European Spine Journal

, Volume 24, Issue 2, pp 369–380

Outcomes and their predictors in lumbar spinal stenosis: a 12-year follow-up

Authors

    • Department of NeurologyUniversity Hospital and Masaryk University Brno
    • CEITEC-Central European Institute of Technology, Masaryk University
  • S. Vohanka
    • Department of NeurologyUniversity Hospital and Masaryk University Brno
    • CEITEC-Central European Institute of Technology, Masaryk University
  • L. Dusek
    • Institute of Biostatistics and AnalysesMasaryk University
  • J. Jarkovsky
    • Institute of Biostatistics and AnalysesMasaryk University
  • R. Chaloupka
    • Orthopaedic DepartmentUniversity Hospital and Masaryk University Brno
  • J. Bednarik
    • Department of NeurologyUniversity Hospital and Masaryk University Brno
    • CEITEC-Central European Institute of Technology, Masaryk University
Original Article

DOI: 10.1007/s00586-014-3411-y

Cite this article as:
Adamova, B., Vohanka, S., Dusek, L. et al. Eur Spine J (2015) 24: 369. doi:10.1007/s00586-014-3411-y

Abstract

Purpose

The aim of this prospective observational cohort study was to evaluate long-term outcomes in patients with mild-to-moderate lumbar spinal stenosis (LSS) and to analyse the predictors of clinical outcomes.

Methods

A group of 53 patients were re-examined after a median period of 139 months. Evaluations were made of subjective clinical outcome, objective clinical outcome and its predictors, any correlation between subjective and objective outcome, and the development of changes in radiological and electrophysiological parameters after 12 years.

Results

Satisfactory objective and subjective clinical outcomes were recorded in 54.7 and 43.4 % of patients, respectively. No statistically significant correlation between objective and subjective clinical outcome was found (Spearman coefficient = 0.225, p = 0.132). Patients with isolated unsatisfactory subjective outcome exhibited the highest Functional Comorbidity Index of all subgroups. Electrophysiological and radiological findings did not demonstrate statistically significant changes after 12-year follow-up. Multivariate logistic regression confirmed only the lowest transverse diameter of spinal canal ≦13.6 mm as an independent predictor of unsatisfactory clinical outcome (OR = 5.51).

Conclusions

Satisfactory objective and subjective clinical outcomes were disclosed in about half of the patients with mild-to-moderate LSS in a 12-year follow-up. The number of comorbid diseases had an unfavourable effect on subjective evaluation of clinical outcome. The lowest transverse diameter of spinal canal proved to be the only independent predictor of deterioration of clinical status in LSS patients.

Keywords

Lumbar spinal stenosisOutcomePrognosisPredictor

Abbreviations

AP

Anteroposterior

BMI

Body mass index

CT

Computed tomography

EMG

Electromyography

FCI

Functional Comorbidity Index

LL

Lower limb

LSS

Lumbar spinal stenosis

MRI

Magnetic resonance imaging

NASS

North American Spine Society

NC

Neurogenic claudication

NIS-LSS

Neurological impairment score in lumbar spinal stenosis

NRS

Numerical rating scale

OR

Odds ratio

ODI

Oswestry Disability Index

ROC

Receiver operating characteristic

Introduction

Lumbar spinal stenosis (LSS) is defined as an osteoligamentous narrowing of the spinal or nerve root canal in the lumbar spine that is clinically symptomatic.

The choice of optimal treatment modality (i.e. conservative vs. surgical) for patients with LSS still remains a matter of debate and a number of facts have to be taken into consideration. The results of several published studies and their reviews justify conservative therapy in patients with LSS [115]. However, there exists a large body of studies and reviews in which results justify surgical therapy in LSS patients [1, 47, 1017]. The benefits of surgery have also been demonstrated in patients of higher age but the rate of complications is higher (especially general complications, dural lesions) [18, 19]. Thus, taking both views into account, the choice of optimal therapy in patients with LSS proves somewhat complicated. Only a few long-term studies document the natural course of LSS and most of them are retrospective, as well as methodologically flawed. Knowledge of the natural course of the disease and predictive factors for further clinical development is key to making the correct decision about optimal treatment procedure. Certain studies have identified a range of predictors: radiological [8, 9], clinical [6, 20] and electrophysiological [21]. One failed to find any [4].

The need for trials documenting the long-term natural course of the disease and predictors of it has been highlighted [1, 14, 15]. In the light of this situation, we organised a study to evaluate long-term outcomes in patients with mild-to-moderate LSS and to analyse a wide range of variables (demographical, clinical, imaging and electrophysiological) as potential predictors of clinical outcomes.

Materials and methods

Study design

This study is a prospective observational cohort trial of patients with mild-to-moderate LSS. Its first results (7-year follow-up) have already been published [21] and further analysis after 12 years is presented in this paper. The study was reviewed and approved by the local medical research ethics committee. Informed and signed consent was given in writing by all subjects.

