumbar decompression surgery; and (4) availability of preoperative imaging data (MRI). The exclusion criteria were (1) CES caused by lumbar vertebral fractures, tumors, deformities, or infections; (2) concomitant cervical or thoracic lesions; (3) other conditions/factors affecting urinary, bowels, or sexual function, such as prostate hyperplasia or cancer radiotherapy; (4) Other diseases that cause sensorimotor disorders of the lower limbs, such as diabetic foot or peripheral neuropathy; (5) long-term use of psychotropic or other drugs causing abnormal defecation; (6) history of spinal, urological, or gynecological surgery; (7) patients with missing imaging data or those not agreeing to follow-up; (8) patients who did not provide written informed consent. Finally, 356 patients from Xijing Hospital were randomly divided into training and validation cohorts at a 2:1 ratio, and 92 patients from 970th Hospital were collected as the testing group.

Neurological recovery is a long-term process, and reports have indicated that further recovery from cauda equina injury requires 3–5 years postsurgery^2,19. Therefore, our study set the postoperative follow-up period as 3–5 years. In this study, all surgeries were performed by experienced orthopedic surgeons with associate senior titles or above. All patients underwent lumbar decompression surgery involving discectomy or laminectomy with fusion, and the operative segment was from L2 to S1. We defined the responsible level as the level with the largest disc herniation or most severe spinal canal stenosis on imaging. The dural sac cross-sectional area (DSCSA) of the responsible level was measured on axial MRI images using Image J digital image viewing software (National Institutes of Health). Two orthopedic surgeons measured the DSCSA, and the interclass correlation coefficient for intra-assessor and interassessor reliability was 0.934 and 0.873, respectively.

Preoperative clinical symptom assessment

We collected potential predictive factors from all patients prior to surgery, including age, sex, height, weight, smoking, and alcohol consumption history, underlying diseases, duration of lower limb symptoms, duration of CES symptoms, etiology type, DSCSA, visual analog scale for pain (VAS; lumbar and lower limb), bladder function, bowel function, sexual function, saddle anesthesia, radiating leg pain, reduced lower limb sensation, lower limb motor disturbances, passive straight leg raising test results, and bone mass. Specifically, bladder dysfunction was further categorized into stress urinary incontinence (e.g. incontinence upon coughing), overactive bladder (e.g. frequency, urgency, or nocturia), and low stream (e.g. abdominal pressure during urination or catheterization). Bowel dysfunction was categorized into difficult defecation (constipation or tenesmus) and fecal incontinence. Saddle anesthesia was classified as none, partial, and complete based on the extent. Motor disturbances were defined as any lower limb strength below Grade 4. We utilized the WHO-based definitions for osteoporosis (T-score <-2.5), osteopenia (-2.5≤T-score≤-1), and normal bone density (T-score>-1)²⁰.

Postoperative CES symptom questionnaire

During the follow-up period, the two researchers, who received uniform language and expression training, conducted telephone follow-up examinations of the patients without knowing the patient’s preoperative data. Patients were asked to complete three questionnaires that included scales on bladder, bowel, and sexual function. This would allow self-assessment of residual symptoms of cauda equina nerve injury. The Urinary Symptoms Profile (USP) scale was used to assess bladder function, with a score ≥1 indicating dysfunction²¹. The Neurogenic Bowel Dysfunction (NBD) score was utilized to evaluate bowel function, with a score ≥6 indicating dysfunction²². Sexual function was evaluated using the Arizona Sexual Experiences Scale (ASEX) to assess sexual function, where a score ≥19, one item scoring ≥5, or three items scoring ≥4 were considered sexual dysfunction²³. Higher scores indicated worsening dysfunction. PPR in CES was defined as the presence of bladder dysfunction and/or bowel dysfunction in patients with or without sexual dysfunction⁹. Due to the diverse factors causing sexual dysfunction and its high incidence in the Asian population²⁴, patients with only one positive symptom of sexual dysfunction were excluded from the PPR cohort.

Statistics

The normal distribution of the continuous variables was tested using the Kolmogorov–Smirnov test. Those that conformed to the normal distribution were represented by mean and SD, and the data were compared by independent sample t-test. Those that did not conform to the normal distribution were represented by the median and interquartile range, and the Mann–Whitney U test was used for data comparison. Categorical variables were expressed as frequency and percentage. Between-group categorical data were compared using the χ² test or Fisher’s exact test.

The Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis with 10-fold cross-validation was used to reduce data dimensionality and select the most valuable predictors (MVPs)²⁵. The nomogram was constructed by integrating all MVPs according to their regression coefficients in the binary logistic regression. Subsequently, accompanied by the Hosmer–Lemeshow test, the calibration curve was used to evaluate the calibration of the nomogram. The discrimination of the nomogram was evaluated using the area under the receiver operating characteristic curve (AUC)²⁶ and concordance index (C-index). Decision curve analysis (DCA) was used to determine the net benefit of the nomogram for clinical decisions²⁷.

A two-tailed P-value less than 0.05 was considered significant. The Statistical Package for Social Sciences (SPSS) 26 software and R 4.2.3 software (R Foundation for Statistical Computing) were used for the complete statistical analyses.

Results

General characteristics

The study flowchart is shown in Figure 1. There were 238, 118, and 92 cases in the training, validation, and testing cohorts, respectively. In the three datasets, 61 patients (25.6%), 26 patients (22.0%), and 29 patients (31.5%) experienced PPR. Supplementary Figure 1 (Supplemental Digital Content 2, http://links.lww.com/JS9/C197) shows the specific numerical values from the follow-up scales. After analyzing data from the 356 patients from Xijing Hospital Hospital, we found that at the last follow-up, the residual rates of CES symptoms were 41.9% (67/160) for bladder function, 44.1% (56/127) for bowel dysfunction, 63.7% (109/171) for sexual dysfunction, and 29.0% (80/276) for saddle anesthesia. A chord Diagram was used to depict the outcomes of bladder dysfunction sub-cohort (stress urinary incontinence, overactive bladder, and low stream) and bowel dysfunction sub-cohort (difficult defecation and fecal incontinence; Supplementary Figure 2, Supplemental Digital Content 3, http://links.lww.com/JS9/C198).

Figure 1.

Flowchart for patient selection. CES, cauda equina syndrome; LDDs, lumbar degenerative diseases.

Demographic and clinical characteristics of the three cohorts are summarized in Table 1. Baseline characteristics were generally similar between the training and validation cohorts.

Table 1.

General characteristics.

