Description of the condition
In 1986, the American College of Rheumatology defined osteoarthritis (OA) as “a heterogeneous group of conditions that lead to joint symptoms and signs, which are associated with defective integrity of articular cartilage, in addition to related changes in the underlying bone at the joint margins" (Altman 1986). OA is by far the most prevalent form of all joint disorders and a major cause of pain and functional disability that affected 151 million individuals globally in 2000 (Symmons 2006), and the number is expected to double by the year 2020 (Lawrence 2004). The prevalence of symptomatic OA (pain on most days of a recent month and radiographic evidence of OA in the affected joint) is approximately 13% in adults aged ≥ 60 years (Dunlop 2001; Lawrence 2004), 6% of all adults aged 30 years for the knee (Longo 2011), 4% in those aged between 55 and 74 years for the hip (Lawrence 2004), and 10% of people aged > 70 years for the hand (Zhang 2002). Because OA is a chronic, progressively degenerative disease of synovial joints lacking disease-modifying medication, operative interventions, primarily in the form of joint arthroplasty have become increasingly popular but also posed a heavy economic and societal burden. By 2030, the demand for primary total hip and knee arthroplasties is estimated to grow by 174% to 572,000 and by 673% to 3.48 million procedures, respectively, in US alone (Kurtz 2007).
The key and most remarkable characteristic of OA is the destroyed integrity of articular cartilage, its abnormalities, however, are not as limited. Furthermore, cartilage itself is both avascular and aneural, OA symptoms, such as pain and stiffness, logically could not be directly generated by it. Therefore, there must exist other important features that are responsible for the pain of OA such as alterations in subchondral bone architecture (increased volume, trabecular thickening and separation, sclerosis), joint margin osteophytes outgrowth on radiograph, and changes in the juxta-articular bone marrow known as bone marrow lesions on magnetic resonance imaging (MRI), all of which have been shown to correlate with the presence of increased bone mineral density (Felson 2007; Li 1997; Lo 2005; Messent 2005; Spector 2003). Increased bone mineral density could produce a stress shielding effect, leading to localised weakening of trabecular micro-structure and enhanced bone turnover (Tissakht 1996). In other words, bone is an ignored important player in the development process of OA. Changes in different joint tissues are not independent with each other. Indeed, not only are they spatially linked, they take place simultaneously and in concert. The correlation between degenerations of cartilage and alterations in subchondral bone as well as bone marrow are suggested and supported by a number of studies (Felson 2003; Guermazi 2003; Muraoka 2007), and provides a possibility that drugs affecting bone metabolism may have benefits in altering the progression of OA. The nature of their exact relationship, however, remained to be elucidated.
Although the aetiology of OA remains elusive, multiple risk factors have been found to be associated with osteoarthritis in epidemiology studies, with biomechanics playing a central role and a complex interplay of other important factors such as age, female sex, previous injury, genetic predisposition, metabolic elements, and so on (Creamer 1997).
The association between symptoms and radiographic examinations of OA is weak (Conaghan 2011). OA is most often diagnosed by an overall clinical impression coming from the patient's age, history, physical examination findings, and radiographic presentation. The most frequently affected joints are weight-bearing joints such as knees and hips, as well as the cervical and lumbosacral spine, and hands. By contrast, the wrist, elbow, and ankle are usually spared (Longo 2011). In recent years, no longer misconceived as a uniform disease which affects only joint cartilage, the theory that OA is a disease of the whole synovial joint has already become widely accepted, meaning that OA may represent failure of the joint as an organ, analogous to renal or cardiac failure (Arden 2006), resulting from an imbalance in the dynamic equilibrium between the breakdown and repair of joint tissues. This new understanding and other discoveries relating to its pathophysiology have driven the division of OA into distinguishable phenotypes, comprising posttraumatic, ageingrelated, genetic, painrelated and metabolic forms (Zhuo 2012).
Although chondrocytes are probably the most important cells involved in the development of osteoarthritis (Longo 2011), the process, generally initiated either by damage to normal articular cartilage or by defective cartilage failure under normal joint loading, is an active interaction between degradation and repair processes of cartilage, bone, and synovium. Under either circumstance, an essentially similar pathway follows, involving a multitude of factors such as various proteolytic enzymes, cytokines and growth factors secreted by chondrocytes as well as synovium (Zhuo 2012), finally leading to the degradation of collagen and proteoglycans (which comprise over 90% of the cartilage macromolecules), synovial inflammation, bone remodeling, and muscle weakness.
