Article Text
PDF
Abstract
Myofascial pain syndrome (MPS) is characterized by the presence of clinically detected myofascial trigger points (MTrPs). Diagnostic ultrasound (US) has been proposed as a method to strengthen the reliability of MTrP localization, thus potentially improving the efficacy and safety profile of interventional procedures. The objective is to evaluate the benefit and safety profile of any US-guided interventional procedure for MPS. Medline, Embase, PubMed, the Allied and Complementary Medicine Database (AMED), and Web of Science were systematically searched from their inception to May 2020 for any randomized controlled trial that evaluated treatment benefit and safety of any US-guided interventional procedure for MPS. The primary outcome of interest was pain severity. Additional outcomes of interest were function and adverse effects. The risk of bias was assessed using the Risk of Bias V.2.0 tool. eleven studies met all inclusion and exclusion criteria. Two studies (n=174) with a high risk of bias revealed some evidence supporting US guidance over blinded interventions for improvement in pain and function. Eight studies (n=483) with varying risks of bias were of head-to-head comparisons of different US modalities. These studies revealed that US-guided local anesthetic injections were inferior to US-guided pulse radiofrequency and US-guided dry needling (DN). US-guided DN was also found to be superior to US-guided platelet-rich-plasma injections but inferior to US-guided miniscalpel. Meanwhile, one study (n=21) with some concerns of bias found that US-guided local anesthetic injections were superior to non-steroidal anti-inflammatory drugs for pain outcomes and fewer adverse events. All US-guided procedures resulted in zero or minimal self-limited adverse events. Issues with clinical relevance, limited sample sizes, and small point estimates warrant more high-quality research to better characterize the possible value of US-guided injections.
- ultrasonography
- diagnostic techniques and procedures
- chronic pain
Statistics from Altmetric.com
Introduction
Myofascial pain syndrome (MPS) is a common regional musculoskeletal pain syndrome that can cause local or referred pain.1–3 MPS is characterized by myofascial trigger points (MTrPs), which are hard, palpable, discrete, and localized nodules located within taut bands of skeletal muscle that can be painful on compression.4 The prevalence of MPS varies from 21% of patients seen in general orthopedic clinics and 30% of patients seen in general medical clinics.5 6 In specialty pain management centers, as many as 85%–93% of patients present with myofascial pain.7 MTrPs are most commonly distributed in the neck, shoulder, and pelvic girdle; specific muscles include the trapezius, scalene, sternocleidomastoid, levator scapulae, and quadratus lumborum.8
Traditionally, the diagnosis of MPS has relied solely on clinical findings proposed by Travell and Simons.8 However, it was found that such criteria were used inconsistently in research.9 Subsequently, attempts have been made to standardize the clinical diagnostic criteria using survey consensus data of physicians.10 11 Despite these efforts, physicians are still required to rely on the presence (or absence) of a collection of clinical signs and symptoms to determine the existence of a MTrP. In fact, multiple reviews have found poor agreement and reliability of manual palpation for MTrP localization.12 13 These diagnostic limitations have direct treatment implications as common intervention-based therapies for MPS are centered on targeting MTrPs.
Diagnostic ultrasound (US) imaging is a promising method that can be used to strengthen the reliability of MTrP localization. Combined with a physician’s clinical assessment, US imaging (via Bright-mode, Doppler, or elastography) can provide objective evidence of active and latent MTrPs by identifying areas with sonographic characteristics distinct from normal tissue.11 14–17 Such characteristics have been previously described, and despite no definite consensus in the literature, common descriptions include: spherical/elliptical shape and hypoechogenic on Bright-mode; stiffer (reduced vibration amplitude) on elastography; and high peak systolic velocity, low peak diastolic velocity, retrograde diastolic flow, increased blood volume, and increased outflow resistance on Doppler. In patients with MPS, the exact percentage of those with sonographic changes consistent with MTrPs is unclear. One study visualized 49.4%±14.6% MTrPs via Bright-mode and 64.4%±2.4% MTrPs via elastography; however, visualization varied greatly depending on MTrP location.18 Another study was able to visualize all MTrPs via Bright-mode.19 It is plausible that the improved diagnostic accuracy obtained from US imaging may improve the efficacy of interventional procedures that rely on correct localization of MTrPs as compared with the “blind” technique. US may also offer more accurate insertion of the needle used for injection and may also help to prevent adverse consequences.