		CES Postoperative recovery, N (%)			Cohort, N (%)
Variables	The study cohort (training and validation, n=356), N (%)	Poor recovery (n=87)	Complete recovery (n=269)	P	Training (n=238)	Validation (n=118)	P	External testing (n=92)
Age, years				0.181			0.603
<40	87 (24.4)	16 (18.4)	71 (26.4)		62 (26.1)	25 (21.2)		21 (22.8)
40–60	191 (53.7)	47 (54.0)	144 (53.5)		125 (52.5)	66 (55.9)		35 (38.0)
>60	78 (21.9)	24 (27.6)	54 (20.1)		51 (21.4)	27 (22.9)		36 (39.1)
Sex				0.032			0.447
Female	161 (45.2)	48 (55.2)	113 (42.0)		111 (46.6)	50 (42.4)		35 (38.0)
Male	195 (54.8)	39 (44.8)	156 (58.0)		127 (53.4)	68 (57.6)		57 (62.0)
BMI, kg/m2 (mean±SD)	24.17±2.80	24.51±2.77	24.07±2.80	0.204	24.21±2.86	24.11±2.68	0.775	24.13±3.07
Follow-up time, months (range)	46 (36–64)	46 (36–59)	47 (36–64)	0.379	46.5 (36–64)	47.5 (37–63)	0.162	46 (36–59)
Smoking	80 (22.5)	19 (21.8)	61 (22.7)	0.871	54 (22.7)	26 (22.0)	0.889	31 (33.7)
Alcohol use	34 (9.6)	7 (8.0)	27 (10.0)	0.583	22 (9.2)	12 (10.2)	0.780	13 (14.1)
Diabetes	19 (5.3)	7 (8.0)	12 (4.5)	0.269 F	11 (4.6)	8 (6.8)	0.394	6 (6.5)
Hypertension	71 (19.9)	24 (27.6)	47 (17.5)	0.040	49 (20.6)	22 (18.6)	0.666	23 (25.0)
Root symptoms time, weeks				0.246			0.491
<6	116 (32.6)	22 (25.3)	94 (34.9)		73 (30.7)	43 (36.4)		29 (31.5)
6–12	53 (14.9)	14 (16.1)	39 (14.5)		35 (14.7)	18 (15.3)		16 (17.4)
>12	187 (52.5)	51 (58.6)	136 (50.6)		130 (54.6)	57 (48.3)		47 (51.1)
Duration of CES, days				<0.001			0.259
<2	31 (8.7)	10 (11.5)	21 (7.8)		16 (6.7)	15 (12.7)		10 (10.9)
2-7	57 (16.0)	18 (20.7)	39 (14.5)		41 (17.2)	16 (13.6)		19 (20.7)
7-30	173 (48.6)	24 (27.6)	149 (55.4)		116 (48.7)	57 (48.3)		27 (29.3)
>30	95 (26.7)	35 (40.2)	60 (22.3)		65 (27.3)	30 (25.4)		36 (39.1)
VAS Back (mean±SD)	4.86±1.84	4.79±1.76	4.88±1.87	0.596	4.87±1.90	4.84±1.71	0.849	4.37±1.46
VAS Leg (mean±SD)	5.14±2.07	5.10±2.41	5.16±1.95	0.743	5.10±2.19	5.23±1.80	0.765	4.79±1.81
Incontinence	83 (23.3)	54 (62.1)	29 (10.8)	<0.001	60 (25.2)	23 (19.5)	0.230	23 (25.0)
Overactive bladder	107 (30.1)	61 (70.1)	46 (17.1)	<0.001	77 (32.4)	30 (25.4)	0.180	30 (32.6)
Low stream	100 (28.1)	52 (59.8)	48 (17.8)	<0.001	67 (28.2)	33 (28.0)	0.971	28 (30.4)
Difficult defecation	86 (24.2)	39 (44.8)	47 (17.5)	<0.001	61 (25.6)	25 (21.2)	0.356	26 (28.3)
Fecal incontinence	44 (12.4)	26 (29.9)	18 (6.7)	<0.001	32 (13.4)	12 (10.2)	0.377	13 (14.1)
Saddle anesthesia				<0.001			0.986
None	80 (22.5)	17 (19.5)	63 (23.4)		54 (22.7)	26 (22.0)		21 (22.8)
Partial	205 (57.6)	35 (40.2)	170 (63.2)		137 (57.6)	68 (57.6)		39 (42.4)
Complete	71 (19.9)	35 (40.2)	36 (13.4)		47 (19.7)	24 (20.3)		32 (34.8)
Sexual dysfunction	171 (48.0)	71 (81.6)	100 (37.2)	<0.001	116 (48.7)	55 (46.6)	0.705	42 (45.7)
Radiculalgia				<0.001			0.433
None	56 (15.7)	19 (21.8)	37 (13.8)		35 (14.7)	21 (17.8)		10 (10.9)
Unilateral	217 (61.0)	36 (41.4)	181 (67.3)		143 (60.1)	74 (62.7)		51 (55.4)
Bilateral	83 (23.3)	32 (36.8)	51 (19.0)		60 (25.2)	23 (19.5)		31 (33.7)
Sensitive deficit				0.002			0.476
None	44 (12.4)	17 (19.5)	27 (10.0)		31 (13.0)	13 (11.0)		11 (12.0)
Unilateral	226 (63.5)	42 (48.3)	184 (68.4)		154 (64.7)	72 (61.0)		46 (50.0)
Bilateral	86 (24.2)	28 (32.2)	58 (21.6)		52 (22.3)	33 (28.0)		35 (38.0)
Motor deficit				0.164			0.517
None	205 (57.6)	48 (55.2)	157 (58.4)		140 (58.8)	65 (55.1)		46 (50.0)
Unilateral	113 (31.7)	25 (28.7)	88 (32.7)		71 (29.8)	42 (35.6)		30 (32.6)
Bilateral	38 (10.7)	14 (16.1)	24 (8.9)		27 (11.3)	11 (9.3)		16 (17.4)
Passive straight leg raising test				0.044			0.521
None	126 (35.4)	40 (46.0)	86 (32.0)		82 (34.5)	44 (37.3)		31 (33.7)
Unilateral	186 (52.2)	36 (41.4)	150 (55.8)		129 (54.2)	57 (48.3)		43 (46.7)
Bilateral	44 (12.4)	11 (12.6)	33 (12.3)		27 (11.3)	17 (14.4)		18 (19.6)
Bone mass				0.713			0.119
normal	136 (38.2)	32 (36.8)	104 (38.7)		94 (39.5)	42 (35.6)		34 (37.0)
osteopenia	172 (48.3)	45 (51.7)	127 (47.2)		107 (45.0)	65 (55.1)		37 (40.2)
osteoporosis	48 (13.5)	10 (11.5)	38 (14.1)		37 (15.5)	11 (9.3)		21 (22.8)
Image responsibility level				0.715 F			0.526 F
L2/3	8 (2.2)	3 (3.4)	5 (1.9)		5 (2.1)	3 (2.5)		3 (3.3)
L3/4	24 (6.7)	5 (5.7)	19 (7.1)		17 (7.1)	7 (5.9)		6 (6.5)
L3-5	7 (2.0)	2 (2.3)	5 (1.9)		3 (1.3)	4 (3.4)		10 (10.9)
L4/5	196 (55.1)	51 (58.6)	145 (53.9)		137 (57.6)	59 (50.0)		48 (52.2)
L4-S1	13 (3.7)	4 (4.6)	9 (3.3)		9 (3.8)	4 (3.4)		2 (2.2)
L5/S1	108 (30.3)	22 (25.3)	86 (32.0)		67 (28.2)	41 (34.7)		23 (25.0)
Etiology type				0.061			0.293
LDH	257 (72.2)	201 (74.7)	56 (64.4)		176 (73.9)	81 (68.6)		65 (70.7)
LSS	99 (27.8)	68 (25.3)	31 (35.6)		62 (26.1)	37 (31.4)		27 (29.3)
DSCSA, mm2
Mean±SD	71.44±25.70	70.32±24.28	71.80±26.18	0.640	71.13±25.76	72.05±25.68	0.751	72.00±27.97
Median (IQR)	68.82 (52.54–87.35)	68.58 (52.73–88.60)	70.82 (51.88–86.70)	0.747	70.65 (52.69–83.60)	67.24 (51.55–93.30)	0.832	71.58 (53.61–92.19)