Description of the intervention
Similar to many musculoskeletal disorders, the management of osteoarthritis includes physical measures (physical therapy and exercise programs) (Bartels 2007; Fransen 2008; Fransen 2009, Verhagen 2007), medical treatment (acetaminophen, non-steroidal anti-inflammatory drugs, opioids, glucosamine and chondroitin, intra-articular steroid or hyaluronate injections) (Bellamy 2006a; Bellamy 2006b; Cepeda 2006; Nüesch 2009; Towheed 2006; Towheed 2009), and surgical therapies (osteotomy, arthroscopic surgery, and total joint replacement) (Brouwer 2007; Jacobs 2001; Jacobs 2005; Jolles 2006; Laupattarakasem 2008; Nakama 2012; Reichenbach 2010; Zhang 2010). However, unlike diseases such as osteoporosis and rheumatoid arthritis, where dramatic success had been achieved in preservation of function and structure as well as symptom relief by means of pharmacological interventions, current mainstream medical treatment for OA is in essence symptomatically palliative with only modest efficacy and yet frequent or serious adverse effects, especially with long-term use (ACR 2008; Hochberg 2012; Hunter 2011; Zhang 2008).
Inspired by disorders such as rheumatoid arthritis, for which the development of the disease-modifying anti-rheumatic drugs has successfully altered the disease course and clinical management, an array of new etiologically targeted agents, collectively named as disease-modifying anti-OA drugs after their potential via a variety of mechanisms and targets including subchondral bone, cartilage and synovium, to slow, halt, or even revert the disease progression as well as to provide symptomatic benefit, have been emerging, and this may well be the preferable approach that is in accordance with the concept of OA as a disease entity involving the whole joint (Hunter 2011; Verbruggen 2006).To link the whole joint tissue to the pathophysiology of OA offers the possibility of introducing new etiologically targeted pharmacological interventions to the treatment regimen. Clinical studies both in animals and humans have suggested that a wide array of such agents are promising candidates of possessing the ability to modify OA progression as well as provide concomitant improvement in OA symptoms (Hunter 2011). Some of these candidate drugs include those targeting cartilage (inhibitors of matrix metalloproteinases (MMPs), inducible nitric oxide synthase, and cathepsin K, glucosamine, chondroitin, diacerein, avocado soybean unsaponifiables, growth factors, doxycycline, vitamin D and vitamin E), systemic agents inhibiting inflammatory pathways (licofelone, cytokines, antibodies of tumour necrosis factor, steroids), and disease-modifying anti-OA drugs targeting subchondral bone (calcitonin, bisphosphonates, and strontium ranelate) (Bondeson 2011; Fajardo 2005; Hunter 2011; Zhuo 2012).
How the intervention might work
Bisphosphonates are a class of drugs commonly used for the prevention and treatment of osteoporosis (Wells 2008a; Wells 2008b; Wells 2008c) as well as to treat similar bone diseases that feature bone fragility, such as Paget's disease, bone metastasis, multiple myeloma, osteogenesis imperfecta, hypercalcaemia, and so on (FDA 2011).
Bisphosphonates are named based on the fact that they have two phosphonates, and their structural similarity to pyrophosphate, a byproduct of cellular metabolism and the mechanism by which bone mineralization is regulated (Drake 2008). They work by directly impairing osteoclast activity, and thus decrease physiological bone resorption, via mechanisms of either increasing cellular apoptosis or influencing their metabolic pathway (Fleisch 1998). In the meantime, bisphosphonates also appear to have a positive effect on osteoblasts (Plotkin 2008). Bisphosphonates are, however, differential both in structure, in effect, and in mechanism of action (Rogers 2004). Early developed bisphosphonates do not contain nitrogen group (etidronate, clodronate, and tiludronate), while the second- and third-generation ones are nitrogen-containing (alendronate, neridronate, ibandronate, pamidronate, risedronate, and zoledronic acid,). The nitrogen-containing bisphosphonates work primarily by inhibiting activation of enzymes that utilize pyrophosphate, such as farnesyl pyrophosphate synthase, a key regulatory enzyme and determinant in the regulation of core osteoclast cellular activities and anti-resorptive capacity, and are more potent than the simple not-containing-nitrogen bisphosphonates, which induce apoptosis of osteoclasts by disrupting the use of adenosine triphosphate (ATP) as the source of energy (Beek 2003; Kimmel 2007).