Our previous review consisted of two studies, both demonstrating statistically significant improvements in pain and function using US-guided injections over blinded injections.20 However, the criteria used in our previous review was limited to comparisons between US-guided injections versus blinded injections for the upper trapezius muscle.20 This review serves as an update and expansion to our previous review on US-guided injections.20 Given the rapidly evolving field of US-guided procedures, we expected there to be numerous studies published since our last search in 2016. Additionally, we have since expanded our criteria to include any US-guided interventional procedure on MTrPs and to any injection site. The objective of our review was to evaluate the benefit and safety profile of any US-guided interventional procedure for MPS.
Methods
A systematic review investigating the benefits and safety of any US-guided interventional procedure was conducted. This review is registered in PROSPERO (registration number: CRD42020184891) and was reported in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.
Data sources and searches
We performed English-based searches in Medline, Embase, PubMed, AMED, and Web of Science using combinations of the following key terms: “Myofascial Pain Syndrome”, “Trigger Points”, “Injections”, “Dry Needling”, “Ultrasound”, “Ultrasonography”. Using similar key terms, we also performed English-based searches in the following clinical trial registries: United States (ClinicalTrials.gov), Canada (https://health-products.canada.ca/ctdb-bdec/index-eng.jsp), Europe (https://www.clinicaltrialsregister.eu/), and Australia (https://www.australianclinicaltrials.gov.au/). All databases were searched from their inception to May 2020 and were repeated twice to ensure robustness. The reference and citation lists of included studies were hand-searched and if a new study was found to be appropriate for inclusion into the review, it was reviewed by DD and KJQC and included if it met our criteria.
Design and study selection
At least two independent reviewers (DD and KJQC) conducted citation identification through abstract and full-text screening, as well as study selection. Disagreements were resolved through a third assessor (DK). The following inclusion and exclusion criteria were used to review the search and retrieve potentially relevant studies.
Types of studies
Any randomized controlled trial (RCT) in the English language that evaluated treatment benefit and safety of any US-guided interventional procedure for myofascial pain was included. Studies must have been peer reviewed and have at least one arm that investigated an US-guided interventional procedure.
Types of participants
Any patient population described to have pain from a myofascial origin, which could be further characterized as a taut band, local tenderness on palpation, local pain that heightens with use, pain recognition, referred pain, local twitch response, restricted range of motion, reproducible pain pattern, or weakness without atrophy. This established the diagnosis of MPS and the presence of MTrPs clinically. Patients of any age, sex, and gender were included.
Types of interventions
Any US-guided interventional procedure into any injection site deemed to be a trigger point or source of myofascial pain was included. Any US modality (eg, gray-scale/Bright-mode, Doppler, elastography, etc) was included. Interventional procedures were considered to be any injection (eg, local anesthesia, prolotherapy, botulinum toxin, etc) or procedure that penetrated the skin (eg, dry needling (DN), pulsed radiofrequency (PRF), etc) to any site of myofascial pain.
Type of comparators
All comparator groups were included. Comparator groups were stratified under the following three categories: blinded (non-US-guided) interventional procedures, all other treatments involving interventional procedures (eg, other US-guided injections), and non-interventional therapies (eg, pharmacological and non-pharmacological interventions).
Types of outcome measures
The primary outcome of interest was pain severity, which could be directly measured using validated scales, such as the visual analog scale (VAS) or numerical rating scale (NRS).21 More indirect measurement scales that included a pain component were also included (eg, quality of life scales).22 23
Secondary outcomes included function and adverse events. Function could be measured using any validated scale that measures the interference of occupational, recreational, social, emotional, or general (eg, activities of daily living and instrumental activities of daily living) function.24–27 We considered any treatment or nontreatment-related adverse events. Examples of treatment-related adverse event include injection site pain, blood loss, infection, and spasm. Clinical judgment was used to differentiate this from non-treatment-related adverse events, especially if adverse events may have been attributed to disease status or environmental factors.
Exclusion criteria
Studies were excluded if they were a non-RCT study (eg, review articles), if patients had pain secondary to systemic comorbidities such as cancer or diabetes, if non-human subjects were involved, or if no US-guided intervention was used.
Data extraction and quality assessment
At least two independent reviewers extracted aggregated data for demographics, descriptions of interventions, and all outcomes to pre-designed abstraction forms. Disagreements were resolved by repeated review and consensus. Registration data from trial protocols, if available, were used to obtain any missing information. If specific values were not reported, values were estimated from any given graphs. When data were not extractable, the primary authors were contacted.
