Preoperative Predictors for Postoperative Clinical Outcome in Lumbar Spinal Stenosis

Preoperative Predictors for Postoperative Clinical Outcome in Lumbar Spinal Stenosis Systematic Review SPINE Volume 31, Number 18, pp E648 E663 2006, Lippincott Williams & Wilkins, Inc. Timo J. Aalto, MD,* Antti Malmivaara, MD, DMSc, Francisco Kovacs, MD, PhD, Arto Herno, MD, DMSc, Markku Alen, MD, DMSc, Liisa Salmi, MA, MS(Hons), Heikki Kröger, MD, DMSc,** Juan Andrade, MD, Rosa Jiménez, MD, PhD, Antti Tapaninaho, MD, DMSc, Veli Turunen, MD,** Sakari Savolainen, MD, DMSc, and Olavi Airaksinen, MD, DMSc From the *Department of Surgery, Kuopio University, Kuopio, Finland; Department of the Physical and Rehabilitation Medicine, Kuopio University Hospital, Kuopio, Finland; Finnish Office for Health Care Technology Assessment/National Research and Development Centre for Welfare and Health, Helsinki, Finland; Departamento Cientifico. Fundacion Kovacs. Palma de Mallorca, Spain; Department of Health Sciences, University of Jyvaskyla, Jyvaskyla, Finland; Kuopio University Library, Kuopio University Hospital Medical Library, Kuopio, Finland; **Department of Surgery/Orthopaedics, Kuopio University Hospital, Kuopio, Finland; Bone and Cartilage Research Unit, University of Kuopio, Kuopio, Finland; Rehabilitation Department, Hospital de Jaen, Jaen, Spain; Research Department, Hospital Hermanos Ameijeiras, and Instituto Superior de Ciencias Medicas de la Habana, La Habana, Cuba; Hospital Lasaretti, Kuopio, Finland; and Department of Neurosurgery, Kuopio University Hospital, Kuopio, Finland. Acknowledgment date: July 6, 2005. First revision date: February 7, 2006. Acceptance date: March 30, 2006. Supported by EVO-grant 2004, 2005 and 2006 (Kuopio University Hospital), a grant from the Finnish Cultural Foundation (Finland) and the Kovacs Foundation (Palma de Mallorca, Spain). No commercial party having a direct or indirect interest in the subject matter of this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. The manuscript submitted does not contain information about medical device(s)/drug(s). Foundation, Professional Organizational funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript. Address correspondence and reprint requests to Timo J. Aalto, MD, Department of the Physical and Rehabilitation Medicine, Kuopio University Hospital, P.O. Box 1777, FIN-70211 Kuopio, Finland; E-mail: timo.aalto@kuh.fi Study Design. Systematic review. Objective. To define preoperative factors predicting clinical outcome after lumbar spinal stenosis (LSS) surgery. Summary of Background Data. LSS is the most common reason requiring lumbar spine surgery in adults older than 65 years. There are no published systematic reviews on this topic. Methods. A literature search was done until April 30, 2005. Included were randomized controlled or controlled trials or prospective studies dealing with operated LSS. The preoperative predictors had to be presented. Included articles were assessed as high-quality (HQ) and low-quality studies. The predictors in HQ studies were considered as the main results. Results. A total of 21 articles were included. Depression and walking capacity were predictors according to 2 HQ studies. Predictors reported in 1 HQ study were cardiovascular/overall comorbidity, disorder influencing walking ability, self-rated health, income, severity of central stenosis, and scoliosis. Conclusion. Depression, cardiovascular comorbidity, disorder influencing walking ability, and scoliosis predicted poorer subjective outcome. Better walking ability, self-rated health, higher income, less overall comorbidity, and pronounced central stenosis predicted better subjective outcome. Male gender and younger age predicted better postoperative walking ability. The predictive value may be outcome specific; thus, the use of all relevant outcome measures is recommended when studying predictors of LSS. Key words: stenosis, predictor, outcome measure, clinical outcome, surgery. Spine 2006;31:E648 E663 Lumbar spinal stenosis (LSS) is the most common reason requiring lumbar spine surgery in adults older than 65 years. 1 Good to excellent results have been reported on average in 64% of cases. 2 Studies evaluating the predictors of LSS surgical outcome generally have been retrospective and have considered only a limited number of potential predictors. 3 An attempted meta-analysis of surgery for LSS in 1992 could identify no predictors, including only 3 clearly prospective studies. 2 In the metaanalysis conducted in 1997 comparing surgical procedures for degenerative LSS, there was also the finding that patients with multiple symptoms had poorer overall result irrespective of the surgical method, 4 indicating that preoperative factors may independently predict postoperative outcome. Even though prognostic aspects have been better considered in articles published during the last decade, there has been no systematic review of predictors in LSS. Systematic reviews are applicable to all types of research design, and a methodology has been developed also for studies on prognostic variables (factors). 5 The preoperative predictors are crucial because they are the only information available before the operation. Preoperative symptoms, radiologic and clinical findings, and other patient-centered relevant matters combined with the physician s experience all influence the decision on whether surgical treatment or even a consultation is appropriate. The purpose of this review was to identify preoperative predictors of prospectively studied outcome of LSS E648

Lumbar Spinal Stenosis Aalto et al E649 surgery and to define their predictive value. The secondary interest was also to define the clinically relevant main outcome measures by which these predictive values can be determined. Methods Study Selection Search Strategy. Computerized literature searches for the surgery, treatment outcome, and prognosis of spinal stenosis were performed until April 30, 2005 (Table 1, details in Appendix 1). The type of article was indicated by the National Library of Medicine, MeSH term prospective studies. Articles with the free text word prospective were also included. Inclusion Criteria for Selection of Studies 1. Randomized controlled trial (RCT), controlled trial, or prospective cohort study 2. Study had to deal with operated lumbar spinal stenosis patients describing preoperative predictors for only surgical treatment 3. The study had to include clinical outcome measures Quality Assessment and Data Extraction. Every included article was assessed by 2 independent reviewers according to predefined assessment criteria (Appendix 2). The criteria were adapted from Borghouts et al and modified for our review. 