Background

Lumbar spinal stenosis is a well recognised and severely debilitating spinal condition. It is generally attributed to a narrowing of the spinal canal, nerve root canals, or intervertebral foramina, usually as a consequence of age-related degenerative changes to the spine anatomy, including bone, ligaments, facet joints, and intervertebral disc. Clinical symptoms are believed to result from compression and/or ischaemia of vascular and neurological tissues in the spine. Neurogenic claudication is the most typical symptom, and may be accompanied by lower limb pain, numbness, paraesthesia, or weakness, usually exacerbated by standing or walking [1, 2]. The available literature on the subject highlights that symptoms are exacerbated by extended positions or walking and relieved by flexed positions or sitting on patients with radiologically confirmed lumbar spinal stenosis [2]. Patients with lumbar spinal stenosis have greater mobility limitation than patients with knee or hip osteoarthritis, and this results in important reductions in both functional ability and quality-of-life [3].

Lumbar spinal stenosis has become a commonly diagnosed and treated condition of the spine—it is estimated that approximately one-fifth of adults aged 65 years or older will have symptoms of neurogenic claudication due to severe lumbar spinal stenosis [4, 5]; this condition becoming the most common reason for individuals older than 65 years undergoing spinal surgery [1, 6]. In fact, surgical management has become the standard procedure in the management of symptomatic lumbar spinal stenosis, and as a result, this procedure is currently the fastest growing surgical procedure worldwide [6, 7]. Recent report yielded various degrees of clinical symptoms in relation to radiological findings [5], which highlights the importance of correlation of both clinical and radiological findings when targeting the surgical procedures. A recent systematic review of randomised clinical trials of surgery for lumbar spinal stenosis has shown that although patients will experience decreases in pain and disability following surgery, over a quarter will have further spinal surgery 1 year after having spinal surgery, suggesting they may not experience full recovery [8]. Moreover, recent studies have shown that there is no correlation between the severity of clinical symptoms and dural cross-sectional areas [9, 10] or between increased canal diameter following surgery and symptom improvement (i.e., back or leg pain, functional status, and neurological claudication) [11]. It is, therefore, also unclear whether the initial benefits of surgery in terms of pain and disability are sustained over the years, or whether patients will experience increase in symptoms over time. However, randomised clinical trials are not the ideal design to infer long-term course, as they usually provide shorter follow-up data and more stringent inclusion criteria, if compared to cohort studies.

We, therefore, aimed to systematically review the literature to identify cohort studies assessing the long-term course of pain and disability in patients with lumbar spinal stenosis who have undergone surgery. To our knowledge, this is the first systematic review to assess the clinical course of lumbar spinal stenosis managed surgically.

Methods

Data sources and searches

This review was prospectively registered on PROSPERO (registration number CRD42013005988). MEDLINE, CINAHL, and Embase databases were searched from inception to July 2014, to identify eligible studies. Search terms are available in Additional File 1. In addition to the electronic searches, citation tracking was conducted and the reference lists of the included studies in relevant systematic reviews were checked.

Study selection

No language or geographic restrictions were included in the search strategy, but non-English studies were included in the review only when translation was available. Independent reviewers screened titles and abstracts (CF, MF, CGM, and PHF) for inclusion. The full text of potentially eligible studies was then obtained and assessed by three independent reviewers (CF, MF, and RP) for inclusion against our criteria. Disagreements were resolved by consensus.

To be eligible for inclusion in the review studies needed to explicitly report that participants had a primary diagnosis of lumbar spinal stenosis of any duration. Diagnosis had to be defined either by imaging techniques or clinically by the presence of signs and symptoms.

All prospective surgical cohort studies with at least 3-month follow-up that reported pain or disability outcomes were included. Studies that only reported recovery rates or percentage change in pain or disability were excluded.