At least two independent reviewers assessed each study for bias using the Risk of Bias V.2.0 tool by the Cochrane Collaboration. Using the tool, bias in the following five domains were assessed: bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in measurement of the outcome, and bias in selection of the reported result. Disagreements were resolved through a third assessor.
Data synthesis and analysis
Given the substantial clinical heterogeneity of included studies, we deemed a meta-analysis to be inappropriate. Instead, a narrative synthesis of the findings was performed. Outcomes were stratified by comparing US-guided interventional procedures with the following comparators: (1) blinded interventional procedures, (2) all other treatment involving interventional procedures, and 3) non-interventional therapies. Adapted from the Cochrane Handbook,28 the duration of follow-up was defined as the time between the completion of the intervention and the measurement of the outcome, and categorized as:
Immediate term: less than 1 week.
Short term: longer than 1 week and less than 6 weeks.
Long term: longer than 6 weeks and up to 1 year.
If studies reported the same outcome multiple times during one time point (eg, VAS reported at 2 weeks and 4 weeks), the longest follow-up was used. Moreover, if there was more than one study in each category of data synthesis, we performed a sensitivity analysis to separate RCTs based on (1) any risk of bias and (2) only low risk of bias. All data are reported as means and SD.
Moreover, a rating of clinical relevance was determined for each outcome based off its minimally clinically important difference (MCID).21–27 MCIDs were determined from a literature search for primary studies of chronic pain populations. MCIDs were used as a threshold such that outcomes below the MCID were considered unlikely to be clinically relevant. Outcomes exceeding the MCID were considered likely to be clinically relevant. A rating of unclear clinical relevance was given to outcomes with insufficient information to calculate MCID (eg, unreported mean difference).
Results
Out of the 990 articles retrieved from Medline, Embase, PubMed, AMED, and Web of Science databases, 518 duplicates were removed. The 472 remaining articles proceeded through title and abstract screening, after which 30 articles were selected for full-text screening. Of these 30 articles, 20 articles did not meet all inclusion and exclusion criteria and 10 studies were included (see online supplemental table 1: Summary of Excluded Studies). Hand searches of reference lists of all included studies yielded one eligible study (Bubnov 2013, Medical Acupuncture). A repeat search did not yield any additional results. Two clinical trials were ongoing at the time of the search, both of which were in their preliminary stages.29 30 Overall, 11 studies were used for this systematic review (figure 1; table 1).18 31–40
Supplemental material
Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram.
Characteristics of included studies
All included studies were single-center parallel-group RCTs with two arms. Studies were conducted in China,33 Korea,34 35 37 38 Turkey,36 40 Ukraine,18 32 39 and the USA.31 The specific interventions used included: US-guided local anesthetic injections,31 32 34–38 40 US-guided DN,18 32 33 39 US-guided PRF,35 38 US-guided platelet-rich-plasma (PRP) injections,39 US-guided miniscalpel release,33 US-guided saline injections,40 blinded local anesthetic injections,37 blinded DN,18 and oral non-steroidal anti-inflammatory drugs (NSAID).34 All interventional procedures were performed once per identified MTrP. However, one exception was with Farrow et al31 who included a comparator group that received US-guided local anesthetic injections to non-MTrPs. Otherwise, four studies used MTrPs in the trapezius muscle as their intervention site.36–38 40 Single studies used MTrPs in shoulder muscles,32 neck muscles,33 neck and back muscles,31 brachialis muscles,34 gastrocnemius muscles,35 and pterygopalatine muscles18 as their intervention site. One study did not specify the location of MTrPs.39
Outcome measures used included VAS pain score,18 31–34 36 37 39 40 Shoulder Pain and Disability Index (SPADI),37 neck disability index (NDI),33 37 40 NRS pain score,35 38 short form 36 physical component summary (SF-36 PCS) score33 35 36 38 and mental component summary (SF-36 MCS) score,33 35 36 38 pain pressure threshold (PPT),36 and neck pain and disability scale (NPDS) score.36 A detailed description of each outcome measure and the derivation of MCID is included in online supplemental appendix 1.
US-guided interventions versus blinded interventions
There were two studies in this category (n=174).18 37 Both studies had a high risk of bias. The first study by Bubnov and Wang included an US-guided DN group (n=91) and a blinded (via clinically established landmarks) DN group (n=42) to the pterygopalatine muscle.18 In the immediate-term, the US-guided group showed a significant reduction (p<0.001) in VAS pain scores from 7.2±3.8 to 1.1±0.48 (pain level decreased in 84% of the subjects). Meanwhile, a significant reduction (p<0.001) from 7.4±1.94 to 2.7±1.30 (pain level decreased in 64% of the subjects) was observed in the blinded group. However, no between-group analyses were performed. Function and adverse events were not reported.