6 Thirteen criteria were divided into six categories as follows (points of each category in parenthesis): study population (0 2), study size (0 2), number of prognostic factors (0 2), follow-up including dropouts (0 4), outcome measures used/reported (0 2), and analysis (0 1) (Appendix 2). Each item of a selected study which met these criteria was assigned a (1 point, positive). If the item did not meet the criteria or was insufficiently or not described at all, no points were assigned. Thus, the maximum score was 13 points. Those articles scoring 7 points ( 50%) or higher were considered to be high quality (HQ) studies, and articles scoring less than 7 points were rated as low quality (LQ). The predictors presented in each original article were extracted, and their predictive value was determined as neutral, positive, or negative in each original article. The uniformity of the double assessment was verified (score, predictors, and their predictive value); and in the case of disagreement, a third opinion was sought. The predictors originating from HQ articles were considered to be the main results of this study. Table 1. Databases Used in Literature Search 1) PubMed/Medline (National Library of Medicine, Bethesda, MD) 2) Cochrane Library 3) NHS Centre for Reviews and Dissemination (University of York, York, UK) databases DARE, NHSEED, HTA 4) Medic (Terkko, National Health Sciences Library, Helsinki, Finland) 5) Indice Medico Español (Spanish National Database, including papers published in Spanish; starting 1971) 6) LILACS (Latin American and Caribbean Health Sciences Data Base, including papers published in Spanish and Portuguese; starting 1982) Results All the original studies in the results section are HQ studies and the reported predictors have a statistically significant association (P 0.05) with the outcome, unless otherwise stated. Details of the studies are presented in Appendix 3. Results of the Search A total of 885 abstracts from databases were found. All the potential articles were read, and ultimately 21 were included (Appendix 4). More than 85% of studies were excluded mostly because they were retrospective, the indication was not LSS, or they were not original studies. Methodologic Quality Eight (38%) of the 21 articles were graded as HQ (Appendix 3A). Quality percents of each article are presented in Appendix 3. The percentages of articles fulfilling each of the 13 criteria are presented in Table 2. There was a clear variation in the amount of predictors and used outcome measures. Five of eight HQ studies had a minimum of 2-year follow-up. Table 2. Methodologic Quality: Percentage of Articles Meeting the Criteria, High Quality (HQ) and Low Quality (LQ) Criteria Study population (0 2 p) Selection of study population ( 1 p) Description of inclusion/ exclusion criteria All Articles (n 21) n (%) HQ Articles (n 8) n (%) LQ Articles (n 13) n (%) 20 (95) 8 (100) 12 (92) 13 (62) 6 (75) 7 (54) ( 1 p) No. of prognostic factors (0 2 p) 3 (0 p) 4 (19) 4 (31) 3 6( 1 p) 5 (24) 5 (38) 6 ( 1 p) 12 (57) 8 (100) 4 (31) Study size (0 2 p) 100 patient-year/ subgroup (0 p) 15 (71) 4 (50) 11 (85) 100 patient-year/ 3 (14) 1 (12.5) 2 (15) subgroup ( 1 p) 200 patient-year/ 3 (14) 3 (37.5) subgroup ( 1 p) Follow-up (0 4 p) 24 mo 11 (52) 5 (62.5) 6 (46) ( 1 p) Dropouts 20% (0 p) 4 (19) 1 (12.5) 3 (23) 10% 20% ( 1 p) 5 (24) 2 (25) 3 (23) 0% 10% ( 1 p) 3 (14) 2 (25) 1 (8) Perfect ( 1 p) 9 (43) 3 (37.5) 6 (46) Outcome measures (0 2 p) Used 4; at least 1 objective and subjective ( 1 p) 7 (33) 7 (87.5) 0 (0) Reported 4 outcome 4 (19) 4 (50) 0 (0) measures ( 1 p) Data analysis ( 1 p) 6 (29) 3 (37.5) 3 (23)

E650 Spine Volume 31 Number 18 2006 Predictors Predictors, their predictive values, used outcome measures, and number of original articles are presented in Table 3. Radiologic Predictors. Pronounced stenosis of the spinal canal (anteroposterior diameter 6 mm in myelography) predicted less postoperative pain (5-year follow-up). 7 Dural tube cross-sectional area 70 mm 2 in magnetic resonance imaging was associated with better postoperative Oswestry score compared with patients having cross-sectional area 70 mm 2. 8(LQ) Stenosis classification (acquired/mixed), 9 (acquired/congenital/mixed), 10 type of stenosis (lumbar stenosis; lumbar stenosis and herniated disc; lateral recess stenosis), 11 and minimal/ mean canal diameter 9 had no predictive values in terms of several outcome measures. A greater degree of preoperative scoliosis predicted more postoperative back pain; the extent of preoperative spondylolisthesis showed no consistent association with back/leg pain and satisfac- Table 3. Predictors in Lumbar Spinal Stenosis Predictor Predictive Value Outcome Measures No. of Original Articles HQ LQ Comorbidity, health and disability Depression Neg* Satisfaction (2-yr follow-up), symptoms (2-yr follow-up) Pain, walking capacity (2-yr follow-up) 2 1 Better reported walking capacity Pos* Walking capacity (2-yr follow-up), satisfaction 2 (2-yr follow-up) Disorder influencing walking ability Neg Pain (5-yr follow-up) 1 Cardiovascular comorbidity Neg Walking capacity (2-yr follow-up), satisfaction 1 (2-yr follow-up), symptoms (2-yr follow-up) Less comorbidity Neg* Global function 1 More extensive medical Neg Satisfaction 1 comorbidity 1 1 Better self-rated health Pos Walking capacity (2-yr follow-up), satisfaction 1 (2-yr follow-up), symptoms (2-yr follow-up) Symptoms Shorter duration of symptoms in Pos Pain (2-yr follow-up) 1 1 lateral stenosis 2 3 Back pain Neg Satisfaction 1 2 2 Continuous pain 1 Radiologic predictors Pronounced central stenosis Pos Pain (5-yr follow-up) 1 1 1 Preoperative scoliosis Neg Back pain (2-yr follow-up) 1 1 Extent of preoperative 2 spondylolisthesis Type of stenosis 3 Coexistent disc herniation and 1 LSS Aorta calcification 1 Education, work, income Time off work 1 1 Workload 3 1 Higher income Pos Walking capacity (2-yr follow-up), satisfaction 1 (2-yr follow-up), symptoms (2-yr follow-up) Expectations More ambitious expectation of Pos* Physical function, satisfaction 1 physical function Greater no. of pain relief Neg* Pain relief 1 expectations Other Age 4 7 Younger age Pos* Walking capacity 1 Gender 4 7 Male Pos* Walking capacity 1 Findings in clinical examination 2 Obesity 1 Smoking 1 2 Marital status 1 indicates not significant; HQ, high quality; LQ, low quality; Neg, negative; Pos, positive. *Six-month follow-up. Predictors having significant predictive value in 1 original HQ study.