Bold indicates statistical significance.

CES, Cauda equina syndrome; DSCSA, Dural sac cross-sectional area; F, Fisher’s exact test; IQR, Interquartile range; LDH, Lumbar disc herniation; LSS, Lumbar spinal stenosis; VAS, Visual analog scale for pain.

Identification of optimal predictive factors for poor postoperative CES outcomes

Using LASSO regression with 10-fold cross-validation, the optimized hyperparameter λ was identified based on the bivariate deviation. Based on the optimal λ, six nonzero coefficients of preoperative features were selected as MVPs for PPR, in order: stress urinary incontinence, overactive bladder, low stream, difficult defecation, fecal incontinence, and saddle anesthesia (Fig. 2).

Figure 2.

Feature selection using the least absolute shrinkage and selection operator (LASSO) analysis with 10-fold cross-validation. Lambda (tuning parameter) selection of deviance in the LASSO regression based on the one standard error criteria (right dotted line) and the minimum criteria (left dotted line) (A). LASSO coefficient profiles of the candidate features. The intersecting curves represent the number of features retained at that log (lambda) value, and six predictors with nonzero coefficients were selected according to the one standard error criteria (B). Note. a, stress urinary incontinence; b, overactive bladder; c, low stream; d, difficult defecation; e, fecal incontinence; f, saddle anesthesia; g, gender; h, diabetes; i, duration of CES symptoms; j, smoking; k, bone mass.

In the training cohort, the AUCs (95% CI) of the six MVPs were 0.749 (0.671–0.828), 0.767 (0.694–0.841), 0.696 (0.615–0.778), 0.647 (0.563–0.732), 0.608 (0.520–0.696), and 0.590 (0.501–0.680), respectively. In the validation cohort, the AUCs (95% CI) of the six MVPs were 0.770 (0.650–0.889), 0.756 (0.640–0.873), 0.740 (0.623–0.857), 0.611 (0.481–0.741), 0.632 (0.499–0.766), and 0.682 (0.552–0.812), respectively. In the testing cohort, the AUCs (95% CI) of the six MVPs were 0.670 (0.544–0.796), 0.690 (0.568–0.812), 0.706 (0.585–0.827), 0.646 (0.519–0.773), 0.598 (0.467–0.729), and 0.606 (0.475–0.736), respectively.