The latter study by Kang et al37 included an US-guided local anesthetic injection group (n=21) and a blinded local anesthetic injection group (n=20) to the trapezius muscle. In the short-term, a significant between-group difference (p=0.003) in VAS pain scores was found in favor of the US-guided group (−1.92±0.56) over the blinded group (−1.20±0.85). Significant between-group differences in the SPADI and the NDI were also observed. SPADI scores improved by 20.14±8.90 in the US-guided group versus 9.70±16.39 in the blinded group (p=0.018). NDI scores improved by 11.14±4.19 in the US-group versus 5.85±7.80 in the blinded group (p=0.012). However, no statistically significant between-group differences were observed with respect to muscle strength and range of motion. Adverse events were not reported.
US-guided interventions versus other US-guided interventions
There were eight studies in this category (n=483).31–33 35 36 38–40
Head-to-head comparisons of different US-guided interventions
Six studies were head-to-head comparisons of different US-guided interventions.32 33 35 38–40 Three studies had high risk of bias,32 39 40 two studies had some concerns of bias,35 38 and one study had a low risk of bias.33
Analan et al40 included an US-guided local anesthetic injection group (n=29) and an US-guided saline injection group (n=29) to the trapezius muscle. In the short term, both groups experienced statistically significant improvements in VAS pain scores and NDI scores from baseline. However, no significant between-group differences were observed for VAS (4.5±3.41 vs 4.44±2.97; p=0.95) and NDI (16.3±16.26 vs 21.5±18.93; p=0.34) scores. Two patients reported injection-site pain (3.4%) and one patient reported swelling (1.7%), but it was unclear from which groups.
US-guided local anesthetic injections were also compared with US-guided PRF and to US-guided DN. Two studies investigated the former comparison.35 38 Cho et al38 included an US-guided local anesthetic injection group (n=18) and an US-guided PRF group (n=18) to the trapezius muscle. In the short-term, a significant between-group difference (p<0.001) in NRS pain scores was found in favor of the US-guided PRF group (6.2±1.1 to 2.5±0.8) over the US-guided local anesthetic injection group (6.5±0.9 to 4.2±1.2). In the long-term, significant between-group differences in NRS pain scores (p<0.001) and SF-36 PCS and SF-36 MCS scores (p=0.001; p<0.001) were observed in favor of the US-guided PRF group (6.2±1.1 to 3.2±0.9; 32.8±3.0 to 40.7±4.6; 35.6±2.7 to 44.6±5.1) over the US-guided local anesthetic injection group (6.5±0.9 to 5.3±1.0; 31.9±3.0 to 34.3±3.1; 37.6±3.1 to 39.1±3.5). Patients experienced no adverse events. Meanwhile, Park et al35 employed similar methods but for the gastrocnemius muscle. In the immediate term, a significant between-group difference (p<0.001) in NRS pain scores was found in favor of the US-guided local anesthetic injection group (5.2±1.0 to 2.0±0.6; n=20) over US-guided PRF group (5.5±0.9 to 5.0±0.7; n=20). In the short term, significant between-group differences in NRS pain scores (p<0.001) and SF-36 PCS and SF-36 MCS scores (p<0.001; p=0.002) were observed in favor of the US-guided PRF group (5.5±0.9 to 2.4±1.0; 31.0±4.5 to 41.7±7.3; 34.8±3.7 to 44.0±5.9) over the US-guided local anesthetic injection group (5.2±1.0 to 4.0±0.8; 32.0±3.3 to 32.9±2.8; 36.8±3.5 to 37.8±3.6). Two patients reported dizziness and nausea (5.0%) in the US-guided injection group.