The primary reason that bisphosphonates hold the OA therapeutic potential is their established pharmacology of altering bone remodeling through a direct inhibitory effect on the osteoclast activity (function and number), occurring at the subchondral bone level of affected joints, where altered bone mineral quality and density as well as bone turnover similar to that of osteoporosis was observed (Spector 2003). This is further demonstrated by a clinical study (Bingham 2006) in which patients with progressive knee OA were associated with higher urinary levels of C-terminal telopeptide of type II collagen (CTX-II), a marker of cartilage degradation, as well as by a number of animal studies, in which the classic spontaneous model of Duncan-Hartley guinea pig (Bendele 2005; Meyer 2001a; Meyer 2001b) and other models were used to investigate the anti-resorptive effects of bisphosphonates on OA (Hayami 2004; Muehleman 2002), and displayed that when measured by size and severity of cartilage damage and osteophyte outgrowth, risedronate, a bisphosphonate, slowed OA progression by as much as 30 to 40%. In addition, it was demonstrated that bisphosphonates may exert immuno-modulation, and thus, inhibition of pro-inflammatory cytokines, which is also a major participant in the osteoarthritic process (Frith 2001). Data suggested that inflammation may also play an important role in the pain of OA (Felson 2003; Haynes 2002), and both knee effusions and synovial thickening are more common in patients with knee pain than those without knee pain (Hill 2001). Bisphosphonates may thus relieve knee pain by way of antiinflammatory effects (Frith 2001).
Why it is important to do this review
OA is the most common and prevalent form of arthritis and a leading cause of disability; however, a remarkable gap exists with respect to desirable pharmacologic interventions, considering that current pharmacological treatment options have been burdened with issues such as low to moderate efficacy, long term safety concerns, and inability to alter the course of the disease.
The field of disease-modifying osteoarthritis drugs is rapidly evolving, in which bisphosphonates is one important class of candidates. Cochrane reviews have already analysed some possible agents such as diacerein (Fidelix 2009), doxycycline (da Costa 2012), glucosamine (Towheed 2009), and chondroitin in the treatment of OA (Singh 2006). Based on the fact that there have been several published and on-going prospective, double-blind randomised controlled trials (RCTs) involving bisphosphonates, we believe it is clinically relevant to systematically review and discuss the current evidence of effects of bisphosphonates in the treatment of OA.
To assess the benefits and harms of bisphosphonates for the pharmacological treatment of the hip, knee and hand OA. Both symptomatic and structural outcomes will be evaluated.
Criteria for considering studies for this review
Types of studies
RCTs and quasi-RCTs (method of allocating participants to a treatment which is not strictly random: e.g. by date of birth, hospital record number, alternation) evaluating bisphosphonates for treating the hip, knee and hand OA.
Types of participants
Adult patients aged 45 years and above, of either sex, with clinically and/or radiologically confirmed primary OA of the knee, hip, or hand. Studies testing conditions such as secondary OA, and inflammatory ones such as rheumatoid arthritis will be excluded. If mixed population are tested, we will include them if the subgroup data for primary OA are available.
Types of interventions
RCTs, or quasi-RCTs comparing bisphosphonates with placebo, no intervention and active comparator. There will be no restrictions on route of administration or dosage. We will include all durations of treatment.
Types of outcome measures
- Pain with a hierarchy of 11 levels
- Pain overall
- Pain on walking
- Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale
- Pain on activities other than walking
- WOMAC global scale
- Lequesne osteoarthritis index global score
- Other algofunctional scale
- Patient's global assessment
- Physician's global assessment
- Other outcome
- No continuous outcome reported
When more than one is reported, we will take the highest on the list e.g. choose pain overall over pain on walking.