Patient population

The study group was recruited from a total of 151 patients with clinically symptomatic LSS diagnosed, treated and observed in the Department of Neurology of the University Hospital, between 1998 and 2002. The patients were considered for inclusion in the study if they fulfilled the following criteria:
  • Clinically symptomatic LSS (NC and/or radicular pain, e.g. low back pain radiating below the knee to one or both limbs).

  • Mild-to-moderate LSS suitable for conservative therapy at entry examination.

  • Presence of central degenerative LSS (an osteoligamentous narrowing of the lumbar spinal canal) on at least one level, established by CT. The following conditions did not prevent inclusion: degenerative spondylolisthesis, radiographic instability of the lumbar spine and degenerative lumbar scoliosis.

  • Correlation between clinical and radiological findings.

  • Age under 75 years at entry examination.

The assessment of LSS severity is based on the severity of clinical impairment and its impact on patient disability. We evaluated pain by means of a numerical rating scale (0–10), counting mild pain: 0–3, moderate pain: 4–6. Disability was assessed by means of the ODI with mild disability: 0–40 % and moderate disability: 41–60 %. Walking distance was also factored in, with mild impairment: more than 201 metres, moderate impairment: 21–200 m. The severity of LSS was then derived from order categorization of individual parameters.

The exclusion criteria were as follows:
  • Presence of hip and/or knee joint arthrosis limiting walking.

  • Isolated herniated nucleus pulposus, isolated lateral or foraminal stenosis, spondylolysis and isthmic spondylolisthesis.

  • Presence of arteriosclerotic peripheral vascular disease of the lower extremity limiting walking.

  • Presence of diabetes mellitus or other disease causing polyneuropathy (to eliminate the effects of polyneuropathy on electrophysiological and clinical neurological examination).

  • Presence of any other serious comorbidity (e.g. malignancy, stroke, etc.).

Study treatment

Most of the patients were given multimodal conservative treatment (analgesics, back exercises, physical therapy and the use of a lumbosacral corset), especially at the beginning of the study. Epidural steroid injections were administered to only four patients during the 12-year follow-up. The conservative therapy was changed and adjusted with respect to the specific patient’s condition and complaints during follow-up, and most of these patients were observed regularly at intervals of between 3 and 6 months. Surgery consisted of a posterior decompressive laminectomy with or without fusion (with or without instrumentation). The type of operation was based on our own clinical experience after taking into consideration the patient’s age, general health status, activity level and also radiological findings (the presence of spondylolisthesis, instability and scoliosis). Operation was suggested to a patient if deterioration of clinical status or severe pain during follow-up had been observed.

Clinical examination, history and scores

All patients were given a detailed clinical neurological examination, including full medical history at initial and final visit. Symptoms were confirmed by an experienced neurologist specialising in LSS.

We investigated the presence of NC and resting pain (low back pain with/without leg pain), BMI, average daily pain score for lower back and/or lower limb measured on an 11-point numerical rating scale (0–10), presence and severity of lower limb paresis, positivity of straight leg raising test (Lasègue’s sign), deep tendon reflexes, tactile and vibratory sensation in the LL, walking test (evaluation of a 10-m walk, including time taken to walk the distance as quickly as possible without help), and running test (evaluation of a 10-m run, including time taken to do the run as quickly as possible without help).

The neurological impairment score in lumbar spinal stenosis (NIS-LSS) was evaluated. This is a scoring system for the assessment of neurological impairment in the lower limbs of patients with LSS; the total score ranges from 0 (inability to walk) to 33 points (no impairment) [22].

The Functional Comorbidity Index (FCI) was calculated at final visit. This index of comorbid diseases, with physical function as the outcome of interest, contains 18 diagnoses scored by adding the number of “yes” answers. The score ranges from 0, indicating no comorbid illness, to 18, indicating the highest number of comorbid illnesses [23]. The FCI does not take into consideration the severity of the diagnoses.

The Oswestry Disability Index was assessed at initial and final visit to evaluate the disturbance of common activities of daily living through LSS, to quantify subjective problems and report the level of disability.

Evaluation of clinical outcome

The clinical outcome was assessed using objective criteria and subjective self-evaluation (by patients).

Objective clinical outcome for the LSS patients was classified as satisfactory (stable or improved clinical status) or unsatisfactory (deterioration in clinical status).

The criteria for unsatisfactory objective clinical outcome were the same as those used in evaluation after 7 years:
  • New onset of neurogenic claudication becoming evident in the course of follow-up.

  • Occurrence or progression of lower limb paresis.

  • A minimum 15-point increase in the ODI.

  • Indication for surgical treatment based on deterioration of clinical status, severe pain and agreement of the patient to surgery.

For definition of unsatisfactory objective outcome at least one of these criteria had to be met.

Evaluation of the subjective clinical outcome was based on self-reported status with respect to LSS. Patients were asked to assess their status in comparison with that before entry into the study by using the Likert-type five-category scale: much worse, somewhat worse, about the same, somewhat better, much better. Unsatisfactory subjective outcome was defined as a worsening of patient status (much worse or somewhat worse).

Radiological examination

The LSS patients were examined radiologically according to the following protocol at initial visit:
  • A plain radiograph of the lumbar spine was taken, with assessment of the presence of spondylarthrosis, scoliosis and degenerative spondylolisthesis.