Establishment of the nomogram

Multifactor logistic regression confirmed that overactive bladder, low stream, and difficult defecation were independent predictors of PPR in CES. A forest plot of odds ratios was created to visualize the logistic regression analysis results (Supplementary Figure 3, Supplemental Digital Content 4, http://links.lww.com/JS9/C199). According to the coefficients of the six items (1.302, 1.339, 1.748, 1.171, 1.133, and 1.414), the nomogram was generated using the ‘rms’ package of R software (Fig. 3). The length of the lines in the nomogram reflected the importance of predictors. Low stream exhibited the highest weight, indicating it as the most significant factor impacting PPR in CES, followed by complete saddle anesthesia and overactive bladder. In contrast, fecal incontinence was the least influential factor. The low stream was set at 100 points in the nomogram. Based on the ratio of regression coefficients, the scores for stress urinary incontinence, overactive bladder, low stream, difficult defecation, fecal incontinence, and saddle anesthesia were 74.47, 76.65, 100, 66.91, 64.80, and 81.11 (partial saddle anesthesia was 34.28), respectively. When using the nomogram to predict the risk of PPR in CES patients, the total score can be calculated by summing up the six preoperative scores of the corresponding symptoms. Subsequently, the risk of PPR can be determined by extending a downward vertical line from the total score.

Figure 3.

Nomogram for predicting the risk of poor postoperative CES outcomes.

Multiple validations of the nomogram

In the internal validation using the training cohort, the nomogram presented a high accuracy in estimating PPR in CES patients after surgery, with an AUC of 0.896 (95% CI: 0.852–0.939; Fig. 4A). The result of the Hosmer–Lemeshow test (χ²=7.055, P=0.217) and a close-to-ideal calibration curve indicated good calibration of the nomogram (Fig. 4B). DCA showed significantly higher net benefits than the six MVPs (Fig. 4C).

Figure 4.

Evaluation of the nomogram. The ROC curve (A), the calibration plot (B), and the decision curve analysis (C) of the nomogram in the training cohort. The ROC curve (D), the calibration plot (E), and the decision curve analysis (F) of the nomogram in the validation cohort. The ROC curve (H), the calibration plot (I), and the decision curve analysis (G) of the nomogram in the testing cohort. In calibration plots, the calibration curve is expected to fall along the ideal line corresponding to a perfectly calibrated nomogram; the solid purple line represents the apparent accuracy of the nomogram without correction for overfitting, while the solid black line represents the bootstrap-corrected nomogram. In DCA plots, ‘all’ refers to the assumption that all patients experienced PPR, while ‘none’ assumes that no patient experienced PPR. Note. a, stress urinary incontinence; b, overactive bladder; c, low stream; d, difficult defecation; e, fecal incontinence; f, saddle anesthesia. AUC, area under the curve.

In the validation cohort involving 118 patients, the AUC of the nomogram was 0.919 (95% CI: 0.863–0.974; Fig. 4D), indicating outstanding accuracy. Apart from the results of the Hosmer–Lemeshow test (χ²=4.988, P=0.289), a strong consistency between observed and predicted values further verified the excellent calibration of the nomogram (Fig. 4E). Similarly, DCA demonstrated the clinical utility of the nomogram better than MVPs (Fig. 4F).

In the testing cohort, the Hosmer–Lemeshow test (χ²=3.658, P=0.723) and an AUC of 0.848 (95% CI: 0.758–0.937; Fig. 4H) indicated good calibration and accuracy, respectively. Additionally, the calibration curve displayed close alignment between the observed and predicted probabilities (Fig. 4I). DCA presented conclusions similar to those of the training and validation cohorts (Fig. 4J).

Determination of the optimal cutoff value for the nomogram

Based on data from all 448 patients enrolled in this study, relevant parameters such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), negative likelihood ratio (LR−), positive utility index (UI+), negative utility index (UI−), diagnostic odds ratio (DOR), and Youden index under different nomogram cutoff points are listed in Table 2. As per the rule of the most balance between sensitivity and specificity, the optimal cutoff value of the nomogram was identified as 148.02 (sensitivity 0.853 and specificity 0.774). In other words, patients with a nomogram score of ≥148.02 were identified as high-risk PPR patients and recommended for early repair of cauda equina injuries. Additionally, clinicians may need to develop individualized postoperative rehabilitation plans and provide more comprehensive psychological support.

Table 2.

Parameters for evaluating the performance of nomogram total score at different cutoff points.