- Physical function with a hierarchy of eight levels
- Global disability score
- Walking disability
- WOMAC disability subscore
- Composite disability scores other than WOMAC
- Disability other than walking
- WOMAC global scale
- Lequesne osteoarthritis index global score
- Other algofunctional scale
When more than one is reported, take the highest on the list e.g. choose global disability score over walking disability.
- Radiographic joint structure changes according to the given hierarchy
- Minimum joint space width
- Median joint space width
- Semi-quantitative measurement
- Quality of Life
- Number of patients experiencing any adverse event
- Patients who withdrew because of adverse events
- Patients experiencing any serious adverse events
- Percentage of "progressors" (i.e. number of severe joint space narrowing, which is most often defined as a narrowing of greater than 0.5mm (Ravaud 1998)）
- Joint arthroplasty
- Bone marrow lesions
If a study provided more than one pain or function scale, we will refer to a previously described hierarchy list and extract those highest on it (e.g. global Visual Analog Scale (VAS) pain takes precedence over WOMAC osteoarthritis index pain subscale, and WOMAC disability subscale over Lequesne osteoarthritis index score etc) (McAlindon 2000).
In order to map safety profile of each drug, we intend to present specific adverse events for each disease modifying anti-OA drugs candidate included; meanwhile, whenever possible, the following three outcomes will be analysed as well: patients who withdrew because of adverse events, patients experiencing serious adverse events, and patients experiencing any adverse event.
We will extract data at the end of the trial. If multiple time-points should be provided, we will extract data based on short-, intermediate-, and long-term follow-up, which are pre-defined as ≤ 6 months, 7 to 24 months, and ≥ 24 months. We expect different clinical effects in the short versus long term. In the short term, pain-related outcomes may be more obvious, whereas structural changes may only be detected in the long term using X-ray method. However, with the use of more sensitive MRI, structural changes may be quantitated at an earlier time, and data concerning MRI outcomes may be combined with joint space width (JSW) on radiograph using the standardised mean difference (SMD).
Search methods for identification of studies
We will search the Cochrane Central Register of Controlled Trials (The Cochrane Library, current issue), MEDLINE (1946 to present), and EMBASE (1980 to present). No language or publication restrictions will be applied.
We will combine the MEDLINE (subject-specific strategy with the "Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE: sensitivity-maximizing version (2008 revision); Ovid format" as illustrated in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011). We will modify the MEDLINE search strategy (Appendix 1) for use in other databases.
In order to identify recently completed, unpublished and ongoing trials, we will search Current Controlled Trials (http://www.controlled-trials.com/) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/).
Searching other resources
We will check the reference lists of published studies to identify additional trials. Where necessary, we will contact authors of identified studies and companies who produce relevant products for additional information. Furthermore, we ill screen the websites of regulatory authorities such as the U.S. Food and Drug Administration (FDA) (http://www.fda.gov/) and the European Medicines Agency (EMA) (http://www.ema.europa.eu/ema/) for information on adverse events. We will search for errata or retractions from included studies published in full-text on PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) and report the date this was done within the review.
Data collection and analysis
Selection of studies
Two authors (ZQ, YW) will assess the results of the searches. To determine which studies are to be assessed further, we will independently scan the titles, abstracts and key-words of every record retrieved. We will retrieve full-text articles for further assessment. Where necessary, we will contact the original authors of identified studies for missing information. Differences in opinion will be resolved by discussion, or a third author will be consulted. If bisphosphonates were included in any multi-armed RCTs, we will extract all relevant data and include them in our analyses.
Data extraction and management
Two authors (ZQ, YW) will independently assess each trial and extract information on trial design, study population, interventions and outcomes using a data extraction form specifically designed for this review. Differences will be resolved by referring back to the original article and reaching a consensus. When necessary, we will seek further information from the trial authors.
During the data extraction process, if a mixture of change-from-baseline scores (change scores) and final measurement (final values) are to be identified, we will put priority on the change scores on the ground that change scores remove a component of between-person variability, and was also considered to have a less skewed distribution (Deeks 2011). However, if final values are explicitly reported and change scores need to be imputed, we will include final values, considering that change scores and final values could be combined in a meta-analysis when using the (unstandardised) mean difference method.
We also intend to extract data at all periods of follow-up for each outcome in all trials so as to make the most use of the information available, and present a summary effect over different time points predefined as short, medium, and long-term.