Data extraction

For each study, summary data were obtained on sample source, sample size, patient characteristics, outcomes (pain and disability), duration of follow-up, and inception time, if applicable. Measures of central tendency (e.g., mean or median) and dispersion (e.g., standard deviation, standard error or 95 % confidence intervals) were extracted for pain and disability outcomes. Outcome data were re-scaled to a common 0–100 scale to facilitate between-study comparisons (e.g., means and standard deviations of pain scores given on a 0–10-point scale were multiplied by 10; means and standard deviations of disability scores given on a 24-point scale were multiplied by 4.1666 or 100/24). When insufficient data were reported on outcome measures, authors were contacted by e-mail for further details. If data on dispersion were not provided by the authors, standard deviations were imputed from similar studies.

Quality assessment

Methodological quality of included studies was assessed using an adaptation of the methodological criteria suggested by Altman [12] and include two items on sampling, two on completeness of follow-up, and one item on description of prognostic outcomes. These criteria have been used in a previous systematic review on prognosis of low back pain [13]. The results of the methodological quality assessment for each study are presented as percentage (Table 1).

Table 1 Characteristics of included studies
Full size table

Data synthesis

To accommodate the different time points for outcome assessment in the included studies, pain and disability were modelled as a continuous function of time. For all analyses, time was treated as time from surgery. If studies reported more than one measure of pain intensity (e.g., back and leg pain), the more severe measure at baseline was included in the analyses. In addition, secondary analyses were performed for back pain and leg pain separately.

Pooled estimates of outcomes were obtained using generalised estimating equations to account for the dependence of repeated observations (follow-ups) within studies. The observations from each study were assigned a weight equal to the inverse square of the mean SE of all observations from that study. The effect of time was clearly non-linear, so it was modelled using fractional polynomial regression [14]. The regression models were used to generate pooled point and interval estimates of outcomes at baseline and 3, 6, 12, 24, and 60 months.

Results

From a total of 10,741 titles, 69 publications reporting on 64 cohort studies were included in the review (references [15–17] are multiple publications reporting follow-up assessments of the same cohort, as are [18, 19] and [20, 21]). These studies provided data on 3774 participants (Fig. 1). Table 1 presents the main characteristics of all included studies (complete references of included studies can be found in Additional File 2). In more than 50 % of the included studies, participants had persistent symptoms of lumbar spinal stenosis (n = 38; 57 %). For the remaining 31 studies, symptom duration was not reported. In approximately one-fifth of the included studies, the recruited participants had central canal stenosis (n = 12 studies), whereas most studies (n = 55 studies; 80 %) included a mixed population (i.e., central and lateral stenosis); and 3 % (two studies) included only participants with a diagnosis of lateral canal stenosis. In most studies, the main complaint of included participants was intermittent neurological claudication with or without pain (n = 42 studies; 61 %), followed by low back pain with or without leg pain (n = 11 studies; 16 %) and radicular pain only (n = 3 studies; 4 %). In 13 studies (19 %), main complaints were not reported or unclear.

Fig. 1
figure 1

Flow chart showing process of selection of studies

Full size image

Methodological quality

All included studies presented follow-up data for at least one outcome measure at 3 months or later (n = 69), but only approximately half (n = 40; 58 %) reported enough data on clinical prognosis of lumbar spine stenosis to be included in the meta-analyses. In over three quarters of the studies, the follow-up included at least 80 % of the sample (n = 55 studies or 81 %), however, only one-third of included studies (n = 25 studies or 37 %) clearly included a representative sample of participants with spinal stenosis and in only 61 % of the studies was the sample well defined (i.e., inclusion and exclusion criteria provided).

Clinical course of pain and disability

Of the 64 included studies (69 publications), 31 provided sufficient data on disability and 39 provided sufficient data on pain to be included in the meta-analysis. Follow-up time ranged from 3 to 72 months post-surgery. In all except one study [22], baseline assessments were performed pre-operatively at the time of hospital admission. Surgical admission was, therefore, regarded as the inception time in the mixed-model analyses. The study by Bednar [22] which did not report pain or disability at the time of surgical admission was excluded from the analyses. Ha et al. [23] did not report enough data to be included in the analyses and was also excluded.