Cutoff points	Sensitivity	Specificity	PPV	NPV	LR+	LR-	DOR	UI+	UI-	Youden’s index
100	0.922	0.660	0.486	0.961	2.710	0.118	23.041	0.449	0.634	0.582
101.19	0.914	0.711	0.525	0.959	3.160	0.121	26.058	0.480	0.682	0.625
110.93	0.888	0.738	0.542	0.950	3.388	0.152	22.312	0.481	0.701	0.626
131.71	0.879	0.744	0.545	0.946	3.434	0.162	21.171	0.480	0.704	0.623
134.28	0.879	0.747	0.548	0.947	3.475	0.162	21.510	0.482	0.707	0.626
141.45	0.853	0.768	0.563	0.938	3.680	0.191	19.286	0.480	0.720	0.622
145.91	0.853	0.771	0.566	0.938	3.728	0.190	19.616	0.483	0.723	0.625
148.02	0.853	0.774	0.569	0.938	3.778	0.189	19.955	0.486	0.726	0.628
151.12	0.828	0.780	0.568	0.928	3.764	0.221	17.030	0.470	0.724	0.608
157.76	0.802	0.822	0.612	0.922	4.511	0.241	18.710	0.491	0.758	0.624
165.99	0.802	0.825	0.616	0.923	4.589	0.240	19.102	0.494	0.761	0.627
166.91	0.793	0.825	0.613	0.919	4.540	0.251	18.109	0.486	0.759	0.618
176.65	0.767	0.831	0.614	0.911	4.549	0.280	16.246	0.471	0.757	0.599

PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR−, negative likelihood ratio; UI+, positive utility index; UI-, negative utility index; DOR, diagnostic odds ratio.

Discussion

This study used the largest available sample size to date to analyze long-term postoperative outcomes in CES, establishing the first pretreatment tool to discern patients who will experience PPR. The main advantage of our model is that it allows physicians in any resource-limited hospital to use history and physical examinations to predict the long-term postoperative recovery of patients with CES undergoing surgery. We did not include variables, such as urodynamic data or residual urine volume, which requires additional testing, like invasive studies or ultrasound measurement of bladder, respectively. In addition to placing a physical and economic burden on patients, the effectiveness of such measurement tools varies depending on the literature⁵. Measurement of the bladder residual urine volume may often provide inaccurate information; nearly 60% of patients with confirmed CES have less residual urine than the currently recognized diagnostic threshold of 200 ml^28,29. Hellstrom and Podnar et al. demonstrated a significant variability between the results of invasive urodynamic testing and patients’ actual perception of urinary symptoms^30,31. In other words, this study confirmed that medical history and signs cannot be ignored, which is conducive to diagnosing CES and highlights the high value of predicting postoperative outcomes.

Bladder dysfunction was subdivided into three parts in our study, and we found that it occurred mainly in the overactive bladder region (66.9%), with fewer functional impairments associated with stress urinary incontinence (51.9%) or low stream (62.5%). The dysfunction may be attributed to continuous nerve damage in CES, primarily affecting detrusor muscle innervation and function rather than pelvic floor strength or urethral patency³². Urinary symptoms vary based on muscle impairments in which sphincter dysfunction causes stress urinary incontinence, and detrusor reflex relaxation causes urinary retention³³. Kennedy et al.³⁴ confirmed preoperative bladder dysfunction as a factor affecting PPR after surgery for CES and highlighted a significant association between the severity of preoperative bladder dysfunction and overall residual functional impairments. Chau et al.³⁵ reported that retention-type CES (CESR), representing the degree of bladder dysfunction, was considered a predictive factor for PPR after surgery, which aligns with the results of this study. CESR is mainly manifested as overflow urinary incontinence synonymous with ‘low stream’ in this study and was the highest contributor in our nomogram.

