|Home | About | Journals | Submit | Contact Us | Français|
A structured literature review.
Widely recognized classification criteria for rheumatologic disorders have resulted in well-defined patient populations for clinical investigation.
We sought to determine whether similar criteria were needed for back pain disorders by examining variability in eligibility criteria in published studies
Studies involving radiculopathy due to lumbar herniated disc (HD) and for neurogenic claudication due to lumbar spinal stenosis (LSS) were identified. Randomized controlled trials published between January 1, 2006 and October 1, 2008 in select peer reviewed journals were retrieved, their eligibility criteria were identified and categorized.
Twelve eligible HD studies were identified. Thirteen unique categories of eligibility criteria were identified with a mean of 3.9 (+/−2.0) and a range from 0 to 8 categories per study. More categories were present for studies that included nonsurgical (5.6 +/− 2.5) treatment for studies with only surgical treatment (2.6 +/− 1.7) p= 0.04). Seven LSS studies met eligibility criteria, and 9 unique categories were identified. A mean of 5.0 (+/−2.2) categories with a range from 2 to 7 was used per study.
Wide variation in the number and type of eligibility criteria from randomized clinical trials of well defined back pain syndromes was identified. These results support the need for developing and disseminating international classification criteria for these clinical conditions.
The last two decades have witnessed an explosion of controlled trials involving a wide range of low back pain conditions. Prior to this, there were few high quality randomized controlled trials of common back pain treatments. Well conducted meta-analyses of low back pain treatments routinely concluded that reliable data were too sparse to draw firm conclusions 1-4.
Immense improvements in the quality and quantity of research have taken place in recent years. Major randomized controlled trials have been conducted and published in highly rated peer reviewed journals 5-7. It is tempting to imagine that high quality meta-analyses with well grounded, clear conclusions for many low back conditions and treatments are just a few years off.
In contrast with many rheumatological conditions, such as rheumatoid arthritis 8 or complex regional pain syndrome 9, the literature on low back pain disorders has developed without widely recognized diagnostic and classification criteria. The development of internationally recognized classification criteria for rheumatological disorders has contributed to their pathophysiologic understanding and the development of new therapeutic targets 10. Classification criteria are also critical for conducting clinical trials, epidemiologic studies and other clinical investigations. The use of established classification criteria ensures recruitment of a more clinically homogeneous sample of subjects 10. Imprecise definitions of specific spinal conditions may have clinical consequences as physicians attempt to apply the results from trials with potentially different patient populations to their own patients. To document and address the potential impact of different case definitions, we examined variation in eligibility criteria in recently published studies for two seemingly well defined back pain syndromes, radiculopathy arising from a lumbar herniated disc (HD) and neurogenic claudication due to lumbar spinal stenosis (LSS). Our goals were to document variability in classification of the same entities across studies and to identify possible classification criteria for future evaluation.
We performed a structured literature review using PubMed to identify potential clinical studies of radiculopathy due to a lumbar herniated disc and neurogenic claudication due to lumbar spinal stenosis. Two independent literature searches were conducted by one author (SG). Search terms on radiculopathy due to HD included: sciatica, lumbar radiculopathy, and lumbar disc herniation. Search term on neurogenic claudication due to LSS included: degenerative lumbar stenosis, lumbar spinal stenosis, neurogenic claudication and intermittent claudication. We focused exclusively on randomized controlled trials comparing two treatments because strict eligibility criteria are particularly important for such studies. Studies that included mixed pathological conditions were excluded. For both conditions, additional search criteria included: 1) age 19 or older and 2) English literature. As methodology in back pain literature has been said to have recently improved, the search was also restricted to manuscript published between January 1, 2006 and October 1 2008. We augmented the electronic search by including studies from the reference lists of retrieved papers. To specifically identify high quality clinical trials, the literature search was limited to publications from journals with an impact factor of at least 1. To avoid duplication we excluded secondary analyses of previously published studies if the parent study met eligibility criteria for this analysis.
Data abstraction from eligible studies involved reviewing the Methods sections to identify inclusion and exclusion criteria. Eligibility criteria were then extracted and classified into one of the following four groups: a) demographic characteristics; b) symptoms; c) physical examination signs; d) and investigative procedures. Categories were identified and classified within each of these four broad groups. Eligibility criteria that were not directly related to the diagnosis of the underlying clinical condition but rather related to the design of the study (e.g. symptom duration, failure of previous treatments, history of previous surgery, etc.), or to the treatment provided (e.g. allergies to studied medication, specific contraindications to the procedure, etc.) were excluded. The entire selection process was reviewed with a second author (JK). Discrepancies and ambiguities were discussed and consensus was achieved.
