Effectiveness and economic evaluation of chiropractic care for the treatment of low back pain: a systematic review protocol.

BACKGROUND
Chiropractic care is a common treatment for low back pain (LBP). Previous studies have failed to clarify the relative cost-effectiveness of chiropractic care in comparison with other commonly used approaches because previous attempts to synthetize the economic literature has only included partial economic evaluations. The objective of this project is to estimate the clinical effectiveness and cost-effectiveness of chiropractic care compared to other commonly used care approaches among adult patients with non-specific LBP.


METHODS/DESIGN
Two systematic reviews will be conducted to identify 1) randomized controlled trials and 2) full economic evaluations of chiropractic care for low back pain compared to standard care provided by other healthcare providers. We will conduct searches in specialized electronic databases for randomized controlled trials and full economic evaluations published between 1990 and 2014 using a combination of keywords and MeSH terms. This will be supplemented by a search of the gray literature. Citations, abstracts, and relevant papers will be screened for eligibility by two reviewers independently. Studies will be critically appraised using 1) the Cochrane risk of bias tool and 2) the Drummond (BMJ) checklist. Results will be summarized using Slavin's qualitative best-evidence synthesis approach. Data relating to the primary outcomes of the effectiveness study will be evaluated for inclusion in meta-analyses. The costs will be standardized to the same currency (USD) and adjusted to the same year for inflation. The incremental cost-effectiveness, incremental net benefit, and relevant confidant intervals will be recalculated in order to facilitate comparison between studies.


DISCUSSION
Our review will evaluate both the clinical effectiveness and the cost-effectiveness associated with chiropractic care for LBP. A more precise estimate of the cost-effectiveness of chiropractic care for LBP relative to other forms of conservative care is needed for decision-makers and third-party payers to offer best care options for LBP. Our results will facilitate evidence-based management of patients with LBP and identify key areas for future research.


SYSTEMATIC REVIEW REGISTRATION
The protocol is registered on PROSPERO ( CRD42014008746 ).


Background
Low back pain (LBP) remains a leading cause of disability worldwide, accounting for over 10% of the total of 'years lived with disability' [1]. LBP is the most common occupational injury in Canada and United States [2,3]. It is the leading cause of work absenteeism and ranks sixth among health problems in terms of direct medical costs in North America [4].
The incidence of non-specific LBP has not significantly increased in the last four decades [5][6][7]. However, a drastic increase in the number of certificates of illness and benefits paid for chronic disabilities resulting from LBP has been reported in industrialized countries since the 1980s [8]. Such increase in disability level has had an alarming impact on costs due to lost productivity, wage replacement, and health care utilization. According to the 2010 Global Burden of Disease Study, low back pain disability-adjusted life years increased from 58.2 million in 1990 to 83.0 million in 2010 [9]. With the hope of reducing the significant health and economic burden associated with LBP, researchers have examined the effectiveness of numerous treatment options, including manual therapy [10][11][12][13][14].
Opinions vary widely on what causes LBP and how best to manage it [15]. It is estimated that over 85% of patients with LBP have symptoms that are 'non-specific' in nature since they cannot reliably be attributed to a specific disease or anatomical structure [16]. Perhaps as a result, relatively few treatment modalities for the management of LBP have been shown to achieve superior and sustained improvements in pain, physical function, and disability [17,18]. An example is spinal manipulative therapy (SMT), which recent reviews did not find significantly more effective than and other modalities [11,19,20]. SMT is often a core component of chiropractic care [21], but chiropractic care is not restricted to the use of SMT [22], and a range of other treatment modalities may be offered exclusively or in combination with SMT to potentially compliment or enhance treatment outcome. Moreover, SMT is not performed exclusively by chiropractors but is used extensively worldwide by a range of other health care professionals [11,20]. In the case of SMT, studies on the effectiveness may guide clinicians (chiropractors and others) in their choice of treatment modality, but offer little information to patients, policy makers, and third-party payers about the clinical effectiveness and cost-effectiveness of standard care offered by different providers.
Considering this, older reports [23,24] have pointed to chiropractic care as efficient for the treatment of LBP because of the relatively low fee for service, the use of 'low tech' therapies such as manual therapy, and the low usage of costly investigations such as advanced diagnostic imaging. However, when compared with medical and physiotherapy care, economic reviews have not been able to support this or provide clear guidance to informed decision-making regarding the many available provider options [25,26]. This divergence in findings may be partly due to the limited number of studies of acceptable methodological quality [25,27] and partly because the previous systematic reviews have only included partial economic evaluation (cost description, cost analysis, and cost-outcome description). The last systematic review of economic and clinical effectiveness studies of chiropractic care was completed nearly a decade ago and provided limited guidance [25]. We are conducting this review with the hope of including more high quality studies.
When evaluating standard care for LBP, decisions to recommend any one option should preferably be based on the clinical effectiveness, the cost-effectiveness, the safety of the approach, and patient preference [11,19,26]. Only full economic evaluation (cost-effectiveness analysis, cost-utility analysis, cost-benefit analysis) of standard care practice can provide adequate information about resource inputs (costs) and outputs (health outcomes) [28] and evaluate whether healthcare resources are being used optimally [29]. In order to better inform patients, policy makers, and third-party payers about the clinical effectiveness and cost-effectiveness of standard chiropractic care for LBP in comparison to usual standard care provided by other health care professionals, an evidence synthesis is indicated.

