This study identified patients with good preoperative function and HRLQoL, short duration of symptoms, narrow dural sac area, spondylolisthesis, and absence of consumption of analgesics to be most likely to achieve a good outcome after decompression for spinal stenosis. An example of this is that patients having leg pain less than 2 years and self estimated walking distance of > 100 m are almost 3 times more likely to be satisfied with operative outcome at 1-year follow-up compared with patients having leg pain exceeding 2 years and very poor preoperative walking distance (< 100 m).
An important finding was the negative effect of long duration of symptoms. Longstanding pain and loss of function in these elderly patients may be difficult to treat. Duration of symptoms is a potentially modifiable factor by giving better information to patients, doctors and healthcare policy makers for earlier diagnosis and intervention. Two meta-analyses exploring prognostic factors in spinal stenosis surgery did not find that duration of symptoms was a significant factor (Turner et al. 1992
, Aalto et al. 2006
), and a previous study from our institution found a trend of inferior outcome in patients with duration of symptoms exceeding 4 years (Jönsson et al. 1997
). Recent studies have shown duration of symptoms to be an important prognostic factor. Ng et al. (2007)
showed that patients with a duration of symptoms of less than 3 years to have a better prognosis than those with a longer duration of symptoms. In an analysis from the SPOR trail, Radcliff et al. (2011) found that patients treated both operatively and nonoperatively with a symptom duration of ≥ 1 year improved less than those with a shorter duration of symptoms. There have been few long-term studies on the outcome of surgery for spinal stenosis, but in a prospective study, Amundsen et al. (2000)
did not find that duration of symptoms had any influence on outcome after a 10-year follow-up. Thus, there is contradicting information in the literature about the predictive value of the duration of symptoms preoperatively.
In addition to the negative psychological effect of longstanding pain and inferior function on the patient, chronic spinal nerve compression has been shown in experimental studies to initiate demyelinization and change of the phenotype of the nerve, perhaps limiting the ability to improve neuronal function (Gupta et al. 2004
, Chao et al. 2008
). Compression of the cauda equina can also reduce blood flow to spinal nerves (Olmarker et al. 1989
It is important, however, to keep in mind the natural history of spinal stenosis; pain and function vary over time (Johnsson et al. 1992
). Thus, if surgery is performed very early, some patients will probably be operated who would also improve without surgery. In a randomized study by Malmivaara et al. (2007)
, about half of the patients who were assigned to nonoperative treatment experienced improvement during the follow-up period. It is important for the spinal surgeon to discuss these facts with the patient and the results of our study can provide some help in that discussion.
Another important finding is the negative prognostic value of poor preoperative function, as patients with very low functional ability are less satisfied at 1-year follow-up and gain less in terms of pain and HRLQoL (EQ-5D). This implies a true risk of over-treatment as some—especially older—patients can accept a low level of function and, as mentioned earlier, the natural history of spinal stenosis is quite variable. The impact of comorbidities on function should also be weighed in, as a patient with severe heart insufficiency or chronic obstructive pulmonary disease would probably experience limited gain in function by having a decompression for spinal stenosis.
We found no difference in outcome in patients with and without spondylolisthesis when compared group wise; however, having spondylolisthesis predicted significant reductions in leg pain in the multivariable analysis, as patients with spondylolisthesis had reduced leg pain (by 16 mm) on the VAS scale. Admittedly, this study was not designed to answer the question of how to treat spondylolisthesis, but patients in the spondylolisthesis group were older and had smaller dural sac area—and despite this, significant pain reduction was observed in the multivariable analysis. No patients were fused in the spondylolisthesis group, as all the patients had low-grade spondylolisthesis with essentially the same clinical presentation as the patients with spinal stenosis without spondylolisthesis. Previous studies have often included patients with and without concomitant spondylolisthesis subsequently treated with decompression with or without fusion (Yukawa et al. 2002
, Athiviraham et al. 2011
), perhaps making the cohorts from these studies more heterogenous and evaluation of prognostic factors more difficult. We believe that it was justified to include patients with low-grade slip, in our study as the clinical syndrome is essentially the same.
Change in intensity of back pain correlated with the narrowness of the dural sac area (patients with smaller dural sac area tended to improve more). The narrowness of the dural sac area only explains a small proportion (5%) of the improvement in back pain, however. Also, patients with large dural sac areas had higher levels of leg pain at follow-up. Some earlier studies have also shown that patients with more severe compression of the cauda equina improve more (Herno et al. 1994
, Aliashkevich et al. 1999
). However, the correlation between symptoms and the dural sac area has been considered to be poor. Boden et al. (1990)
showed that 20% of asymptomatic subjects over 60 years old had spinal stenosis on MRI. More recently, Haig et al. (2007)
showed that imaging could not differentiate symptomatic individuals from asymptomatic individuals. Yukawa et al. (2002)
showed that the severity of central canal narrowing at a single level did not appear to limit the postoperative improvement in functional and patient-reported outcome. Recently, Kanno et al. (2012)
also showed a correlation between severity of symptoms preoperatively and dural sac area on an axially loaded MRI in patients planned for surgery. Also, in a surgical cohort, Ogikubo et al. (2008)
showed that dural sac area is a powerful predictor of preoperative walking ability, pain, and HRLQoL. In their 10-year follow-up study, Amundsen et al. (2000)
could not establish a relationship between the severity of the stenosis and outcome. Also, 3 other recent studies have failed to show any relationship between preoperative symptoms and MRI findings (Sirvanci et al. 2008
, Zeifang et al. 2008
, Sigmundsson et al. 2011
). Barz et al. (2010)
have shown that packed nerve roots (positive sedimentation sign) occur consistently in spinal stenosis.
In many countries, decompressive surgery for lumbar spinal stenosis is the most commonly performed spinal operation. The surgical technique is well established, but indications for surgery, diagnosis, and imaging are not well defined—with considerable geographical variation (Bederman et al. 2011
). After surgery, most patients improve in function, pain, and quality of life, and surgical treatment has recently been shown to be superior to nonoperative treatment (Malmivaara et al. 2007
, Weinstein et al. 2008
). Despite this, many patients have residual symptoms and HRQoL inferior to that of the background population (Strömqvist et al. 2009
, Jansson et al. 2009
). The results of our study can be used to improve patient information and selection of patients for surgery.
Some of the factors known to have prognostic value could not be included in our study: most importantly, depression (Sinikallio et al. 2011
), smoking (Sandén et al. 2011
), and BMI (Athivariham et al. 2011). In our opinion, the most important factors for prediction of outcome are the clinical findings that most often lead to surgery, i.e. leg pain, back pain, and preoperative walking distance. Our cohort of patients was elderly with high pain levels, inferior function, low HRLQoL, and very narrow central dural sac area. All these factors can be used to reduce the surgeon selection bias.
Our analysis has identified multiple predictors of outcome in a well-described patient database using both patient- and imaging-related factors with a prospective follow-up. In addition, the study has identified modifiable predictors of outcome, thus possibly enabling improvement in patient care.
We believe that the results of our study can be included in the preoperative patient information as an aid for the patient and the surgeon in shared decision-making. Some caution must, however, be advocated in generalizing the results from a single center cohort to broader surgical practice, as surgeons elsewhere would perhaps treat these patients differently and this relates mainly to our patients with concomitant spondylolisthesis. Another concern is the lack of a control group, as we do not know whether the predictors identified in our cohort would predict the same outcome in patients not undergoing surgery or undergoing different types of surgery.