Abstract
Introduction
Low back pain is among the most frequent causes for medical appointments and increased health cost every year. Treatment for this disorder should be evidence-based through systematic reviews (SR) and meta-analysis (MA). However, methodological mistakes frequently seen in those studies put their credibility into question.
Materials and Methods
A literature search was conducted using established databases and gray literature for systematic reviews only, involving low back pain as population and surgical treatment as intervention. A previous protocol has been registered in the local research committee. SR for cervical or thoracic spine disorders were excluded. Four certified spine surgeons independently extracted data. Study quality was assessed through measurement tools specific for SR (PRISMA and AMSTAR). For both questionnaires, studies were rated as poor, fair, good, very good, and excellent according to their quality percentage (0–30, 30–50, 50–70, 70–90, > 90%, respectively). For each SR, data were extracted for population, intervention, primary outcomes (VAS, Oswestry disability index, SF-36 questionnaire) and secondary outcomes (fusion, complications, return to work). When a meta-analysis was performed, the outcomes were considered as conclusive or inconclusive. Similar studied diseases, interventions, and outcomes were grouped and analyzed independently.
Results
A total of 40 SR of low back pain with at least one surgical treatment were included. After a throughout analysis of the studies, most popular search database was Medline, followed by Cochrane and EMBASE. Majority of the SR were published within the last 10 years (90%). Average number of included studies was 17.7 per systematic review. According to the PRISMA quality tool, 7.5% of the SR were rated as poor, 30% fair, 37.5% good, 20% very good, and 5% excellent. AMSTAR rated 22.5% of the SR as poor, 15% as fair, 30% good, 25% very good, and 7.5% excellent. Most frequent mistakes identified by the PRISMA questionnaire was the lack of a review protocol followed by absent analysis of risk of bias within studies. AMSTAR indicated unreported conflict of interest as major error, followed by the lack of a previously reported protocol. Six analyzed lumbar spine diseases were identified and grouped as degenerative disk disease (DDD) (32.5%), spondylolisthesis (SL) (15%), lumbar stenosis (LS) (12.5%), lumbar disk herniation (LDH) (7.5%), spondylosis (S) (27.5%), and painful facet disease (PFD) (5%). Three SR presenting same population (DDD), intervention (arthroplasty vs. fusion) and outcome (Oswestry disability index) had a positive meta-analysis favoring the total disk replacement technique. Two SR analyzing fusion techniques for spondylolisthesis (posterolateral fusion vs. posterior lumbar interbody fusion) had positive meta-analysis for higher fusion rates with the interbody fusion technique.
Conclusion
Systematic reviews for lumbar spine diseases are still not effective to determine conclusive optimal treatment mostly due to methodological mistakes and lack of scientific evidence.