Quantity of Research Available
The literature search strategy initially identified 377 articles of potential. Following screening titles and available abstracts, 13 full text articles were retrieved. In addition, four articles were identified in the grey literature. Upon review, four guidelines and one RCT met the selection criteria. The 12 excluded articles consisted of two narrative reviews, two commentaries, two studies examining irrelevant interventions, one duplicate of an excluded study, three studies that lacked sufficient methodological quality, one study examining an irrelevant population, and one survey study. A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart describes the selection procedure of the included studies of this review (Appendix 1). No systematic reviews or meta-analyses studies met the inclusion criteria.
Summary of Study Characteristics
Randomized Controlled Trials
Characteristics of the included RCT are tabulated in Appendix 2, Table A2.1.
Study design
The design of RCT consists of three parallel treatment arms, and while examiners were blinded, patients could not be blinded to the type of mattress they were sleeping on.6 A stratifying program was used for random allocation. The aim was to equalize baseline data of the three groups with respect to age, sex, duration and severity of LBP, the number of LBP-related days off work in the past 12 months, and daily physical workload.6
Population
The RCT inclusion criteria was for patients between the ages of 18 and 60 years with largely stable chronic low back pain (LBP) for at least six months. It was also required that the pain had to either dominate in the morning or be equal throughout the day. Additionally, any accompanying leg pain had to be at a constant ratio to the back pain. Patients with other serious illnesses that may impact sleep or patients that were already sleeping on a study mattress were also excluded.6
Intervention and Comparators
The interventions of the trial were described in limited detail. The firm mattress consisted of a foam core, surrounded by three layers of cotton with no springs. The water bed was built with four horizontal layers of fibers which stabilized the movement of water after one second. The foam mattress was made of a temperature-sensitive pressure relieving material that molds to the persons shape after a few seconds. One of these three interventions was used for one month.6
Outcomes
Interviews before and after the intervention were used to obtain integer scores for lower back pain (0–10), leg pain (0–10), and an activities of daily living (ADL) score (0–30). An increased ADL score reflects a decrease in daily function level. The number of hours slept was also obtained by the post-intervention interview.
Guidelines and Recommendations
Four guidelines were identified as meeting the inclusion criteria.7–10 Study characteristics of the included guidelines are summarized in Appendix 2, Table A2.2.
Origin of reports
One included guideline from 2011, was a Canadian guideline and originated in Edmonton, Alberta.8 This guideline was based upon eight previously published guidelines including previous versions of two guidelines included in this report.7,10 Two identified guidelines originated in the United States, one from the Agency for Healthcare Research and Quality (AHRQ), in Rockville, MD, published in 20117 and one from the American Pain Society in Glenview, IL published in 2009.9 The fourth guideline included in this report is from Europe from the Working Group on Guidelines for Prevention in Low Back Pain published in 2006.10
Interventions
All of the identified guidelines included various interventions for prevention and/or treatment of LBP and all include recommendations on mattresses.7–10 The Canadian recommendations,8 the European evidence-based guidelines,10 and the American Pain Society guidelines10 based their recommendations on the same single RCT published in 2003.11 This RCT also contained more quantitative information on the definition of firm and medium-firm mattresses. The firmness of mattress (Hs) was rated according to the European Committee for Standardization scale. The scale starts at 1.0 (firmest) and stops at 10.0 (softest). The firm mattress used in the 2003 RCT was Hs 2.3 while the medium-firm mattress was rated Hs 5.6.11 No identified guidelines contained interventions specific to neck pain.
Grading of recommendations and levels of evidence
The schemes used by the included guidelines for grading recommendations and levels of evidence are summarized in Appendix 3. Two included guidelines graded recommendations and assigned a level of evidence to the identified literature.7,8 Guidelines from the AHRQ either Strongly Recommend or Strongly Not Recommend based upon an evidence level A, Moderately Recommend or Moderately Not Recommend based upon an evidence level B, Recommend or Not Recommend based upon an evidence level C and either Recommend, Not Recommend or have No Recommendation for evidence level I depending on consensus of the guideline development group (GDG).7 The identified Canadian guidelines were based upon previously published guidelines and graded the previously published recommendations as Do, Do Not Do, or Do Not Know. Evidence levels were graded based upon the highest quality study identified either systematic review (SR), randomized controlled trial (RCT), case series (CS), guideline (G), or expert opinion (EO).8 The American Pain Society did not have graded recommendations and rated evidence as good, fair or poor.9 The European evidence-based guidelines, in contrast, did not have levels of evidence but graded recommendations as Level A-D based upon the quality and quantity of the supporting literature.10
Summary of Critical Appraisal
Critical appraisal of the included RCT is tabulated Appendix 4.