Lumbar Spinal Stenosis Aalto et al E651 tion. 12 Preoperative scoliosis had no association with walking capacity, symptom severity, or satisfaction. 3 Amundsen et al assessed many radiologic parameters but could detect no association with outcome (degenerative changes at different parts of vertebral segment, occurrence of pseudospondylolisthesis, occurrence of transition vertebrae, type of stenosis, occurrence of redundant nerve roots, degree of narrowness, root affliction, and measures of different dimensions of the spinal canal). 13 Health, Functional Ability, and Somatic Comorbidity. Better preoperative walking capacity predicted better postoperative walking capacity 3,14 and improved satisfaction (related to current state). 3 Greater preoperative cardiovascular comorbidity (and overall comorbidity only in univariate analysis) predicted worse walking capacity, treatment satisfaction, and more severe symptoms. 3 Preoperative better self-rated health (scale: excellent, good, fair, poor) was associated with better outcome (walking capacity, symptoms, treatment satisfaction). 3 A disorder affecting walking ability (hip or knee arthrosis, neurologic disease, cardiopulmonary disease) predicted worse postoperative pain. 7 At the 6-month follow-up, more extensive medical comorbidity associated with poorer treatment satisfaction, 15(LQ) and less comorbidity with better global function. 14 With respect to comorbidity, only 1 LQ study reported no association with outcome. 9(LQ) Depression and Psychologic Variables. Baseline depression (Zung Depression Scale) was associated with more pain at the 6-month follow-up. 14 Depression measured with a 3-item scale was associated with worse treatment satisfaction (P 0.05) and more severe symptoms in a multiple linear regression model; in univariate analysis, an association with poorer walking capacity was also noted. 3 A higher depression score (Self-Rating Depression Scale) was associated with poorer global satisfaction in univariate but not in multivariate analysis. 15(LQ) Patients with many preoperative expectations, particularly among patients with low baseline function, reported more improvement in postoperative function than patients with few expectations. More preoperative ambitious expectations for postoperative physical function were also associated with improved postoperative functional ability and satisfaction. Conversely, having more numerous pain relief expectations was associated with more postoperative pain. 14 Patients with unrealistic preoperative expectations of the outcome of their surgery (general health, subjective functional ability, life satisfaction) tended to have lower levels of postoperative satisfaction. 16(LQ) Symptoms. In lateral stenosis, shorter duration of symptoms predicted less postoperative pain. 17,18(LQ) In central stenosis, 1 article reported a nonsignificant tendency toward better outcome (sciatica-claudication 4 years), 7 but no association with outcome was reported in 4 articles. 9,13,19 20(LQs) Back pain predominance was associated significantly with worse satisfaction related to results of surgery. 15(LQ) Patients with no preoperative back pain had a tendency toward better outcome, 7,21(LQ) however, without significant values presented. In 2 studies, localization of pain had no association with outcome (several outcome measures). 3,20(LQ) Other Predictors. Age had no association with outcome in 4 HQ studies 3,7,13,22 and in 7 LQ studies. 9,10,15,18,20,21,23 Iversen et al reported a significant association between younger age and better postoperative walking ability. Men had better postoperative walking ability. 14 Gender had no association with outcome in 4 HQ studies 3,11,13,22 and in 7 LQ studies. 9,10,15,18 20,23 Education had no association with outcome, but higher income associated with poorer postoperative walking ability, treatment satisfaction, and greater symptom severity. 3 No associations were observed with the findings in clinical examination, 3,13 marital status, 13 obesity, 11 sedentary work (however, more sedentary workers returned their previous work), 17 workload (not specified), 13,18(LQ),22 time off work, 13,18(LQ) and smoking. 9(LQ),13,20(LQs) Regular consumption of analgesics in patients with central stenosis predicted more postoperative pain, 18(LQ) however, with no statistical significance presented. The recovery of extensor hallucis longus (L5) power was associated with diagnosis (central stenosis, lateral stenosis, and herniated nucleus pulposus [HNP]). In the 2-year follow-up, the total recovery was 0%, 0%, and 56%, respectively. The partial recovery was 0%, 80%, and 28%, respectively. The result was reported statistically significant between diagnoses, including the HNP group. 19(LQ) Lateral stenosis predicted nonsignificant inferior results compared with LSS and LSS HNP. 11 Discussion The predictive value (negative, positive, neutral) of each predictor in the original study was simply extracted by patient-centered outcome measures such as pain, functional ability, and satisfaction as described in the original articles. In earlier meta-analyses, 2,4 the overall outcome was evaluated on a 3-grade scale (excellent-good, fair, poor) of pain and physical activity, but the specific predictive value of each predictor with respect to separate main outcome measures was not reported, obviously because of very heterogenous use and limited reporting of outcome measures. In addition to pain and disability, satisfaction as an outcome measure is also recommended for standardized use in low back pain research, 24 which is increasingly used in recent decade. We think that identifying first the predictors, and then reported specific predictive values according to used main clinically relevant outcomes, is important because one predictor may predict better postoperative functional ability, but not less pain, for example. Thus, if the predictive value of the predictor is in that way outcome-specific, the comparing