  • CT axial scans at three levels (L3–S1) were performed. The CT criteria for central stenosis have been described in detail in a previous study of ours [21]. Anteroposterior and transverse interarticular diameters were measured at three levels (L3–S1) and a morphological evaluation was also made (presence of spondylosis, facet joint arthrosis and hypertrophy of the ligamenta compromising nervous structures). The CT scans were assessed by an experienced neuroradiologist. The lowest anteroposterior and transverse diameters at L3–S1 levels and the number of stenotic levels were evaluated.

The patients were also examined radiologically at the end of study (after 12 years). An MRI of the lumbar spine was performed to determine the number of stenotic levels. The MRI scans were assessed by an experienced neuroradiologist. The presence of central stenosis on MRI of the lumbar spine was considered in morphological terms as encroachment on the thecal sac with compression of nervous structures. When MRI was contraindicated in patient (e.g. due to pacemaker or claustrophobia), CT axial scans at three levels (L3–S1) were performed, as at initial visit.

Electrophysiological examination

Electrophysiological examination of the lower extremities included motor and sensory conduction studies, F-waves, soleus H-reflex and needle EMG from the L4–S1 myotomes. The presence of radiculopathy in patients with LSS was investigated. Electrophysiological examinations were performed at initial and final visits. A detailed description of the electrophysiological examination is included in one of our previous papers [21].

Statistical approaches

Standard robust summary statistics were used to describe primary data: relative and absolute frequencies, median and 5th–95th percentile range. The ML-χ2 test was employed to compare experimental variants in categorical variables and the non-parametric Mann–Whitney U test was used for comparison of variants on the basis of continuous variables.

The diagnostic power of potential predictors of unsatisfactory outcome in LSS was assessed by means of ROC curves. These ROC analyses were based on the binormal assumption. MedCalc 11.1.0.0 (MedCalc Software 1993–2009) was used to estimate the sensitivity and specificity of potential predictors in relation to unsatisfactory outcome of LSS. Sensitivity and specificity estimates were supplied with corresponding confidence intervals based on binomial distribution.

Univariate and multivariate logistic regression was applied to assess association of the predictive factors examined and the risk endpoint (i.e. binary coded unsatisfactory outcome of LSS). Odds ratios at 95 % confidence limits were calculated and tested in Wald χ2 test. Parameters with predictive power (providing at least p < 0.10 in univariate logistic regression) were then examined for mutual correlation and interaction terms were coded and tested for significantly correlated pairs of variables. If used in the models, effective cut-off values of continuous variables were optimised on the basis of ROC analysis. The final set of potential predictive factors and interaction terms entered an objective stepwise selection algorithm in multivariate logistic regression (driven by maximum likelihood ratio test).

Correlation of objective and subjective outcome was assessed by means of Spearman correlation coefficient, while the effect of the FCI was tested by Mann–Whitney U test. Comparison of radiological and electrophysiological parameters between initial and final examination was performed, with the statistical significance of difference tested by McNemar test for categorical data and by Wilcoxon signed-rank test for continuous parameters.

Results

Patient characteristics and follow-up

Seventy-one patients fulfilled the inclusion criteria and were enrolled into long-term follow-up. Fifty-three patients underwent clinical examination at their check-up after 12 years (46 of them underwent clinical examination after 7-year follow-up); 18 patients did not take part (7 of them died, 2 changed address, 9 gave no reason for absence). Radiological examination after 12-year follow-up was performed in 50 patients (2 patients refused this examination and 1 patient was too obese for the MRI gantry) and electrophysiological examination in 46 patients (7 patients refused this examination). Figure 1 summarises subject recruitment. The period of follow-up was 139 months (median); range 99–162 months.
https://static-content.springer.com/image/art%3A10.1007%2Fs00586-014-3411-y/MediaObjects/586_2014_3411_Fig1_HTML.gif
Fig. 1

Study flowchart including subject recruitment

Clinical outcome

Objective clinical outcome

The LSS patients were divided into two subgroups based on the above-mentioned criteria for objective clinical outcome. An unsatisfactory clinical outcome (worse clinical status) was disclosed in 24 patients (45.3 %) and a satisfactory outcome in 29 patients (54.7 %) based on objective criteria. The basic characteristics of LSS patients on entry, including the two subgroups, appear in Table 1. Both subgroups (with satisfactory and unsatisfactory objective outcome) were comparable in characteristics such as duration of follow-up, sex, age and BMI.
Table 1

Initial characteristics of LSS patients

Parameters1

All patients (N = 53)

Objective clinical outcome after 12-year follow-up

Satisfactory (N = 29)

Unsatisfactory (N = 24)

p2

Follow-up (months)

139 (112; 151)

139 (101; 157)

139 (115; 147)

0.604

Patient characteristics

 Sex, men

23 (43.4 %)

13 (44.8 %)

10 (41.7 %)

0.817

 Age at entry (years)

55 (42; 71)

56 (42; 71)

55 (42; 71)

0.485

  ≤55

27 (50.9 %)

14 (48.3 %)

13 (54.2 %)

0.669

  >55

26 (49.1 %)