The identified RCT6 provided a CONSORT diagram, which documented the 40 percent dropout of participants. Patient characteristics were tabulated before and after randomization and after intervention. The dropouts did not create significant differences between groups in respect to any of the tabulated patient characteristics. Before intervention initiation 68% of the dropouts were from the patients allocated to the waterbed treatment while during the intervention 67% of the dropouts were from the patients allocated to the firm mattress. This uneven distribution of dropouts may be a source of study bias. The study clearly defined the role of blinded investigators, the allocation process, patient eligibility, outcome measures, and the statistical methods used. The interventions were described but not sufficiently for precise replication of the study. The study did not include a discussion of possible adverse events or limitations of the trial. The authors analyzed the data for two possible scenarios, one where the patients that dropped out of the study were perfectly representative of their assigned group and one where the patients that dropped out of the study were assigned the worst 90th percentile scores. The second analysis was used because patients that removed themselves from the study did so because of more pain or less sleep. An analysis that did not include dropouts was not reported. The self-reported, subjective outcome measures of back and leg pain, combined with the inability to blind study participants to the intervention, introduces the potential for bias in this study. The mattress interventions were supplied and installed in the participant’s home by mattress industry sponsors.
Critical appraisal of the included guidelines is tabulated in Appendix 5.
Three of the four included guidelines detailed a literature search methodology.7,9,10 Two of these three described a systematic literature search including selection criteria.7,9 The remaining set of guidelines were based upon previously published guidelines and did not describe a literature search methodology or provide details for a supplementary literature search that was mentioned.8 Two guidelines described attempts to improve stakeholder representation in their respective GDGs.7,10 The included guidelines varied in stating explicitly the scope,7–9 objectives8,10 and target audience.7,8,10 One guideline failed to include a conflict of interest statement.8 The bibliography of the AHRQ guidelines were contained in a separate document.7 While these guidelines do provide methods for recommendation formulation, there is no explicit link between the recommendations and the evidence used to formulate them.7
Summary of Findings
The statistically significant findings and author’s conclusions of Bergholdt et al., 20086 are summarized in Appendix 6.
This RCT compared a waterbed mattress, a foam mattress, and a firm mattress for the treatment of chronic LBP. The study did not find any statistically significant differences between the water bed and foam mattress in any outcomes. When patients who discontinued intervention were assigned the worst 90th percentile scores the firm mattress produced a statistically significant increase in LBP scores, leg pain scores, an increase in ADL, and a decrease in reported sleep hours compared to the water bed and foam mattress. When patients who discontinued the intervention were considered to have no effect, only a difference in LBP scores and sleep hours remained statistically significant. No analysis was done without including patients who discontinued the intervention. The authors conclude that the hard mattress resulted in worse outcomes but that the differences were small.
Relevant recommendations of the included guidelines are summarized in Appendix 7. The levels of evidence and grades of recommendations used below are described in Appendix 3.
Guidelines from the AHRQ included separate recommendations for acute, subacute and chronic LBP with interventions of mattresses, specific beds and use of optimal sleeping surfaces (e.g. bedding, water beds, and hammocks). All of the recommendations were based upon Evidence Level I (consensus), and were either No Recommendation or Not Recommended. The use of specific beds was Not Recommended for treatment of any category of LBP, while mattresses and the use of optimal sleeping surfaces had No Recommendation for any category of LBP.7
The Canadian guidelines,8 used the European evidence-based guidelines10 as a reference for the included mattress related recommendations. The Canadian guidelines had a Do Not Know recommendation for any specific type of mattress,8 while the European evidence-based guidelines stated no recommendation for LBP prevention using mattress interventions [Level C] but suggested that chronic LBP may be reduced with a medium-firm rather than a hard mattress [Level C].10 For chronic LBP the American Pain Society guidelines cited one RCT as evidence that a firm mattress is slightly inferior to a medium-firm mattress for pain-related disability and pain while in bed with no other pain related outcome differences [fair].9 The Canadian recommendations,8 the European evidence-based guidelines,10 and the American Pain Society guidelines10 based their recommendations on the same single RCT published in 2003.11 None of these guidelines cited the RCT included in this report, Bergholdt et al. (2008),6 in recommendation formulation.7–10
Limitations
There was a lack of evidence identified supporting the effectiveness of different mattress types for chronic back and neck pain. The absence of identified evidence may be a function of the limitations of the search strategy, however systematic literature searches of the included guidelines also identified a lack of evidence. The identified RCT was limited by the types of interventions, the subjective nature of the self-reported outcomes and the inability to blind study participants to the intervention. No evidence regarding mattress interventions for chronic neck pain was identified.