E652 Spine Volume 31 Number 18 2006 or pooling data of articles studying the same predictors with different kind of outcome measures is not judged in our opinion. Consequently, when the predictor has negative or positive predictive value, the used outcome measure should report in the same context, as well as the postoperative time of follow-up. In all 8 HQ articles of our review, a minimum of 6 predictors were studied in each article, and 7 of 8 articles used 4 outcome measures (Table 2). Better self-rated health, higher income, and greater cardiovascular comorbidity were the only predictors that had an influence on all used 3 main outcome measures (walking capacity, treatment satisfaction, symptom severity) of the same original study. Depression and walking capacity were the only predictors reported in 2 HQ studies. Depression predicted poorer outcome. Walking capacity was studied as a predictor but also used as an outcome measure. A better preoperative walking capacity predicted better postoperative satisfaction and walking capacity. Our main intention was to reach those predictors presented in only prospective studies. We considered only statistically significant predictive values (P 0.05), even if it was reported in only 1 HQ study. Nonsignificant predictive values might due to factors such as limited number of patients and different determinations of outcome measures and are not at all a conclusive proof that the predictor in question indeed possesses no predictive value. Therefore, even though there were a few additional reports not showing predictive value for the same predictor even with the same outcome measures, we did rely on a statistically significant result in a HQ study reporting significant predictive value. As an example of this, male gender and younger age were associated with better postoperative walking ability in 1 HQ study. 14 There were 4 HQ and 7 LQ articles reporting no predictive value for both age 3,7,9,10,13,15,18,20 23 and gender. 3,9 11,13,15,18 20,22,23 Two of the former HQ studies reporting neutral predictive value for age 3,7 had walking capacity as an outcome measure, in contrast to the one showing an association. Respectively with gender, only one of the latter HQ studies 3 used and reported walking capacity in contrast to the one having association. These two predictors did not predict pain, satisfaction, and functional ability except walking. In general, there were no opposite predictive values (i.e., positive vs. negative) with respect to any factor in our review. In literature, female gender was reported as a negative predictor according to Beaujon score (measuring pain, neurologic deficit, medications, and quality of life as a summary score) and low back pain; 25 back and leg pain improvement, and Oswestry Disability Index. 26 Katz et al stated that women are operated at a more advanced stage in the course of their disease but fared as well as men. 27 Male gender doubled the relative risk of dying after the operation according to retrospective study by Jansson et al. 28 The other mentioned factors in that study were age, surgical complexity, and time period when operation was done. Also, age over 80 years and complexity of surgery increased postoperative mortality risk. However, comorbidity was not adjusted. 28 Since many predictors were continuous or ordinate variables in the original studies, we were obliged to name some predictors with such elusive terms as less comorbidity or better reported walking capacity, which terms are used in original articles because obviously no more exact cutoff values could not be defined. The more clear-cut predictors were depression, a disorder influencing walking ability and male gender. An example of a statistically continuous predictor is shorter preoperative duration of symptoms, which predicted less postoperative pain in lateral stenosis only. The reported mean durations of back and leg pain were 54 months (range, 1 480) and 30 months (range, 1 150), respectively. 17 In another example, long preoperative duration of sciatica in lateral stenosis associated with worse outcome (pain), among patients having a mean duration of symptoms of 30 months. 18(LQ) No exact cutoff values, after which the prognosis is poorer, could be determined. Furthermore, the back and leg pain were not reported separately in those original articles. However, leg pain duration of less than 30 months may be considered as a positive predictor in lateral stenosis, although this result need to be confirmed in future studies. Predominance of back (as opposed to leg) pain was a negative predictor for satisfaction related to result of surgery in 1 LQ study 15(LQ) being in agreement with other reports not included in this review. 25,29 Increased comorbidity has been reported to be a predictor for poor outcome, 30 and underlying subclinical vascular factors may be involved in the complaints after LSS surgery. 9(LQ) Cardiovascular comorbidity was the most common reason (46%) of death after LSS surgery, the total fatality rate within 30 days being 3.5 per 1000 operations. 28 Arinzon et al reported poorer surgical outcome with elderly LSS diabetic patients compared with elderly nondiabetics, but the outcome also depended on other comorbidities. 31 In our review, depression, cardiovascular comorbidity, and a disorder affecting walking ability were specific comorbid predictors. When a symptomatic comorbid disease is present as a confounding factor in addition to LSS, the question arises do the outcome scores measure in a valid manner the change of pain/functional ability attributable to the decompressive LSS operation? In our opinion, there is a speculative possibility that the negative predictive value of somatic comorbid conditions may be at least partly the straight influence of comorbid disease to outcome measures. For example, the differentiation of pain source in symptomatic coxarthrosis and LSS is sometimes difficult. Cardiovascular symptoms have been reported to limit walking capacity directly, but the mechanism