Assessment of risk of bias in included studies
Two authors (ZQ, YW) will independently assess risk of bias using the criteria described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). The following sources of bias will be assessed.
- Randomization process: assessment for selection bias
- Allocation concealment process: assessment of selection bias
- Blinding of participants and care personnel: assessment for performance bias
- Blinding of outcome assessors: assessment for detection bias
- Incomplete data bias
- Selective reporting bias
- Major imbalances in key baseline characteristics
We will grade each potential source of bias as 'high risk', 'low risk' or 'unclear risk' and provide a quote from the study report together with a justification for our judgment in the 'Risk of bias' table. We will summarise the risk of bias judgements across different studies for each of the domains listed. We will consider blinding separately for different key outcomes where necessary (e.g. for unblinded outcome assessment, risk of bias for all-cause mortality may be very different than for a patient reported pain scale). As well, we will consider the impact of missing data by key outcomes.
Where information on risk of bias relates to unpublished data or correspondence with a trialist, we will note this in the 'Risk of bias' tables.
When considering treatment effects, we will take into account the risk of bias for the studies that contribute to that outcome.
We will present the figures generated by the risk of bias tool to provide summary assessments of the risk of bias. Where necessary, disagreements will be resolved by discussion.
Measures of treatment effect
We expect to find both event (dichotomous) data (e.g. adverse events, and percentage of 'progressors') and continuous data (e.g. pain and function related outcomes, and joint space width). We will use risk ratios (RRs) with 95% confidence intervals (CIs) for reporting dichotomous data. We intend to calculate the risk difference (RD) and convert the RD into the number needed to treat for an additional beneficial outcome (NNTB) or the number needed to treat for an additional harmful outcome (NNTH) if the RR is statistically significant. We will express continuous data as mean differences (MDs) with 95% CIs. When different scales are used to measure the same conceptual outcome (e.g. disability), standardised mean differences (SMDs) will be calculated instead, with corresponding 95% CIs. We will back-translate SMDs to a typical scale (e.g. 0 to 10 for pain) by multiplying the SMD by a typical among-person standard deviation (e.g. the standard deviation of the control group at baseline from the most representative trial).
Unit of analysis issues
We expect the unit of analysis in this review to be individual patients, but will be alert to other potential unit of analyses issues, such as the inclusion of patients undergoing more than one intervention, repeated observation from more than one time-point, and multiple observations for the same outcome (e.g. total adverse events). Regarding multiple time points observation, we aim to additionally perform analysis based on different periods of follow-up. Should any such issues arise, we will seek further advice from methodological experts. Where multiple trial arms are reported in a single trial, we will include only the relevant arms. If two comparisons (e.g. drug A versus placebo and drug B versus placebo) are combined in the same meta-analysis, we will halve the control group to avoid double-counting.
Dealing with missing data
When necessary, we will seek missing data, particularly denominators and standard deviations, from the authors of the primary studies. We will perform intention-to-treat (ITT) analyses wherever possible. We will not impute missing standard deviations (SDs) unless they can be inferred from standard error (SE), CI, t values and P values. If missing data is imputed, we will undertake a sensitivity analysis to explore the effect of including the imputed data.
Assessment of heterogeneity
Besides the visual inspection of forest plot analyses, we will examine statistical heterogeneity using the Chi
Assessment of reporting biases
Provided data can be pooled from 10 or more trials, we will examine potential publication biases by producing funnel plots (Egger 1997). For studies published after 1st July 2005, we will screen the International Clinical Trials Registry Platform (ICTRP) search portal of the World Health Organisation (WHO) (http://apps.who.int/trialsearch/) for the a priori trial protocol. We will evaluate whether selective reporting of outcomes is present (outcome reporting bias).
We intend to perform all the meta-analyses using random-effects model, because it was suggested that this model is the most appropriate for the majority of meta-analyses when studies included in the analyses were gathered from the published literature, in which circumstance, the trial data typically would not be in accordance with the assumptions of the fixed-effect model, i.e., all studies share a common true effect and all observed variation only reflects sampling error. Therefore, while producing the same point estimate as fixed-effect model, random-effects model tends to produce more conservative confidence intervals better reflecting the uncertainty around the point estimate (Borenstein 2009; Schmidt 2009).