Objectives
The two main objectives of this review are 1) to estimate the extent to which chiropractic care is effective for adult patients with non-specific low back pain compared to other conservative care approaches (for example, medical care and physiotherapy) and 2) to estimate the cost-effectiveness of chiropractic care for adult patients with non-specific LBP compared with other conservative care approaches.

Eligibility criteria
To be eligible for inclusion, studies must meet the following criteria: 1. The study design is: a randomized controlled trial for the clinical effectiveness studies; a full economic evaluation (including costeffectiveness, cost-utility, cost-benefit analyses, and cost-minimization analysis alongside a clinical trial [30]) for the economic studies. 2. The population under study is composed of adult patients (≥18 years) with non-specific LBP with or without sciatica of any duration. Studies reporting multiple pain locations or spinal pain without separate results for LBP will be excluded. 3. The intervention is chiropractic care. Studies that evaluate chiropractic care as part of a combined, multidisciplinary approach will be excluded unless the chiropractic care part is evaluated separately. Studies that evaluate specific treatment modalities (for example, SMT) will be excluded. 4. The comparator is non-surgical, usual conservative care delivered by other healthcare providers (for example, medical therapy, physical therapy, or acupuncture). Studies including surgical treatment of LBP as the only comparator will be excluded. 5. The outcome must include -for the clinical effectiveness studies -one or more of the following primary or secondary effect measures:

Primary outcomes
Pain (for example, visual analog scale, numerical rating scale, McGill pain score) Functional status (for example, Roland-Morris questionnaire, Oswestry Disability Index) Global improvement (for example, the number of patients reporting to have recovered)

Secondary outcomes
Health related quality of life (for example, SF-36, EuroQol) Return to work Adverse effects -For the economic studies: an incremental measure of the extra budget required to improve an additional unit of outcome (that is, an incremental cost-effectiveness ratio or an incremental net benefit measure) with the exception of cost-minimization studies.
6. Studies must be published in English or French. 7. Studies without full-text manuscript available (for example, abstracts, conference proceedings, presentations) and duplicate study reports will be excluded. Published study protocol will be registered but not included in the data analysis.