15 (51.7 %)

11 (45.8 %)

 BMI

28.7 (22.6; 40.9)

29.4 (23.8; 37.7)

28.1 (21.6; 42.5)

0.574

 Obesity (BMI > 30)

18 (34.0 %)

12 (41.4 %)

6 (25.0 %)

0.210

Clinical characteristics

 Neurogenic claudication at entry

28 (52.8 %)

16 (55.2 %)

12 (50.0 %)

0.707

 Resting low back pain at entry

38 (71.7 %)

22 (75.9 %)

16 (66.7 %)

0.459

 Lower limb paresis at entry

18 (34.0 %)

10 (34.5 %)

8 (33.3 %)

0.930

 Positive Lasègue’s sign

17 (32.1 %)

10 (34.5 %)

7 (29.2 %)

0.680

 Abnormal tactile sensation in LL

27 (50.9 %)

15 (51.7 %)

12 (50.0 %)

0.901

 NIS-LSS

29 (17; 33)

29 (15; 33)

28 (19; 33)

0.661

 ODI at entry

40.0 (6.0; 68.0)

42.0 (6.0; 64.0)

38.0 (12.0; 68.0)

0.720

 Pain at entry (NRS)

4.0 (1.0; 7.0)

4.0 (1.0; 7.0)

4.5 (1.0; 8.0)

0.392

 Ability to perform walking test

43 (81.1 %)

23 (79.3 %)

20 (83.3 %)

0.196

 Walking test(s)

9.2 (6.8; 17.8)

8.0 (6.8; 16.0)

9.5 (6.0; 22.4)

0.092

 Ability to perform running test

31 (58.5 %)

17 (58.6 %)

14 (58.3 %)

0.885

 Running test(s)

4.8 (3.0; 8.5)

4.6 (3.0; 8.5)

5.0 (2.4; 9.4)

0.404

 Lumbar spine surgery before entry

7 (13.2 %)

3 (10.3 %)

4 (16.7 %)

0.499

Radiology

 Lowest AP diameter

11.0 (7.4; 13.7)

11.1 (8.6; 13.8)

10.8 (7.1; 13.6)

0.362

 Lowest transverse diameter

13.4 (6.6; 18.7)

14.6 (6.6; 19.6)

12.2 (6.6; 16.6)

0.020

 Number of stenotic levels

 1

12 (22.6 %)

7 (24.1 %)

5 (20.8 %)

0.241

 2

25 (47.2 %)

16 (55.2 %)

9 (37.5 %)

 3

16 (30.2 %)

6 (20.7 %)

10 (41.7 %)

 Spondylarthrosis

46 (86.8 %)

25 (86.2 %)

21 (87.5 %)

0.890

 Degenerative spondylolisthesis

15 (28.3 %)

7 (24.1 %)

8 (33.3 %)

0.459

Electrophysiology

 H-reflex—normal

28 (52.8 %)

18 (62.1 %)

10 (41.7 %)

0.088

 H-reflex—unilateral abnormality

15 (28.3 %)

8 (27.6 %)

7 (29.2 %)

 H-reflex—bilateral abnormality

9 (17.0 %)

2 (6.9 %)

7 (29.2 %)

 H-reflex—occurrence

45 (84.9 %)

26 (89.7 %)

19 (79.2 %)

0.288

 H-reflex amplitude—mean

2.1 (0.4; 5.2)

2.1 (0.6; 5.1)

2.1 (0.2; 7.0)

0.323

 H-reflex latency—mean

31.4 (28.0; 38.0)

31.4 (28.5; 34.7)

32.4 (27.5; 38.9)

0.621

 Radiculopathy

38 (71.7 %)

20 (69.0 %)

18 (75.0 %)

0.627

  Monoradicular involvement

15 (28.3 %)

9 (31.0 %)

6 (25.0 %)

0.463

  Pluriradicular involvement

23 (43.4 %)

11 (37.9 %)

12 (50.0 %)

1Categorical data are described by absolute number and percentage of patients in given category; continuous variables are described by median and 5th–95th percentile

2Statistical significance of difference between subgroups is tested by ML-χ2 test for categorical data and by Mann–Whitney U test for continuous parameters. Results significant at α = 0.05 level appear in bold

During the 12-year follow-up, seven patients (13.2 %) underwent surgery (decompressive operation with addition of fusion in indicated cases, with or without instrumentation) because of severe symptoms (failure of conservative treatment). The median time period from entry to surgery was 34 months (range 5–97 months). Three patients underwent a second operation, again for deterioration of symptoms, with intervals after the first operation of 14, 36 and 65 months.

One patient was given an implant for spinal cord stimulation to reduce the pain, to good effect.

Subjective clinical outcome

Twenty-three patients of the conservative treatment group (46 patients) evaluated their outcome as satisfactory and 23 patients as unsatisfactory (much worse or somewhat worse). Three patients of the seven operated upon evaluated their outcome as satisfactory and four of them as unsatisfactory. In summary, satisfactory subjective clinical outcome was disclosed in 43.4 % of the patients with mild-to-moderate LSS.