Subgroup analysis and investigation of heterogeneity
We will assess possible sources of heterogeneity using subgroup analyses as follows, which may be conducted based on:
- first generation versus second and third generation bisphosphonates;
- dosage of bisphosphonates (total amount taken per day, e.g. 5mg/day and 15 mg/day);
- location of the study joint (hands may have a different response in that they are not weight-bearing joints as the hip and knee);
- duration of OA (≤ 6 months, 7 to 24 months, and ≥24 month).
We plan to perform sensitivity analyses in order to explore the influence of the following factors on effect size, by repeating the analysis and:
- excluding unpublished studies;
- excluding studies with unclear or high risk of bias relating to non concealment of allocation, non blinding of outcome assessors, non-ITT analysis, incomplete outcome data;
- using different measures of calculating treatment effects (e.g. MDs versus SMDs) and different statistical models (fixed-effect versus random-effects models).
'Summary of findings' table
We will illustrate key information in terms of the quality of evidence, the magnitude of treatment effect of interest, and an overall grading of the evidence on the basis of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach using 'Summary of findings' tables (Schünemann 2011), which may include the following outcomes.
- Pain relief
- Physical function
- Quality of life
- Radiographic joint structure
- Number of patients with serious adverse events
- Number of patients experiencing any adverse event
- Patients who withdrew because of adverse events
Two authors (ZQ and ZDT) will independently assess the quality of the evidence. We will use the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta-analyses for the prespecified outcomes. We will use methods and recommendations described in the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011b). We will justify all decisions to down- or up-grade the quality of studies using footnotes and we will make comments to aid reader's understanding of the review where necessary.
We will provide the absolute percent difference, the relative percent change from baseline, and the number needed-to-treat (NNT) (the NNT will be provided only when the outcome shows a statistically significant difference).
For dichotomous outcomes, such as serious adverse events, we will calculate the NNT from the control group event rate and the relative risk using the Visual Rx NNT calculator (Cates 2008). For continuous measures, we will calculate the NNT using the Wells calculator (available at the Cochrane Musculoskeletal Group Editorial office).
For dichotomous outcomes, we will calculate the absolute risk difference using the 'risk difference statistic in the Cochrane Collaboration's statistical software, Review Manager 2013, and express the result as a percentage. For continuous outcomes, we will calculate the absolute benefit as the improvement in the intervention group minus the improvement in the control group, in the original units.
We will calculate the relative percent change for dichotomous data as 'Risk ratio - 1' and express it as a percentage. For continuous outcomes, we will calculate the relative difference in the change from baseline as the absolute benefit divided by the baseline mean of the control group.
Interpreting results and reaching conclusions
We will follow the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions for interpreting results and will be aware of distinguishing a lack of evidence of effect from a lack of effect (Higgins 2011b). We will base our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. We will avoid making recommendations for practice and our implications for research will suggest priorities for future research and outline what the remaining uncertainties are in the area.
Assessment of bias in conducting the systematic review
We will conduct the review according to this published protocol and report any deviations from it in the 'Differences between protocol and review' section of the systematic review.
Our gratitude goes to Elizabeth Ghogomu, Lara Maxwell, Laura Laslett for valuable comments and help about the protocol. We also acknowledge the help of Renea Johnston along the way.
Appendix 1. MEDLINE search strategy
- exp OSTEOARTHRITIS/
- (degenerative adj2 arthritis).tw.
- exp Diphosphonates/
- exp Etidronic Acid/
- exp Clodronic Acid/
- exp Alendronate/
- exp Zoledronic Acid/
- 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24
- 4 and 25
- Randomized controlled trial.pt.
- Controlled clinical trial.pt.
- Drug therapy.fs.
- exp Animals/ not Humans/
- 34 not 35
- 26 and 37
Contributions of authors
Wei Yang, Wei Chai, Datong Zeng, Jiying Chen, Cheng Sun, Qi Zhuo and Yan Wang contributed equally to the preparation of the protocol. Yan Wang will be the guarantor of the review.
Declarations of interest
Sources of support
- Chinese PLA General Hospital, China.
- National Natural Science Foundation of China, China.