Information sources and search
A comprehensive literature search will be conducted using indexed subject headings and free text related to the topic of interest in electronic health literature databases, as well as gray literature sources (economic evaluations only), to uncover potentially relevant studies.
With the exception of PubMed, the search will be limited to studies published between 1990 and the search date. Since the volume of literature on back pain is impressive, we restricted search to 1990 (start date) as it corresponds with the first potentially relevant studies in this topic. This start date will enable us to select relatively recent literature that is compatible with the contemporary practice of both chiropractic and comparator providers. The PubMed search will be used to retrieve the most recent publications and restricted to items published on or after 2014. Searches will be conducted in the following electronic databases: Ovid Medline, Ovid AMED, Ovid EMBASE, CINAHL, the Cochrane Database of Systematic Reviews, and PubMed. In addition, we will also search for economic evaluations in the following: Index to Chiropractic Literature (ICL), Cochrane Library, Health Technology Assessment Database, and ECONLIT. Finally, a search of the gray literature for economic evaluations will include the websites of the following organization: Canadian Institute for Health Information (CIHI), Canadian Agency for Drugs and Technologies in Health (CADTH), Canadian Institute of Health Research (CIHR), Tufts Medical Center Cost-effectiveness Analysis Registry, Agency for Healthcare Research and Quality, National Institute for Health Research Health Technology Assessment program, and National Institute for Health and Care Excellence (NICE).
The search strategy will be different for each database, and the RCT filters for PubMed, Ovid Medline, AMED, EMBASE, and CINAHL will be adapted from the Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE [31]. The search in Cochrane will be limited to Cochrane Central (Trials) to exclude other study designs. Two clinical librarians (JB and PH), with experience in searching for systematic reviews, developed a search strategy for each individual database and will conduct the searches. The search strategy for clinical effectiveness and for cost-effectiveness can be found in Appendix 1 and 2, respectively. We will screen the bibliographies of relevant publications, including reviews and meta-analyses, for additional relevant articles.

Study selection
Titles and abstracts of studies identified from the literature search will be combined using Endnote 14 and screened for relevance by two independent reviewers to identify all articles that any reviewer judges potentially eligible. The same reviewers will independently apply eligibility criteria to the full-text manuscript of all potentially eligible studies. Disagreements will be discussed until consensus. Disagreements will be resolved with arbitration by a third reviewer if disagreements persist.

Quality assessment and analysis
All eligible studies on clinical effectiveness will be assessed for methodological quality (risk of bias) by two independent reviewers. Studies assessing clinical effectiveness will be evaluated using 12 criteria recommended by the Cochrane Back Review Group [32]. These criteria include blinding of the patient, treatment provider, and outcomes assessor. Studies that meet at least 6 criteria out of 12 will be considered at low risk of bias, while the others will be considered at high risk of bias.
Studies assessing costs will be evaluated using a recommended tool for health economic evaluations, the Drummond (BMJ) checklist [29,33,34]. This checklist includes 35 items grouped into four broad categories: general issues about study design, data collection, data analysis, and interpretation of results. Any disagreements between reviewers will be discussed until consensus is reached or with arbitration by a third reviewer if disagreements persist. The quality level (low, medium, high) of every study will be determined by agreement between three investigators (AB, MAB, MJS). This will enable the investigators to formulate a qualitative appreciation of the complete study.

Data extraction
Data will be extracted separately by two independent reviewers; any disagreements will be resolved through discussion, with arbitration by a third reviewer if necessary. Authors of potentially relevant studies will be contacted regarding additional information or missing data. Key findings from each study will be summarized and presented in a summary tables. Two separate forms will be used for clinical effectiveness studies and economic evaluations. For clinical effectiveness studies, we will use the Cochrane back review group data extraction form [35]. Extracted variables will include author and year; country; participants, indication, setting; compared treatments; time horizon, outcomes assessed; authors' results; and conclusion.
Data from economic evaluations will be extracted using a customized data extraction sheet (Additional file 1). Extracted variables will include author and year; country; type of economic evaluation; participants, indication, setting; compared treatments; perspective; time horizon, currency price (year); included costs, health effect (pain, functional status, global improvement, health-related quality of life, return to work); mean costs, mean qualityadjusted life years (QALYs); incremental cost-effectiveness statistics; limitations; and authors' conclusion.