Spinal cord or cauda equina nerve injury leads to bowel dysfunction, including intestinal obstruction, constipation (both showing prolonged defecation time), and fecal incontinence (passive leakage)³⁶. While fecal incontinence is a typical symptom of CES, constipation is rarely mentioned¹⁰. Constipation can occur due to various factors, such as age, sex, medication, psychological aspects, diet, daily activities, and lumbar or leg pain, which makes the symptoms less reliable for diagnosing CES¹⁶. However, constipation is not incidental to CES and has been reported as such in various studies^37,38. Constipation and fecal incontinence, as two different manifestations of neurogenic bowel dysfunction, can be explained physiologically and pathologically. The pacemaker and neural network within the intestinal wall involved in regulating the expulsion of feces are transmitted by the intrinsic nervous system. The intrinsic nervous system simultaneously controls the rectal sigmoid colon region; once the S2-4 nerves are damaged, the feces transit time is significantly prolonged, leading to constipation or difficulty in defecation^36,39. Conversely, the external anal sphincter is innervated by the pudendal nerve of the somatic nervous system, which controls voluntary defecation. Damage to the pudendal nerve due to cauda equina injury often results in fecal incontinence and saddle anesthesia^36,39. The different nerve sources cause varied presentations of neurogenic bowel dysfunction. Unfortunately, although some studies have confirmed that bowel dysfunction is one of the factors affecting PPR after CES, the predictive value of constipation and urinary incontinence has not been discussed separately due to the small sample size and univariate analysis^34,40. Our study separated constipation and fecal incontinence as factors impacting postoperative recovery after CES surgery and confirmed the important predictive value of both, presenting a fresh concept for prognostic research of CES patients.

Saddle anesthesia has been confirmed as a predictive factor affecting PPR after surgery for chronic CES^9,41. Our study categorized saddle anesthesia into normal, partial numbness, and complete anesthesia. We found that patients with complete anesthesia were more likely to experience PPR, consistent with the study’s results by Kennedy et al.³⁴ matches. We did not perform a rectal examination during the study, as there is no relationship between rectal examination results and the diagnosis of cauda equina nerve compression⁴². Furthermore, on digital rectal examination, external anal sphincter abnormality occurs in all patients with diminished perineal sensation, whereas the negative predictive value is only 19%, which makes it unnecessary to test anal sphincter tension in patients with reduced perineal sensation, as it may cause unnecessary discomfort¹².

Information on sexual function was not available from the medical records of several patients in whom preoperative sexual function status was collected during follow-up through recollection, which may have caused recall bias (21.7%, 97/448, with preoperative sexual function records). Hence, although univariate analysis identified preoperative sexual dysfunction as a factor impacting PPR after surgery for CES, it was not included in the final model. It is undeniable that incomplete sexual function records are the major obstacle affecting current studies assessing sexual function in CES¹². At the last follow-up, our study reported sustained sexual dysfunction in 63.7% of patients, higher than that (40.1%) reported in 22 studies with an average follow-up period of 3 years¹⁹. Additionally, due to differences in ethnicity and region, the decline in sexual function naturally occurs at an earlier in the Asian population compared to the Western European population²⁴. Therefore, to facilitate cross-ethnic verification of this nomogram by Western scientists in the future, we excluded sexual dysfunction from the model.

In our study, the decompression time was not found to have a predictive value in postoperative recovery, indicating that the duration of compression surgery for CES may not be a determining factor of postoperative recovery. Jha et al.¹³ found that decompression time could not predict ultimate micturition outcomes, but the longer the decompression time, the longer the complete recovery. Moreover, in an experimental study in canines, immediate (2–3 s), early (1–6 h), delayed (24 h), or 1-week decompression surgery did not significantly alter the recovery of the nerve function in the cauda equina⁴³. Qureshi et al.¹¹ found that there were no significant differences in long-term postoperative measurements among patients who underwent surgery for CES within 24 h of symptom onset, between 24 and 48 h, and after 48 h. We further analyzed 269 patients who recovered completely at the Xijing Hospital and found a weak positive correlation between the CES recovery time and the duration of preoperative symptoms after excluding the influence of postoperative care and complications. (Supplementary Table 1, Supplemental Digital Content 5, http://links.lww.com/JS9/C200 and Supplementary Figure 4, Supplemental Digital Content 6, http://links.lww.com/JS9/C201). This indicates that patients who undergo early decompression recover faster, whereas delayed compression only extends the recovery time rather than affecting the extent of recovery.

In this study, our nomogram performed well in predicting ‘poor’ postoperative recovery outcomes in CES patients, with good accuracy, consistency, calibration, and net benefit across training, validation, and external testing cohorts. In addition, we included variables that were rarely used in other studies, such as overactive bladder, bowel difficulties, and incontinence. This tool allows us to better identify patients before surgery who are at risk of incomplete recovery after surgery. For example, if a CES patient has stress urinary incontinence, overactive bladder, and low stream, but no bowel dysfunction or saddle anesthesia, the nomogram score is 251.12 and the risk of poor postoperative recovery is approximately 60%. In addition, depending on whether the patients in the model have symptoms, the total score of the nomogram is calculated and the patients are divided into the low-risk group (nomogram score ≤148.5) and the high-risk group (nomogram score >148.5). For high-risk patients, neurotherapy was carried out in advance, such as embedding light irradiation probes in the spinal canal during surgery. Especially for patients who previously had to wait a long time after surgery to confirm poor recovery results, nerve repair treatment can now be performed in advance without missing the golden period of nerve repair after surgery.