Correlation of objective and subjective clinical outcomes

A correlation of objective and subjective clinical outcomes was performed only for patients treated conservatively, as surgical treatment was one of the criteria for unsatisfactory objective clinical outcome. The correlation between objective and subjective outcome was not significant (Spearman correlation coefficient 0.225, p = 0.132) (Table 2). An agreement between the two outcomes was found in 60.9 % of patients. Unsatisfactory subjective outcome was associated with satisfactory objective outcome in 26.1 %, and the opposite in 13.0 % of patients (Table 3). The patients with unsatisfactory subjective clinical outcome despite satisfactory objective clinical outcome revealed the highest FCI (median FCI = 5) of all subgroups and the difference in FCI between this subgroup and the whole of the remaining sample was statistically significant (p = 0.041).
Table 2

Objective and subjective clinical outcome after 12-year follow-up (only patients without surgery during follow-up: N = 46)

 

Subjective outcome

Total

Satisfactory

Unsatisfactory

Objective outcome

Satisfactory

17 (37.0 %)

12 (26.1 %)

29 (63.0 %)

Unsatisfactory

6 (13.0 %)

11 (23.9 %)

17 (37.0 %)

Total

23 (50.0 %)

23 (50.0 %)

46 (100.0 %)

Spearman correlation coefficient between objective and subjective outcome is 0.225 (p = 0.132)

Table 3

Relationship between number of comorbidities and subjective or objective perception of clinical outcome after 12-year follow-up (only patients without surgery during follow-up: N = 46)

 

N (%)

FCI1

p2

Objective and subjective satisfactory outcome

17 (37.0)

4 (2; 6)

0.660

Only subjective unsatisfactory outcome

12 (26.1)

5 (2; 9)

0.041

Only objective unsatisfactory outcome

6 (13.0)

3 (1; 6)

0.561

Objective and subjective unsatisfactory outcome

11 (23.9)

3 (1; 6)

0.254

1Functional Comorbidity Index—number of comorbid illnesses

2Statistical significance of difference in FCI between given subgroup and the whole remaining sample is tested by Mann–Whitney U test. Results significant at α = 0.05 level appear in bold

Comparison of objective clinical outcomes after 7- and 12-year follow-ups

A comparison of outcomes for 46 patients appears in Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs00586-014-3411-y/MediaObjects/586_2014_3411_Fig2_HTML.gif
Fig. 2

Comparison of objective clinical outcome after 7-year follow-up and 12-year follow up with respect to number/change of fulfilled criteria for unsatisfactory outcome (analysis included only patients who underwent both 7 and 12-year examinations: N = 46)

Satisfactory objective clinical outcome was disclosed in 27 patients (58.7 %) after 7-year follow-up and in 25 (54.3 %) patients after 12-year follow-up. Thirty-eight patients (82.6 %) exhibited the same trend of development (satisfactory or unsatisfactory) after 7 and 12 years, but five patients with satisfactory outcome after 7 years showed unsatisfactory outcome after 12 years and the opposite trend was found in three patients with unsatisfactory outcome who showed satisfactory outcome after 12 years.

Unsatisfactory objective clinical outcome is defined by the four parameters mentioned above; changes in these factors between 7- and 12-year follow-up appear in Fig. 2. The majority of patients (71.7 %) exhibited no change in these factors so their clinical course was considered stable between the seventh and twelfth year of the follow-up.

Radiological and electrophysiological findings during follow-up

The number of stenotic levels was assessed at entry and after 12 years. The median was two stenotic levels at initial and final examination; no statistically significant change was shown for conservatively treated patients (Table 4).
Table 4

Comparison of radiological and electrophysiological parameters after 12-year follow-up (only patients without surgery)

Parameters1

At entry

After 12-year follow-up

p2

Radiology

 Number of stenotic levels3

2 (1; 3)

2 (0; 3)

0.459

Electrophysiology4

 Radiculopathy

27 (69.2 %)

31 (79.5 %)

0.289

 Monoradicular involvement

13 (33.3 %)

10 (25.6 %)

0.508

 Pluriradicular involvement

14 (35.9 %)

20 (51.3 %)

0.109

 Number of affected roots

1 (0; 6)

2 (0; 4)

0.399

1Categorical data are described by absolute number and percentage of patients in given category; continuous variables are described by median and 5th–95th percentile

2Statistical significance of difference between entry examination and control examination is tested by McNemar test for categorical data and by Wilcoxon signed-rank test for continuous parameters

3Only patients without surgery during follow-up and with known information of number of stenotic level at both examinations (N = 43)

4Only patients without surgery undergoing EMG at both examinations (N = 39)

The comparison of electrophysiological findings in unoperated patients after 12 years established no statistically significant changes with reference to the presence of radiculopathy, monoradicular involvement, pluriradicular involvement or the number of affected roots (Table 4).

Predictors of clinical objective outcome

Demographical and clinical predictors

Differences in initial demographical and clinical parameters between the two subgroups (with satisfactory and unsatisfactory objective clinical outcome) were evaluated and no statistically significant difference in any of the potential clinical predictors emerged (Table 1).