Measures of effect estimates
Continuous outcomes measured with the same instrument (that is, pain measured with visual analog scale) will be compared using mean difference, whereas continuous outcomes measured with different instruments (that is, functional status measured with Roland-Morris or Oswestry tools) will be compared using standardized mean difference. For dichotomous outcomes (that is, recovery, return-to-work), a risk ratio will be generated.

Data analysis
Effect measures relating to the primary and secondary outcomes of clinical effectiveness studies with low risk of bias, and no serious flaw will be evaluated for inclusion in the meta-analyses. Outcomes will be assessed at 1, 3, and 12 months and will be categorized according to the time closest to these intervals. In order to minimize clinical diversity, we will stratify by healthcare provider (for example, chiropractic care versus medical care or chiropractic care versus physiotherapy), symptom duration (acute (0 to 6 weeks), sub-acute (6 to 12 weeks), chronic (more than 12 weeks), and mixed/not specified), and outcomes (type of outcome and time of assessment). Heterogeneity will be investigated by subjective interpretation and by statistical testing using the Q and I 2 test. A cutoff of 40% at the I 2 test will determine the limit of acceptable heterogeneity. If the I 2 cutoff is exceeded or the description of the average care provided by the comparator seems too heterogeneous, results will be discussed narratively in the manuscript without pooled estimates. A sensitivity analysis will be performed by including studies with high risk of bias. Funnel plots will be constructed using all data from the primary outcomes regardless of the comparator or follow-up interval in order to evaluate possible publication bias.
For the economic evaluations, the difference in perspective of analysis, type of economic analysis, and healthcare system will be discussed narratively. To allow direct comparisons across countries and years, we will convert reported costs estimates to 2014 United States (US) dollars. International exchange rate based on purchasing power parities (PPP) will be use to convert cost estimates to US dollars, and gross domestic product (GDP) deflators will be use to convert cost estimates to 2014. PPP and GDP are available from the World Economic Outlook Database (http://www.imf.org/external/ data.htm). Results comparing chiropractic to other types of care will be summarized using Slavin's [36] qualitative best-evidence synthesis approach, which assumes that the strength of a relationship between variables is based on the quantity and quality of the evidence available. This approach aims to provide methodological rigor by clearly and concisely articulating the synthesis criteria and was recently used in a number of systematic reviews related to occupational health [37][38][39]. The level of evidence uncovered for the findings of interest will be assessed using a 5-point ordinal scale (strong, moderate, limited, mixed, and insufficient evidences) defined by Slavin [36]. The appropriate level of evidence for each finding will be assessed in a stepwise manner by first determining if criteria for the highest level of evidence (that is, strong) are fulfilled and, if they are, no further evaluation is performed. If those criteria are not fulfilled, those for the next lowest level of evidence are then assessed, continuing until the appropriate level of evidence can be assigned to the various review findings. The criteria for each level of evidence are the following:

Strong evidence
Minimum of three high quality studies; at least three quarters of high and medium quality studies must concur on findings.

Moderate evidence
Minimum of two high quality studies or three of medium and high quality; more than two thirds of all studies must report consistent findings.

Limited evidence
Minimum of one high quality study or two medium quality studies, more than 50% of all studies must report consistent findings.

Mixed evidence
Findings from medium and high quality studies are contradictory.

Insufficient/no evidence
No high quality studies; one or no medium quality studies; any number of low quality studies.

Protocol registration
Our protocol is registered on PROSPERO (CRD4201 4008746), http://www.crd.york.ac.uk/PROSPERO. This manuscript conforms to the PRISMA guidelines [40] that are relevant to the reporting of a systematic review protocol. We present our methods and analysis for the review of clinical effectiveness and our review of economic evaluations separately.

Discussion
Our research team includes French and English investigators. The potential of omitting important studies in other languages is considered very small since chiropractic is of English/American origin and is primarily practiced in the anglophone countries.