Lumbar Spinal Stenosis Aalto et al E653 through which cardiovascular comorbidity leads to persistent pain is only suggested. 3,32 In psychologic disorders, the same question of the effect of comorbidity on the outcome scores arises. Patients with pure stenosis reported better postoperative quality of life than patients having concomitant diseases. 33 Patients with higher distress reported greater disability. 34 Thus, there is still the speculative possibility that the depressed patient enjoys the same objective benefit of surgery but nonetheless reports a poorer outcome because of depression. Another explanation of poorer prognosis of patients with comorbid conditions is that the comorbid disease may influence on postoperative recovery. Whether surgery benefits patients with comorbid conditions, despite of their poorer prognosis, as well as those without can be answered only in a randomized trial. The rather rare objective outcome measures were treadmill test, 8 working ability, 11,13,17,22,35 and neurologic/clinical examination. 11,13,17,22,25,36 No predictors were identified for these outcome measures. The recovery of extensor hallucis longus power was the only objective outcome measure having association with the diagnosis. 19(LQ) Criteria List The criteria of Borghouts et al 6 were modified for the purposes of our review. The quality of a study was based on internal (methodologic) and external (generalizability) validity. The homogeneity of the studies was enhanced by our restriction to prospective design with all patients being at the same operative stage of their disease. Our framework for assessing internal validity of prognostic studies also fulfilled the criteria presented by Altman, including the sampling of patients, defined prognostic variables, follow-up times, outcomes, comparators, and as well as analysis. 5 However, we excluded the intervention itself as the assessment of the effect of surgery demands randomized controlled trial study design. The only predictor with statistically significant predictive value that did not occur in HQ studies was back pain predicting worse patient s clinical satisfaction. 15(LQ) Limitations of This Review A potential limitation might be due to an incomplete literature search. We think, however, that especially PubMed and other five databases include the most important articles of this field. We assessed articles in English, Finnish, French, German, Spain, and Swedish. Articles written in other languages may have been missed. The reviewed literature likely does not include all possible predictors, and it is also possible that all predictors are not yet studied in prospective setting. We focused only in preoperative predictors. In addition to operation type itself, other preoperative predictors (e.g., duration of operation) and also postoperative predictors (e.g., rehabilitation) may have also influence on outcome. This review do not explain the mechanisms in which way predictors influence on outcome. Usefulness of Results and Recommendations The present study may help clinicians in their assessment of prognosis of a patient when considering surgical treatment. However, prognostic studies are not designed to answer the question what is the indication for surgery. This review may be also useful in planning future studies on prognosis after surgical treatment of LSS. The notification that predictive value of one predictor may be different between (main) outcome measures is important to recognize. Attention should be paid to the use of all relevant patient centered outcome measures before and after surgery, and the evaluation of comorbid conditions. However, this may not be feasible if the comorbid disease has a low prevalence in the study population. Conclusion The outcome of surgical treatment of LSS was assessed mainly by using subjective outcome measures. Next predictors, for 2- to 5-year follow-up, were identified (6- month follow-up presented separately). Predictors are italicized. Predictors Reported at Least in 2 HQ Studies 1. Reported walking capacity. Better preoperative walking capacity predicted better postoperative capacity in 2 HQ studies and better satisfaction in 1 HQ study. 2. Baseline depression predicted more pain in 1 HQ study (6-month follow-up); worse symptom severity, less treatment satisfaction, and poorer walking capacity (univariate analysis) were reported in another HQ study, and less global satisfaction (univariate analysis) in 1 LQ study. Predictors Reported in 1 HQ Study Better self-rated health and higher income both predicted less severe symptoms, better walking capacity, and treatment satisfaction; greater cardiovascular comorbidity was a negative predictor with respect to the same 3 outcome measures in the same original study. Less overall baseline comorbidity predicted better global function in 1 HQ study and better treatment satisfaction in 1 LQ study. A disorder influencing walking ability predicted poorer postoperative pain. Pronounced stenosis predicted less postoperative pain in central stenosis (1 HQ study) and better postoperative Oswestry Disability Score in 1 LQ study. Scoliosis predicted more severe postoperative back pain.

E654 Spine Volume 31 Number 18 2006 In 6-month follow-up, preoperative expectations predicted also subjective outcome, and male gender and younger age predicted better postoperative walking ability. The predictive value may be outcome specific; thus, the use of all relevant outcome measures is recommended when studying predictors of LSS. Key Points Prospective studies of preoperative predictors of lumbar spinal stenosis surgery were systematically reviewed. The computerized search identified 21 eligible studies, which were assessed with predefined criteria list as high-quality (HQ) or lowquality (LQ) studies. Eight articles were assessed as HQ studies. The predictors appearing in HQ studies were considered to be the main results. Depression and walking ability were independent predictors in two original HQ studies. Predictors reported in one original HQ study were cardiovascular comorbidity, overall comorbidity, a disorder influencing walking ability, self-rated health, income, severity of central stenosis, and scoliosis. Better walking ability, self-rated health, higher income, less overall baseline comorbidity, and pronounced stenosis predicted better subjective outcome. Depression, greater cardiovascular comorbidity, a disorder influencing walking ability, and scoliosis predicted poorer subjective outcome. Male gender and younger age predicted better postoperative walking ability. The predictive value may be outcome-specific. Thus, the use of all relevant outcome measures is recommended when studying predictors in lumbar spinal stenosis. References 1. Mazanec DJ, Podichetty VK, Hsia A. Lumbar canal stenosis: start with nonsurgical therapy. Cleve Clin J Med 2002;69:909 17. 