Radiological predictors

The initial radiological parameters were evaluated in the two subgroups. Patients with unsatisfactory objective clinical outcome showed a significantly lower median value of the lowest transverse diameter (12.2 vs. 14.6 mm, p = 0.02), while any differences in the lowest AP parameter, number of stenotic levels, the presence of spondylarthrosis and degenerative spondylolisthesis were not significant.

ROC analysis of potential predictors of unsatisfactory objective clinical outcome in patients with LSS disclosed the lowest transverse diameter as the only effective discriminating factor between the two subgroups of patients, at a sensitivity of 75.0 % and specificity of 65.5 % (p = 0.02) when using an optimal cut-off point of 13.6 mm (Table 5; Fig. 3). Normal value for transverse diameter is at least 16 mm (based on our own normal data). We also calculated the odds ratio (OR) of this potential predictor for unsatisfactory objective clinical outcome in LSS patients. Univariate analysis led to statistically significant OR (crude OR) for unsatisfactory outcome at the lowest transverse diameter 13.6 mm or less (OR = 4.91), and multivariate logistic regression proposed this parameter as an independent predictor of unsatisfactory objective clinical outcome (OR = 5.51) (Table 6).
Table 5

ROC analysis of potential predictors for unsatisfactory objective clinical outcome of LSS

Variable

AUC (95 % CI)1

p1

Optimal cut-off2

Sensitivity3 (%)

Specificity3 (%)

Age at entry

0.56 (0.40; 0.71)

0.486

≤47

29.2

86.2

BMI

0.55 (0.38; 0.71)

0.574

≤29.4

70.8

51.7

Clinical characteristics

 NIS-LSS

0.54 (0.38; 0.69)

0.662

≤26.8

45.8

65.5

 ODI at entry

0.53 (0.37; 0.69)

0.721

≤41

66.7

51.7

 Pain at entry (NRS)

0.57 (0.41; 0.72)

0.401

≥4

66.7

48.3

 Walking test(s)

0.65 (0.48; 0.82)

0.093

≥7.8

85.0

43.5

 Running test(s)

0.59 (0.38; 0.79)

0.405

≥4.7

64.3

52.9

Radiology

 Lowest AP diameter

0.57 (0.42; 0.73)

0.362

≤11.3

75.0

48.3

 Lowest transverse diameter

0.69 (0.54; 0.83)

0.020

≤13.6

75.0

65.5

Electrophysiology

 H-reflex amplitude—mean

0.59 (0.41; 0.76)

0.334

≤0.7

36.8

92.3

 H-reflex latency—mean

0.54 (0.36; 0.73)

0.621

≥33.9

36.8

92.3

1Area under the ROC curve with corresponding 95 % confidence interval and statistical significance. Results significant at α = 0.05 level appear in bold

2Optimal cut-off points for binary coding of parameters

3Sensitivity and specificity for given cut-off point

https://static-content.springer.com/image/art%3A10.1007%2Fs00586-014-3411-y/MediaObjects/586_2014_3411_Fig3_HTML.gif
Fig. 3

The lowest transverse diameter as a predictor for unsatisfactory objective clinical outcome after 12-year follow-up

Table 6

Potential predictors for unsatisfactory objective clinical outcome after 12-year follow-up of LSS in logistic regression models

Parameters (categories)

%1

Crude OR2

p2

Multivariate-adjusted OR2

p3

Total

45.3

Patient characteristics

 Sex, men

43.5

1.14 (0.38; 3.39)

0.817

  

 Age at entry (years)

0.98 (0.93; 1.04)

0.545

  

 Age >55

42.3

0.79 (0.27; 2.34)

0.670

  

 BMI

1.01 (0.92; 1.11)

0.804

  

 Obesity (BMI >30)

33.3

2.12 (0.65; 6.91)

0.214

  

Clinical characteristics

 Neurogenic claudication at entry

42.9

0.81 (0.27; 2.40)

0.707

  

 Resting low back pain at entry

42.1

0.64 (0.19; 2.12)

0.461

  

 Lower limb paresis at entry

44.4

0.95 (0.30; 2.98)

0.930

  

 Positive Lasègue’s sign

41.2

0.78 (0.24; 2.51)

0.680

  

 Abnormal tactile sensation in LL

44.4

0.93 (0.32; 2.75)

0.901

  

 NIS-LSS

0.99 (0.90; 1.11)

0.929

  

 ODI at entry

0.99 (0.96; 1.03)

0.753

  

 Pain at entry (NRS)

1.15 (0.87; 1.52)

0.343

  

 Ability to perform walking test

46.5

1.30 (0.32; 5.29)

0.710

  

 Walking test(s)

1.15 (0.96; 1.38)

0.129

  

 Ability to perform running test

45.2

1.10 (0.31; 3.92)

0.886

  

 Running test(s)

1.21 (0.75; 1.95)

0.440

  

 Lumbar spine surgery before entry

57.1

1.73 (0.35; 8.64)

0.502

  

Radiology

 Lowest AP diameter

0.87 (0.61; 1.23)

0.427

  

 Lowest transverse diameter

0.83 (0.71; 0.98)

0.026

  