Pain and disability outcomes are presented in Fig. 2a and b, respectively. At inception (i.e., pre-operatively), the mean weighted pain score across all cohort studies was 63.4 (95 % CI 56.5–70.3). At 3-month post-surgery, pain had decreased to a weighted mean of 33.1 (95 % CI 24.2–41.9). Little further reduction in pain was seen at 6 months (mean 28.2; 95 % CI 19.1–37.4), 12 months (mean 24.5; 95 % CI 15.0–34.0), 24 months (mean 21.8; 95 % CI 12.0–31.5), or 60 months (mean 19.2; 95 % CI 9.2–29.3). The mean weighted disability scores at baseline were 36.9 (95 % CI 32.6–41.3), decreasing to 16.3 (95 % CI 11.8–20.9) at 3 months, 14 (95 % CI 9.3–18.6) at 6 months, 12.9 (95 % CI 8.2–17.6) at 12 months, 12.6 (95 % CI 7.8–17.3) at 24 months, and 12.4 (95 % CI 7.7–17.2) at 60 months.

Fig. 2
figure 2

Pain (a) and disability (b) outcomes after spinal stenosis surgery. Each circle represents the mean pain reported in a single study at a particular time. The area of the circle is proportional to the weight given to the study. The data have been fitted with fractional polynomial regression. The shaded area circumscribes 95 % confidence interval for the regression line

Full size image

The mean standard deviation at baseline was 16.5 (range 6.0–30.3) for pain and 17.3 (range 4.1–62.0) for disability. This indicates a moderate degree of person-to-person variability in outcomes within studies.

At inception, the mean weighted leg pain score was 53 points (95 % CI 43.9–62.2), decreasing to 17.6 (95 % CI 4.6–30.6) 3 months after surgery, and further decreasing at 6 (mean 17.0; 95 % CI 4.0–30.1), 12 (mean 15.9; 95 % CI 2.0–29.8), 24 (mean 13.6; 95 % CI −4.0 to 31.3), and 60 months (mean 6.8; 95 % CI −28.7 to 42.4). For back pain, the mean weighted score pre-surgery was 35 points (95 % CI 23.5–46.4). Three months after surgery, back pain had decreased to 16.4 (95 % CI 0.0–32.8). Small increases in back pain scores were seen at 6 months (mean 16.9; 95 % CI 0.6–33.3) and at 12 (mean 18; 95 % CI 1.8–34.3), 24 (mean 20; 95 % CI 4.3–36.3), and 60 months (mean 26.9; 95 % CI 11.1–42.8).

Discussion

This systematic review included 64 cohort studies assessing post-operative outcomes in 3774 participants with lumbar spinal stenosis. The data show that most participants presented with persistent symptoms of neurological claudication, with or without back or leg pain. Prior to surgery they reported, on average, moderate levels of pain and mild disability. Typically, patients experienced substantial reductions (approximately, 50 %) in pain and disability in the first 3-month post-surgery, but little further improvement over the subsequent 5 years. On average, mild levels of pain and disability persisted at 5 years.

This is the first systematic review on the course of spinal stenosis following surgery. A quantitative approach provided precise estimates of mean pain and disability at 3, 6, 12, 24, and 60 months following surgery. The review included a large number of cohort studies of generally moderate-to-low methodological quality. The main methodological flaw was failure to recruit and clearly describe a representative sample of patients (i.e., consecutive patients presenting for care, or randomly selected patients) observed in more than half of the included studies. About a third of the studies also failed to collect or clearly describe follow-up assessments on at least 80 % of the sample. One-third of the studies (n = 23; 33 %) had to be excluded from the pooled analyses due to incomplete reporting of data. In general, sample sizes were also very small—34 studies reported data on 50 participants or less. One of the main benefits of conducting a systematic review is that it provides pooled analyses of data from many studies—data from a total of 2097 participants were included in the analysis of pain and data from 1773 participants were included in the analysis of disability. Approximately half of the studies included in the pain and disability analyses were of high methodological quality (at least 80 % of total score). This gives us some confidence in the pooled estimates.