Eligible studies were classified into 3 treatment categories: surgical treatment only, surgical versus nonsurgical treatment, and nonsurgical treatment only (Table (Table11 and and2).2). For each clinical condition, the number of studies (mean, median, and range) reporting each identified criteria were reported. Reported criteria across studies were also compared. Unpaired t-tests for unequal variance were used to assess differences between groups.
The literature search on radiculopathy due to HD retrieved 56 references. One additional study was identified by checking reference lists of the retrieved studies. After reviewing the abstracts of these 57 studies, 43 were excluded because they did not meet inclusion criteria. Two more were discarded after reviewing the entire article. Discrepancies between SG and JK arose on 3 occasions and were resolved by reanalyzing the Methods section of the specific study. The Methods section of each of the remaining 12 studies underwent review to identify eligibility criteria.
Thirteen unique categories were identified that accounted for all identified diagnostic related criteria (Table 1). No category was assigned to the demographic characteristic group. Five related to clinical symptoms: 1) presence of accompanying back pain; 2) leg pain greater than back pain; 3) characteristic of leg pain (radicular/sciatic/dermatome distribution); 4) pattern of pain radiation (anterior thigh, below the knee) and 5) pain on Valsalva manoeuvre. Three categories referred to clinical examination: 1) nerve root irritation sign (straight leg raise test [L5 or S1] or femoral stretch test [L3 or L4], contralateral straight leg raise test); 2) presence of neurological deficit; and 3) neurological deficit corresponding to pain pattern. Four categories related to investigations: 1) herniated disc on imaging; 2) herniated disc on imaging corresponding (level + side) with clinical observation; 3) specification of the kind of HD included/excluded (protrusion, extrusion, sequestration, foraminal, etc.) and 4) electromyographic findings. The final category, specification of nerve roots included, could not be specifically attributed to a specific group.
The mean number of diagnostic categories identified per study was 3.9 (standard deviation 2.0, median 4, range 0 to 8). Trials comparing surgery and nonsurgical treatments used more eligibility categories (5.6 +/− 2.5) than trials only comparing surgical techniques (2.8 +/− 1.7). Trials only comparing nonsurgical treatments had a mean of 4.3 (+/− 1.9) categories cited per article. Studies comparing only surgical techniques used fewer criteria than studies involving at least one nonsurgical treatment (2.8 +/−1.7 vs. 4.8 +/− 1.9, p=0.04).
No single diagnostic eligibility category was used in all the studies (Table 3). The most commonly used categories were “radicular pattern of radiation” and “HD on imaging” (each 13/19, 68%). Congruence between pain pattern of radiation and HD on imaging were less commonly used (each 6/19, 32%), while congruence between pain pattern and neurological deficit were each used only twice (11%). There was additional variation in the definition used within a category. For example when the category “specification of nerve roots included” was used the number of defined nerve roots ranged from 1 11 (S1) to 6 12 (from L1 to S1). The definition of the straight leg raise test varied in 2 ways: a) the kind of triggered pain (i.e. any leg pain versus radicular pain) and b) the degree of minimal/maximal angle for the test to be positive.
The search for studies of neurogenic claudication due to LSS retrieved 20 citations. Eleven were excluded after reviewing the abstract, and an additional 2 were excluded after reviewing the entire paper. The remaining 7 eligible studies underwent review to identify eligibility criteria.
Nine unique categories were identified that accounted for all identified diagnostic related criteria (Table 2). One category was demographic: age limitation (usually older than 50). Three categories related to clinical symptoms: 1) presence of accompanying back/buttock/lumbopelvic pain; 2) characteristics of radiating pain (neurogenic claudication / radicular pain); 3) definition of neurogenic claudication (symptoms restricted to pain or include fatigue and sensory loss: factors that made it worse; factors that ease the symptoms). No category was assigned to the clinical examination group. Four categories related to investigations: 1) radiological confirmation of stenosis; 2) explicit description of radiological stenosis (e.g. minimal sagittal diameter); 3) correlation between symptoms and level of radiological stenosis; 4) specific inclusion or exclusion of some other radiological findings (e.g. degenerative spondylolisthesis, instability). The final category could not be specifically attributed to a specific group: exclusion of other spinal (congenital stenosis) or non spinal problems (e.g. vascular, articular or neurogenic).
The mean number of criteria per study was 5.0 (+/−2.2), (median 6, range 2 to 7). No differences were noticed between surgical, nonsurgical and mixed studies (means between 5 and 5.5). On average more categories were used in trials of neurogenic claudication (mean of 5 crtieria used per paper of the 9 possible criteria; 57%) compared to trials on radiculopathy (mean of 3.9 criteria of a possible 13; 28%), p 0.02.