 Lowest transverse diameter ≤13.6

62.1

4.91 (1.49; 16.14)

0.009

5.51 (1.47; 20.61)

0.011

 Number of stenotic levels

 1

41.7

Basal category

  

 2

36.0

0.79 (0.19; 3.22)

0.740

  

 3

62.5

2.33 (0.51; 10.78)

0.278

  

 Spondylarthrosis

45.7

1.12 (0.22; 5.58)

0.890

  

 Degenerative spondylolisthesis

53.3

1.57 (0.47; 5.23)

0.461

  

Electrophysiology

 H-reflex—normal

35.7

Basal category

  

 H-reflex—unilateral abnormality

46.7

1.58 (0.44; 5.64)

0.485

  

 H-reflex—bilateral abnormality

77.8

6.30 (1.09; 36.30)

0.039

5.13 (0.79; 33.50)

0.087

 H-reflex—occurrence

42.2

0.44 (0.09; 2.06)

0.297

  

 H-reflex amplitude—mean

0.92 (0.63; 1.33)

0.651

  

 H-reflex latency—mean

1.09 (0.87; 1.36)

0.461

  

Radiculopathy

47.4

1.35 (0.40; 4.54)

0.628

  

 Monoradicular involvement

40.0

0.74 (0.22; 2.49)

0.628

  

 Pluriradicular involvement

52.2

1.64 (0.55; 4.90)

0.379

  

1Percentage of worsened patients within category given by the predictor (only for categorised predictors)

2Odds ratio with 95 % confidence interval (in parentheses); statistical significance based on Wald’s test

3Only statistically significant multivariate-adjusted ORs are displayed (variables selected by stepwise selection). Results significant at α = 0.05 level appear in bold

Electrophysiological predictors

The presence of radicular involvement found by EMG was established in 71.7 % of patients (69 % with satisfactory outcome and 75 % with unsatisfactory outcome) at entry examination. Pluriradicular involvement was more frequent than monoradicular lesion in patients with unsatisfactory objective clinical outcome (50.0 vs. 25.0 %) compared to the subgroup with satisfactory outcome (37.9 vs. 31.0 %) but this was not statistically significant. We found no statistically significant difference in the soleus H-reflex parameters (amplitude, latency) between the two subgroups, but bilaterally abnormal soleus H-reflex was more frequent in patients with unsatisfactory outcome (29.2 vs. 6.9 %). The calculation of OR for unsatisfactory outcome revealed a significant figure (OR = 6.3) for bilaterally abnormal soleus H- reflex in univariate analysis, but the multivariate logistic regression did not confirm this parameter as an independent predictor of unsatisfactory objective clinical outcome (OR = 5.13, p = 0.087).

Discussion

This prospective study describes long-term outcomes (12-year follow-up) for patients with mild-to-moderate LSS, primarily treated conservatively, and assesses a wide range of potential predictors of clinical outcomes in these patients. To the best of our knowledge, this is the first study that evaluates LSS patients so comprehensively and over such a long period.

Unsatisfactory clinical outcome (worse status) based on objective criteria was found in 45.3 % of patients, and unsatisfactory outcome based on subjective patient evaluation was established in 56.6 % of patients, a slightly higher proportion compared to the 11–33 % reported in other studies [3, 4, 8, 9]. Satisfaction with the course of disease and treatment is naturally very difficult to evaluate and compare. The results are dependent on several different factors, among them the definitions of end-points, satisfactory or unsatisfactory outcomes, duration of follow-up, severity of LSS and type of treatment. The assessment of objective clinical outcome for this study is based on a combination of four strictly defined criteria, selected to minimise the influence of current treatment and age of patient. In other studies, the most frequently referenced end-points are intensity of pain, the ODI, physical function scores, patient’s global assessment, patient’s self-reported improvement, walking ability and neurological deficit [47, 1013].

Surprisingly, we found no significant correlation between subjective and objective clinical assessment of outcome. One of the reasons for this might be the low reliability of patient’s subjective assessment arising out of the long period between the time points compared (patients probably unable to recall clearly how they felt at the time of the initial visit). Comorbidities might also influence subjective assessment of current status. We found that patients with isolated subjective unsatisfactory clinical outcome had the highest FCI. This finding is in agreement with other studies that have demonstrated the unfavourable effects on surgical and nonoperative outcomes of an increasing comorbidity burden, including depression and other psychiatric disorders [2427].

In this study, in the course of a 12-year follow-up seven patients (13.2 %) required surgery due to failure of conservative treatment, fewer compared to previous studies, in which 36–39 % of medically treated patients underwent an operation during a 10-year period [4, 7]. The lower number of conservative treatment failures in our study in comparison with studies with mild-to-moderate LSS may be explained by higher patient motivation towards conservative treatment because the examining physician was a neurologist rather than a surgeon. Comparison of objective clinical outcomes after 7- and 12-year follow-ups implies stability of clinical course, which agrees with previous study results showing no clinically significant deterioration of symptoms during long-term follow-up of nonsurgically treated patients [4, 6].