The surgical techniques varied considerably across studies. Decompression was the most prevalent type of surgery and represented 31 % of the included studies. It was followed by microsurgical decompression, which represented approximately 20 % of the reports. Decompression associated with fusion was performed in approximately 10 % of the included studies. However, 14 % of the studies reported mixed interventions and performed decompression or decompression with fusion depending on the radiological findings and/or clinical symptoms. Likewise, most studies performed a combination of single and multiple spinal-level decompression, according to the patient’s diagnosis. It is possible that these variations introduced between-study heterogeneity in the pooled analysis of outcomes. However, the lack of sufficient data provided by individual studies has prevented subgroup analyses based on the types of surgical technique. However, a recent clinical trial comparing decompression surgery and decompression with fusion showed no superiority of the addition of fusion on pain, disability, walking ability, and quality-of-life up to 5 years after surgery [24]. In addition, a systematic review compared effectiveness in regards to pain and disability among the most common surgical procedures for lumbar spinal stenosis and also reported no significant difference [8]. Therefore, it is unlikely that subgroup analyses based on the type of surgery would have yielded significantly different results.

Lumbar spinal stenosis is a highly debilitating spinal condition and its prevalence will increase over the next decades as the population ages. The number of spinal surgical procedures for spinal stenosis has also increased steadily over the years, possibly due to a scarcity of evidence on the effectiveness of non-operative management of this condition. Past research has shown that there is lack of high-quality evidence on the effectiveness of physiotherapy and non-operative interventions for patients with lumbar spinal stenosis, preventing their inclusion in clinical guideline recommendations [25, 26]. In the US, surgery for spinal stenosis was the fastest growing type of lumbar surgical procedure between 1980 and 2000 [27, 28] and in the last decade alone, Americans have experienced a 15-fold increase in the rate of complex fusions for lumbar spinal stenosis [6]. However, these procedures are known to be associated with important complications, such as need for cardiopulmonary resuscitation or repeat intubation [6], death [6], re-operation, and re-hospitalisation [29]. The need for re-operations following surgical procedures for lumbar spinal stenosis is not rare. In fact, the literature suggests that over one quarter of patients undergoing interspinous process implant will have a re-operation, including a revision of revision of the index procedure or the need to address the problem at a different spinal level [8]. The probability of having a second re-operation may be even greater (hazard ratio 1.58; 95 % CI, 1.41–1.76). Age and presence of comorbidities, however, seem to be associated with a lower chance of having a re-operation [29]. Re-operations and hospital re-admissions are often associated with greater risk of complications and with lower satisfaction with treatment when compared to the first surgical procedure [17, 30]. Our review provides evidence that patients can expect substantial relief of pain and disability in the first 3 months after surgery, but they can also expect long-term recovery to be incomplete. This information needs to be made available to patients when discussing the indication of surgical management for spinal stenosis.

Past research also suggests that patients with lumbar spinal stenosis who also report symptoms of depression and present cardiovascular comorbidities or those resulting in impaired mobility have poorer clinical outcomes [30, 31] and greater chances of re-operation [30]. Likewise, there is compelling evidence showing that increased body weight is associated with worse self-rated quality-of-life and function in patients with lumbar spinal stenosis [32]. The role of these putative predictors could not be further evaluated in our review, as individual patient data were not available. Future studies should explore the predictive value of these as well as other patient-level characteristics on the outcomes of patients who have surgery for spinal stenosis. Most importantly, high-quality randomised trials are needed to provide robust estimates of the size of effects of surgery compared to no treatment.

Conclusion

People with spinal stenosis experience substantial reductions in pain and disability in the first 3 months after surgery. Little further improvement is observed in the following 5 years.