2. Turner JA, Ersek M, Herron L, et al. Surgery for lumbar spinal stenosis: attempted meta-analysis of the literature. Spine 1992;17:1 8. 3. Katz JN, Stucki G, Lipson SJ, et al. Predictors of surgical outcome in degenerative lumbar spinal stenosis. Spine 1999;24:2229 33. 4. Niggemeyer O, Strauss JM, Schulitz KP. Comparison of surgical procedures for degenerative lumbar spinal stenosis: a meta-analysis of the literature from 1975 to 1995. Eur Spine J 1997;6:423 9. 5. Altman DG. Systematic reviews of evaluations of prognostic variables. BMJ 2001;323:224 8. 6. Borghouts JA, Koes BW, Bouter LM. The clinical course and prognostic factors of non-specific neck pain: a systematic review. Pain 1998;77: 1 13. 7. Jonsson B, Annertz M, Sjoberg C, et al. A prospective and consecutive study of surgically treated lumbar spinal stenosis: II. Five-year follow-up by an independent observer. Spine 1997;22:2938 44. 8. Yukawa Y, Lenke LG, Tenhula J, et al. A comprehensive study of patients with surgically treated lumbar spinal stenosis with neurogenic claudication. J Bone Joint Surg Am 2002;84:1954 9. 9. Spratt KF, Keller TS, Szpalski M, et al. A predictive model for outcome after conservative decompression surgery for lumbar spinal stenosis. Eur Spine J 2004;13:14 21. 10. Gunzburg R, Keller TS, Szpalski M, et al. Clinical and psychofunctional measures of conservative decompression surgery for lumbar spinal stenosis: a prospective cohort study. Eur Spine J 2003;12:197 204. 11. Javid MJ, Hadar EJ. Long-term follow-up review of patients who underwent laminectomy for lumbar stenosis: a prospective study. J Neurosurg 1998;89: 1 7. 12. Frazier DD, Lipson SJ, Fossel AH, et al. Associations between spinal deformity and outcomes after decompression for spinal stenosis. Spine 1997;22: 2025 9. 13. Amundsen T, Weber H, Nordal HJ, et al. Lumbar spinal stenosis: conservative or surgical management? A prospective 10-year study. Spine 2000;25: 1424 35; discussion 35 6. 14. Iversen, Daltroy, Fossel, et al. The prognostic importance of patient preoperative expectations of surgery for lumbar spinal stenosis. Patient Educ Couns 1998;34:169 78. 15. Katz JN, Lipson SJ, Brick GW, et al. Clinical correlates of patient satisfaction after laminectomy for degenerative lumbar spinal stenosis. Spine 1995;20: 1155 60. 16. McGregor AH, Hughes SP. The evaluation of the surgical management of nerve root compression in patients with low back pain: 2. Patient expectations and satisfaction. Spine 2002;27:1471 6; discussion 1476 7. 17. Jonsson B, Stromqvist B. Decompression for lateral lumbar spinal stenosis: results and impact on sick leave and working conditions. Spine 1994;19: 2381 6. 18. Jonsson B. Patient-related factors predicting the outcome of decompressive surgery. Acta Orthop Scand Suppl 1993;251:69 70. 19. Jonsson B, Stromqvist B. Motor affliction of the l5 nerve root in lumbar nerve root compression syndromes. Spine 1995;20:2012 5. 20. McGregor AH, Hughes SP. The evaluation of the surgical management of nerve root compression in patients with low back pain: 1. The assessment of outcome. Spine 2002;27:1465 70. 21. Jonsson B, Stromqvist B. Lumbar spine surgery in the elderly: complications and surgical results. Spine 1994a;19:1431 5. 22. Jonsson B, Stromqvist B. Repeat decompression of lumbar nerve roots: a prospective two-year evaluation. J Bone Joint Surg Br 1993;75:894 7. 23. Herkowitz HN, Kurz LT. Degenerative lumbar spondylolisthesis with spinal stenosis: a prospective study comparing decompression with decompression and intertransverse process arthrodesis. J Bone Joint Surg Am 1991;73: 802 8. 24. Deyo RA, Battie M, Beurskens AJ, et al. Outcome measures for low back pain research: a proposal for standardized use. Spine 1998;23:2003 13. 25. Mariconda M, Zanforlino G, Celestino GA, et al. Factors influencing the outcome of degenerative lumbar spinal stenosis. J Spinal Disord 2000;13: 131 7. 26. Herno A, Airaksinen O, Saari T, et al. The predictive value of preoperative myelography in lumbar spinal stenosis. Spine 1994;19:1335 8. 27. Katz JN, Wright EA, Guadagnoli E, et al. Differences between men and women undergoing major orthopedic surgery for degenerative arthritis. Arthritis Rheum 1994;37:687 94. 28. Jansson KA, Blomqvist P, Granath F, et al. Spinal stenosis surgery in Sweden 1987 1999. Eur Spine J 2003;12:535 41. 29. Atlas SJ, Keller RB, Robson D, et al. Surgical and nonsurgical management of lumbar spinal stenosis: four-year outcomes from the Maine Lumbar Spine Study. Spine 2000;25:556 62. 30. Katz JN, Lipson SJ, Larson MG, et al. The outcome of decompressive laminectomy for degenerative lumbar stenosis. J Bone Joint Surg Am 1991;73: 809 16. 31. Arinzon Z, Adunsky A, Fidelman Z, et al. Outcomes of decompression surgery for lumbar spinal stenosis in elderly diabetic patients. Eur Spine J 2004;13:32 7. 32. Kauppila LI, McAlindon T, Evans S, et al. Disc degeneration/back pain and calcification of the abdominal aorta: a 25-year follow-up study in Framingham. Spine 1997;22:1642 7; discussion 8 9. 33. Schillberg B, Nystrom B. Quality of life before and after microsurgical decompression in lumbar spinal stenosis. J Spinal Disord 2000;13: 237 41. 34. Millard RW, Jones RH. Construct validity of practical questionnaires for assessing disability of low-back pain. Spine 1991;16:835 8. 35. Keller RB, Atlas SJ, Soule DN, et al. Relationship between rates and outcomes of operative treatment for lumbar disc herniation and spinal stenosis. J Bone Joint Surg Am 1999;81:752 62.