The search for outcome predictors is very useful to the indication of the probable course of LSS and to help find the optimal therapy. A number of studies have dealt with predictors for postoperative clinical outcome [2529], but predictors for nonsurgical outcome or natural course have been explored in only a few. In this study, radiological severity of LSS (expressed as the lowest transverse diameter of the spinal canal) was found to be the only independent predictor of deterioration in clinical status in mild-to-moderate LSS treated conservatively. The radiological severity of LSS was established as an independent predictor of unsatisfactory clinical outcome only for 12-year follow-up, not for the 7-year follow up [21]. The lowest transverse diameter of spinal canal reflects the severity of the acquired (secondary) stenosis. Some studies confirm the predictive value of severity of LSS [8], but some do not [4, 20]. There was no statistically significant association of number of levels and outcome, which is in agreement with other studies [8, 9].

In our previous study, multivariate logistic regression proposed two electrophysiological variables as mutually independent predictors of unsatisfactory outcome: EMG signs of pluriradicular involvement and averaged soleus H-reflex amplitude <2.8 mV; bilaterally abnormal soleus H-reflex was confirmed as a predictor of unsatisfactory outcome in univariate analysis only, as in the present study [21]. Our two studies have demonstrated that certain electrophysiological parameters have some predictive value, namely abnormalities of the soleus H-reflex and the presence of pluriradicular involvement, and we recommend use of needle EMG and conduction studies as standard examination in patients with LSS. The only available study that deals with electrophysiological predictors in LSS patients found no correlation between severity of EMG findings and future pain or disability [20]. The Haig study and the present study, however, differ in many ways, especially in follow-up period (1.5 years vs. 7 and 12 years) and in definition of outcomes.

Like Amundsen et al., [4] our study found no clinical predictors of unsatisfactory outcome, while the Maine Lumbar Spine Study showed that several clinical variables are predictors of clinical outcome for both surgically and nonsurgically treated patients [6].

The discrepancy in predictors between different studies may be attributed to many factors. The studies differ in the severity of stenosis, the type of conservative treatment, duration of follow-up, parameters evaluated and especially in definition of outcomes. Furthermore, the predictive value may be outcome-specific and objective outcome measures (such as neurological/clinical examination, treadmill test) are used quite rarely in published studies. Thus, there is the possibility that a depressed patient enjoys the same “objective” benefit of treatment but nonetheless reports a worse outcome because of depression [25]. This discrepancy between objective and subjective outcomes was also evident in our study, and so the evaluation of both subjective and objective outcomes appears helpful.

The present study has some limitations. The number of patients recruited is relatively low, a total of 151 patients with clinically symptomatic LSS. The strict inclusion criteria necessary to eliminate factors that might influence outcome (e.g. serious comorbidity, diabetes mellitus, high age) were fulfilled by only 71 patients and only 53 patients completed the follow-up. The drop-out (18 patients) is appropriate to the duration of follow-up. Mode of radiological examination also proved limiting. We used CT scans at three levels at the initial visit, as CT was the standard evaluation for LSS at that time, but MRI of the lumbar spine was used for the check-up radiological examination after 12 years in most of our LSS patients. MRI is recommended today; it is more accurate and carries no radiation load. We decided to compare only the number of stenotic levels established by CT and MRI without comparison of spinal canal diameters. Still, there exists a risk of missing stenosis at higher levels of the lumbar spine when CT is used. A further possible source of inaccuracy in radiological data in this study is the morphological evaluation of MRI scans (and partially CT scans), which remains somewhat subjective. The effectiveness of CT and especially MRI could be improved by using the cross-sectional area of the dural sac. The evaluation of predictors also has its limitations, particularly the choice of criteria for objective clinical outcome. The ODI as a quantitative measure of disability is based on the subjective judgement of patients, but this score was used as one of criteria for objective clinical outcome in this study. Some potential predictors, such as duration of symptoms, walking ability (measured on a treadmill), smoking, income, education, marital status and depression at entry, were not analysed. This study, evaluating outcome predictors in LSS, concentrates only upon outcomes of conservative treatment. It is useful to counsel patients about their likely conservative outcomes, but these factors cannot provide an answer to the question of what kind of therapy is better for patients. The natural history of LSS will probably be close to the results of our study with conservative treatment, but it should be taken into consideration that any treatment may affect this natural history.

However, the advantage of this study is a prospective, long-term, comprehensive (clinical, radiological, electrophysiological) follow-up.

Conclusions

  • For patients with mild-to-moderate LSS, a satisfactory objective clinical outcome (unchanged or better clinical status) was disclosed in about 55 % and satisfactory subjective clinical outcome in 43 % in a 12-year follow-up.

  • The number of comorbid diseases, including psychiatric disorders, had an unfavourable effect on the subjective evaluation of clinical outcome.

  • Electrophysiological and radiological findings did not demonstrate significant changes after 12-year follow-up. The objective clinical course was largely stable between the seventh and twelfth year of follow-up. These findings support an assumption that LSS is a slowly progressive chronic disorder.

  • Radiological severity of LSS (lowest transverse diameter of spinal canal) was predictive of deterioration of clinical status in these patients.

Conflict of interest

None.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014