The presence of imaging findings consistent with lumbar stenosis was used in all studies, while the presence of neurogenic claudication was used in 5/7 (71%) studies. The definition of neurogenic claudication varied considerably across studies (e.g. “fatigue or loss of sensation in the lower limbs aggravated by walking” 13 and “sitting as a better position for symptom severity than standing or walking” 14.) Some studies included pain located in the buttock area as a possible radiation pain likely to have claudicating pattern 13,14. This allows the inclusion of patients who would be ineligible in studies adhering to the classical definition of lumbar pain (i.e. pain located between the 12th costal margin and the buttock folds 15). Detailed radiological definition of LSS was provided in 4/7 (57%) studies. In addition, the definition of radiological LSS varied among studies. Some mentioned dural sac compression 16, while others referred to precise measurement of the lumbar canal 17. Congruence between radiating pain pattern and the radiological level of stenosis was mentioned once. Of note, some radiological findings such as degenerative spondylolisthesis are mentioned in the list of exclusion criteria 18 in some studies while they are specifically mentioned as being included in others 13.
This study documented a high level of variation in diagnostic eligibility criteria for randomized controlled trials involving treatments for radiculopathy due to HD and for lumbar stenosis due to LSS. There was wide variation in the number of criteria used within each study and few criteria were used consistently across studies. These findings support our hypothesis that even for spine conditions that are thought to be well characterized, there is no single, standardized case definition used in the research literature. In addition, differences were found within defined categories, such as the number of nerve roots included or the definition used for the straight leg raise test in the case of radiculopathy or the definition of radiological stenosis for neurogenic claudication.
Some of the categories that were created from our analysis such as “pattern of radiation” (e.g. anterior thigh, below the knee) and “specification of the nerve root included” may appear similar; the precision added by the second proposition was however considered to be crucial, justifying the separation, especially in reference to the L4 nerve root. Pain from the L4 nerve root may radiate below the knee, but the clinical assessment of L4 nerve root requires quadriceps testing for deficit and the femoral stretch test, rather than the straight leg raise test. Some discrepancies were noted in studies where eligibility criteria included pain below the knee and clinical examination related to L5 and S1 nerve root without mentioning the appropriate neurological testing for L4 nerve root 19.
In the absence of widely accepted diagnostic eligibility criteria, each group of authors devised their own construct. This makes generalizing findings across studies and to routine clinical practice a challenge. At a time when other musculoskeletal diseases are considering revision of well established sets of criteria 20,21, the absence of diagnostic and/or classification criteria in the field of low back pain should be considered a major focus for international organizations and clinical investigators. We suggest this process should differentiate among several needs. For epidemiological studies, broad criteria are needed and we suggest they should exclusively rely on symptoms that can be elicited from patients as part of survey research. For clinical studies however, the focus should be put on achieving more homogenous populations. For studies in the primary care setting involving patients with acute symptoms, the addition of clinical findings would be appropriate. Finally, for most clinical trials involving treatments, confirmatory investigative findings from imaging studies or electrodiagnostic tests are needed.
A number of important limitations may limit the interpretation of our results. A broader literature search criteria would have likely retrieved more diagnostic eligibility criteria. However it would also have included a large proportion of studies without any eligibility criteria 22. We believe that finding wide variation in diagnostic eligibility criteria in recent randomized controlled trials published in high quality journals (defined as journals with higher impact factor) strengthens our conclusions. We also could have examined more spine conditions to generalize our findings across a broader range of disorders. However, focusing on two common and presumably well defined conditions, radiculopathy due to a herniated disc and neurogenic claudication due to lumbar spinal stenosis, supports the need for similar efforts for non-specific low back pain disorders as well. Some discrepancies were noted between the two reviewers in term of article selection (3 times) and in term of number of eligibility criteria (5 times). The vast majority were related to imprecision within the abstract. These discrepancies were generally resolved after considering the full text of the article. However, in some cases, imprecise or inconsistent reporting in the Methods section required subjective decisions by the reviewers. Excluding these few criteria would not alter our general findings.
In conclusion, we identified wide variation in diagnostic eligibility criteria for studies of radiculopathy due to HD and for neurogenic claudication due to LSS. These findings support efforts to convene a multidisciplinary, international effort to propose validated classification criteria for these conditions.
Pr J.N. Katz is support by NIH/NIAMS P60 AR 47782 and NIH/NIAMS K24 02123. Dr. Atlas is supported in part by a grant from (P60 AR048094) to the Multidisciplinary Clinical Research Center in Musculoskeletal Diseases from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS).
The authors have no conflict of interest.