Lumbar Spinal Stenosis Aalto et al E655 36. Mariconda M, Fava R, Gatto A, et al. Unilateral laminectomy for bilateral decompression of lumbar spinal stenosis: a prospective comparative study with conservatively treated patients. J Spinal Disord Tech 2002;15: 39 46. Appendix 1 Literature searches conducted up to 30 April 2005. I. PubMed/Medline (National Library of Medicine, Bethesda, MD). Both the controlled thesaurus of Medline, the Medical Subject Headings (MeSH) and free text words (FREE) in all fields were used. The automatic exploring of the MeSH terms was used throughout the search. 1. Spinal stenosis/surgery (MeSH) 2. Spinal stenosis (MeSH) OR spinal stenosis (FREE) 3. Neurosurgical procedures (MeSH) OR orthopedic procedures (MeSH) OR surgery (MeSH) OR surgery (FREE) OR surgical (FREE) OR operated (FREE) OR operation (FREE) 4. 2 and 3 5. 1or4 A filter composed by the information specialist to articles with at least some degree of evidence restricted the search. The filter was formed by using combinations of publication types and MeSH terms: Journal article By the following MeSH terms denoting evidence: bias (epidemiology) OR comparative study OR epidemiologic methods OR evaluation studies OR evidencebased medicine OR (evidence-based medicine AND nursing) OR follow-up studies OR morbidity OR nursing process OR placebos OR prognosis OR quality of health care OR research OR review literature OR statistics OR survival analysis OR the following publication types: clinical trial OR consensus development conference OR controlled clinical trial OR guideline OR meta-analysis OR multicenter study OR practice guideline OR randomized controlled trial OR review OR validation studies. An additional search: 1. Spinal stenosis/psychology (MeSH) 2. Spinal stenosis (FREE) 3. Psychiatr* (FREE) OR psycholog* (FREE) OR mental (FREE) OR percept* (FREE) OR perceive* (FREE) OR patient acceptance of health care ((MeSH) OR consumer satisfaction (MeSH) OR quality of life (MeSH) 4. 2 and 3 5. 1or4 II. Cochrane Library. The search on Cochrane Library was performed as follows: 1. Spinal stenosis (MeSH) 2. Spinal NEXT stenosis 3. 1or2 III. NHS (National Health Service) Centre for Reviews and Dissemination (University of York, York, U.K.) Databases DARE (Databases of Abstracts of Reviews of Effects), NHSEED (NHS Economic Evaluation Database), HTA (Health Technology Assessment). The search on these databases was carried out as follows: 1. Spinal stenosis (All Fields) 2. Expl spinal stenosis (MeSH) IV. Medic (Terkko, National Health Sciences Library, Helsinki, Finland) is the Finnish National Database for Health Sciences Literature Dating from 1978. This database was searched on as follows: 1. Lanneran* (FREE) & ahta*(free) 2. Spinaalisten* (FREE) 3. Spinal stenosis (MeSH) 4. 1or2or3 V. Indice Medico Espanol (Spanish National Database, Including Papers Published in Spanish. Dating from 1971). 1. Estenosis lumbar (FREE) VI. LILACS (Latin American and Caribbean Health Sciences Data Base, Including Papers Published in Spanish and Portuguese. Dating from 1982). 1. Estenosis espinal (Mesh) 2. Cirugia (FREE) 3. 1 and 2 Appendix 2 Detailed Criteria List Detailed criteria list for assessment of the quality of studies on prognostic factors of operated lumbar spinal stenosis (modified from Borghouts et al: The clinical course and prognostic factors of non-specific neck pain: a systematic review. Pain 77([1998] 1 13). Maximum 13 points. High quality article 7 points (pts). One point ( ) for each item (A I), except C maximum 2 pts, D maximum 2 pts, F maximum 3 pts. Study Population A) Selection of population Positive, if patients were (clinically) examined during 1 year before operation. Also positive in case of a heterogenous population (survival cohort) for which subgroups were identified and analyzed. B) Description of inclusion and exclusion criteria Positive, if i) in inclusion criteria, mentioned at least quality of symptoms and radiological diagnostic method (myelography, CT, MRI) AND ii) exclusion criteria mentioned

E656 Spine Volume 31 Number 18 2006 Prognostic factors C) Description of potential prognostic factors a) Positive, if at least 3 main items of 11 reported at baseline, OR 2 main items and 2 minor items OR 1 main items and 3 minor items b) Positive, if at least 6 main items of 11 reported at baseline OR 5 main items and 2 minor items OR 4 main items and 3 minor items Eleven major items: Preoperatively Pain. Descriptive values Localization of complaints (local back pain or radiation towards inferior extremities; number or percentage) Duration of complaints Subjective disability Subjective health assessment Objective disability (e.g., ability to work, physician s clinical assessment, treadmill testing or other acceptable disability measurement) Objective comorbidity (e.g., diseases from medical records; frequency, percentage) Radiological stenosis (graded e.g., severe/moderate/ light, or values describing stenosis presented (e.g., dural sac area); also positive, if coexistent radiological findings in spine presented (discus pathology, spondylolisthesis, scoliosis for example) Indications for operation defined Used operative technique described Psychological prognostic factors presented. Descriptive values. Minor items for example: Age Sex Obesity Smoking Earlier spine operation Neurophysiological or other diagnostic methods used and results presented Also other factors described as having potential prognostic value accepted Study Size D) a) positive, if available information on clinical course, for every (sub)group, was 100 patientyears (number of patients followed-up, multiplied by the follow-up duration [in years]). b) positive in case 2 or more subgroups were similar at baseline and compared for the difference in 1 prognostic factor, and the total available information on clinical course was patient-years for every reported (sub) group (number of patients followed-up, multiplied by the follow-up duration (in years). Also positive if no (sub)group analysis was made in case of a homogeneous cohort at baseline for at least: 6 of the 11 main prognostic factors OR 5 main items and 2 minor items OR 4 main items and 3 minor items Follow-up E) Positive, if the follow-up period was 24 months or more. F) a) Positive if total number of drop-outs/loss to follow-up 20%. b) Positive if total number of drop-outs/loss to follow-up 10%. c) Positive if demographic/clinical information was presented for completers and those lost to followup/drop-outs, or no drop-outs/lost to follow-up. Lost to follow-up/drop-outs: all patients of the assembled cohort minus the number of patients at the main moment of health status measurement for the main outcome measure, divided by all patients in the assembled cohort. Outcome Measures G) Positive, if at least 4 of the following 8 items were used as outcome measures, AND at least 1 subjective AND 1 objective outcome measure used. Subjective outcome measures: 1) pain 2) subjective benefit for surgery, 3) quality of life 4) self-reported functional status. Objective outcome measures: 5) ability to work 6) physician s clinical assessment 7) treadmill testing or in some other way measured walking distance; OR other acceptable disability measurement 8) consumption of analgesics or use of medical services reported. H) positive if frequency or percentage (or mean and SD/ CI) of at least 4 of the following 8 outcome measures were used for each follow-up measurement: Subjective outcome measures: 1) pain 2) subjective benefit for surgery, 3) quality of life 4) self-reported functional status. Objective outcome measures: 5) ability to work 6) physician s clinical assessment 7) treadmill testing or in some other way measured walking distance; OR other acceptable disability measurement 8) consumption of pain killers or use of medical services reported. Data Analysis I) Positive if adjusted proportions were provided. To adjust for prognostic factors, appropriate techniques like logistic regression or Mantel-Haenzel were used (the number of events in the model, should be at least 10 times the number of prognostic variables studied). Also positive if unadjusted proportions of subsets were similar for 50% prognostic factors reported at study and assessed by a Chi-square test or if no subsets were presented in the case of a homogeneous cohort at baseline, for 50% prognostic factors reported at study.