Our study has several strengths. (1) A large sample size ensured the reliability of the conclusions drawn ; (2) the nomogram was tested using independent cohorts from different centers, strengthening its reliability and generalizability to a large extent; (3) based on the nomogram results, we provided the best cutoff point for the nomogram to help clinicians decide on additional neurological treatments; and (4) the selection of the training and external testing cohorts was based on populations from western and eastern China, likely representing a good sample of Chinese patients with CES. However, this study also has limitations. (1) The lack of objective measurement tools to assess CES symptoms prevented further exploration of their impact on postoperative CES recovery; (2) incomplete preoperative psychosocial information makes it impossible to explore its potential impact on postoperative recovery; (3) due to individual cases with follow-up periods close to 6 years, the prevalence of bladder, bowel, and sexual function issues within the study population may potentially increase with advancing age; and (4) although this study carefully explained the questions asked to the patients and excluded patients who did not cooperate well during telephone follow-up, the information obtained via telephone follow-ups may have minor biases.

Conclusions

This study revealed that stress urinary incontinence, overactive bladder, low stream, difficult defecation, fecal incontinence, and saddle anesthesia are MVPs of PPR after decompression surgery in CES secondary to LDDs. Furthermore, we established and validated the first pretreatment nomogram for predicting the risk of PPR, and a score of 148.02 was determined as the optimal cutoff value for the nomogram. This model can help clinicians predict poor long-term recovery in patients after decompression surgery for CES. It will, therefore, aid clinicians in providing appropriate treatment to patients. Furthermore, validation of this model for CES in Western populations is needed.

Ethics approval

This retrospective study meeting the ethical standards of the Helsinki Declarations was approved by the Ethics Committee of the Xijing Hospital (20232023-C-1).

Consent to participate

Written informed consent was obtained from all enrolled patients.

Sources of funding

This work was supported by the ‘Li Jian Action’ project promoted by Talents of Air Force Medical University (2022-fhjsyxrc11).

Author contribution

Q.S.W.: conceptualization, data curation, formal analysis, funding acquisition, investigation, software, validation, writing – original draft, and writing – review and editing; G.D.H.: data curation, formal analysis, software, visualization, writing – original draft, and writing – review and editing; M.Y.W.: data curation, formal analysis, methodology, writing – original draft, and writing – review and editing; Z.W.R.: conceptualization, data curation, and writing – review and editing; W.D.: data curation, methodology, funding acquisition, writing – review and editing; X.L.: data curation, investigation, writing – review and editing; Z.Y.: conceptualization, data curation, investigation, writing – review and editing; S.X.Z.: conceptualization, data curation, investigation, writing – review and editing; B.Y.: conceptualization, data curation, investigation, writing – review and editing; Z.P.T.: conceptualization, data curation, investigation, writing – review and editing; P.P.H.: conceptualization, data curation, investigation, writing – review and editing; F.X.: conceptualization, data curation, investigation, writing – review and editing; B.G.: methodology, funding acquisition, supervision, validation, writing – review and editing, and project administration; X.Y.H.: conceptualization, supervision, resources, writing– review and editing, and project administration; Z.J.L.: conceptualization, funding acquisition, validation, supervision, resources, writing – review and editing, and project administration.

Conflicts of interest disclosure

The authors declare that they have no conflicts of interest.

Research registration unique identifying number (UIN)

1. Name of the registry: Research Registry.

2. Unique identifying number or registration ID: researchregistry9743.

3. Hyperlink to your specific registration (must be publicly accessible and will be checked): https://researchregistry.knack.com/research-registry#home/registrationdetails/656867702e62a900293b390b/.

Guarantor

Zhuojing Luo, MD, PhD. Department of Orthopaedic, Xijing Hospital, Air Force Medical University, Xi’an, China.

Data availability statement

The datasets used and/or analyzed for the present study are available from the corresponding author upon reasonable request.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Supplementary Material