Lumbar Spinal Stenosis Aalto et al E657 APPENDIX 3 Table 4. Details of the Studies Author (year) Reported Prognostic Factor(s) Influence on Outcome Study Title (N, F) Objectives A. High Quality Studies Including Predictors for Postoperative Outcome of Operated LSS (n 7) Frazier (1997) i) Observational study Quality 62% ii) Laminectomy, medial facetectomy when necessary. 23% had concomitant arthrodesis with or without instrumentation iii) Questionnaires by Stucki et al 1996 including back/leg/overall pain, and difficulty ambulating using Likert-scale from 1 (best) to 5 (worst); walking capacity (5 questions); satisfaction (6 questions) Greater preoperative scoliosis Associated with back pain (P 0.05, r 0.29 (6 mo), r 0.25 (24 mo), not with satisfaction, leg pain, or walking capacity Associations between spinal deformity and outcomes after decompression for spinal stenosis To examine associations between radiographic parameters (scoliosis and olisthesis) and outcomes of surgery for degenerative LSS The extent of preoperative spondylolisthesis N 90 F 2 yrs Iversen (1998) i) Observational, multicenter study Quality 62% ii) Decompressive surgery for LSS (23% one-level, 33% two-level, 44% 3 levels; arthrodesis for 29%) iii) Functional status (Sickness Impact Profile), walking capacity (4 questions), pain (6-point scale), postoperative resource use; satisfaction with pain relief (4 questions) and physical function (2 questions) Javid (1998) Quality 62% Jönsson (1993a) Quality 62% Preoperative expectations (e) Greater number of e:s Associated with improved global function (P 0.003) More e:s with pain relief Associated with more pain (P 0.009) Higher baseline pain Associated with more pain (P 0.003) Baseline depression - - (P 0.003) Better baseline global Better global function function (beta 0.5, P 0.0001) Fewer comorbidities - - (beta 0.06; P 0.005) Better walking capacity Better walking capacity (P 0.0001) Younger age - - (P 0.0006) Male gender - - (P 0.003) Higher baseline walking Better satisfaction with capacity walking capacity (P 0.0001) Female gender - - (P 0.025) More ambitious - - (P 0.043) expectation of physical function The prognostic importance of patient preoperative expectations of surgery for lumbar spinal stenosis. N 257 F 6 mos To a) relate patient expectations of surgery to baseline function and pain b) determine how patients expectations and preoperative function interact to predict postoperative outcome i) Observational study ii) Decompressive laminectomy; bilateral laminectomy and medial facetectomy for LSS (n 86; fusion for nine patients) and LSS with HNP (n 61). 23 patients had lateral stenosis iii) Pain, satisfaction, clinical examination, employment Diagnoses as predictive factors: LSS vs. LSS HNP (6 wk; 6 mo; 1 yr; and 1 11 yr follow-ups) Long-term follow-up review of patients who underwent laminectomy for lumbar stenosis: a prospective study To evaluate long-term results (of surgery) in patients with symptomatic spinal stenosis Lateral stenosis vs. (lat. stenosis: a trend N 170 (LSS HNP, LSS) with inferior result) Obesity and sex (at 1-yr follow-up) F 11 yrs i) Observational, consecutive study ii) Decompression without fusion in secondary operation iii) Neurological examination, pain back/leg (4-scale), effect of work, consumption of analgesics, walking capacity In previous operation, 28 lumbar spinal stenosis patients had Repeat decompression of lumbar had decompression (2 also had arthrodesis); 65 nerve roots; a prospective discectomy; diagnosis at the repeat operation were: 2-yr evaluation central LSS (cent) in 20, lateral LSS (lat) in 19, disc F 2 yrs herniation (DH) in 19 and periradicular fibrosis (PRF) in 35 patients Limitation: Diagnoses are based partially on operative findings, being thus peroperative predictors Diagnosis of reoperation Single root affection (due to DH/lat) vs. central LSS Age, gender, type of work Significant better results in patients, who had prolapse or LSS vs. PRF (P 0.0001) Better results in single root affection (P 0.0006) To find prognostic factors which could be applied for second surgical procedure of spine; to report epidemiology of recurrent residual pain in study population (Table continues)