Bubnov et al32 investigated the latter comparison of US-guided local anesthetic injections (n=22) and US-guided DN (n=22) into shoulder muscles. In the immediate-term, a significant between-group difference (p<0.001) in VAS pain scores was found in favor of the US-guided DN group (7.5 to 1.1) over the US-guided local anesthetic injection group (7.4 to 4.2). Moreover, Bubnov and Fodor reported that US-guided dry needling was found to be superior to US-guided PRP injections in the immediate and short term for reducing VAS pain scores (n=20; p<0.01), although neither the number of participants per group nor quantitative data were provided.39 Adverse events were not reported. Finally, Zheng et al included an US-guided miniscalpel release group (n=88) and an US-guided DN group (n=81) to neck muscles.33 In the long term, significant between-group differences in VAS pain scores (p<0.0001), NDI scores (p<0.0001), and SF-36 PCS scores (p=0.024) were found in favor of the US-guided miniscalpel group (6.8±1.6 to 3.8±1.2; 17.9±8.3 to 7.1±2.6; 41.3±14.0 to 55.1±19.0) over the US-guided DN group (7.1±1.8 to 5.8±1.4; 16.0±5.4 to 12.2±5.5; 43.6±16.1 to 47.9±21.5). No significant between-group differences were observed for SF-36 MCS scores (p=0.801). Self-limited minor adverse events (ie, slight pain and diaphoresis) were reported in six patients in the US-guided miniscalpel release group (6.8%) and seven patients in the US-guided DN group (8.6%).
US-guided local anesthetic injections at different injection sites
Two studies investigated US-guided local anesthetic injections at different injection sites.31 36 Both studies had a high risk of bias. Farrow et al investigated pain responses to injections to MTrPs (n=24) versus non-MTrPs (n=27) in neck and back muscles.31 In the immediate term, no significant between-group differences were observed in pain scores on a 0–100 analog scale (p=1.00). Okmen et al investigated pain and function responses to injections to MTrPs in the rhomboid (n=33) versus trapezius (n=32) muscles. In the short term, significant between-group differences in VAS pain scores (p<0.001), PPT (p<0.001), NPDS scores (p<0.001), and SF-36 MCS scores (p=0.002) were found in favor of the rhomboid injection site (−5.56±1.11; 7.86±1.53; −28.75±8.1; 14.63±10.97) over the trapezius site (−2.41±1.15; 3.97±1.38; −12.62±5.39; 5.84±10.90). No significant between-group differences were observed for SF-36 PCS scores (p=0.074).
US-guided interventions versus non-interventional therapies
There was one study in this category (n=21).34 This study had some concerns of bias. The study by Suh et al included one group that received an US-guided injection of local anesthetic and triamcinolone (n=11) to the brachialis muscle, and another group that received a 2-week course of two times a day 500 mg oral naproxen (n=10). In the short term, a significant between-group difference (p<0.05) in VAS pain scores was found in favor of the US-guided group (−2.8±1.4) over the NSAID group (−1.5±1.3). Pain levels decreased in 55% of the US-guided group versus 10% of the naproxen group. Function was not reported. The US-guided group experienced no adverse events, but two patients (20%) in the NSAID group experienced gastric pain or discomfort related to the treatment.
Discussion
A summary of findings, including efficacy, overall risk of bias, and ratings of clinical relevance, is depicted in table 2. Narrative syntheses revealed some evidence supporting US-guidance over blinded interventions for improvement in pain and function.18 37 Head-to-head comparisons of different US modalities revealed no differences in pain and function outcomes between US-guided local anesthetic injections vs placebo.40 With respect to target injection sites, local anesthetic injections targeting MTrPs in rhomboid muscles led to more favorable outcomes than in the trapezius muscles.36 However, no differences were observed when comparing between local anesthetic injections into MTrPs and non-MTrPs.31 Moreover, all other comparisons with US-guided local anesthetic injections demonstrated its inferiority to US-guided PRF and US-guided DN.32 35 38 US-guided DN was also found to be superior to US-guided PRP injections,39 but inferior to US-guided miniscalpel.33 Only one comparison was made between US-guided modalities and non-interventional therapies, which concluded that US-guided local anesthetic injections were superior to NSAIDs for pain outcomes and fewer adverse events.34 All US-guided procedures resulted in zero or minimal self-limited adverse events.
Summary of findings and clinical relevance
Seven studies had a high risk of bias,18 31 32 36 37 39 40 three studies had some concern of bias,34 35 38 and only one study had a low risk of bias33 (figure 2; figure 3). Potential biases arising from the randomization process (81%) as well as selection of the reported result (91%) were the most prevalent. Of the statistically significant findings, 5 findings were below the MCID, 15 findings exceeded the MCID, and 2 findings lacked sufficient information to make a rating of clinical relevance. Studies also consisted of limited sample sizes and considerable clinical heterogeneity, affecting the overall interpretation of results. For instance, only one study was available for the comparison between US-guided injection of local anesthetic (n=11) versus NSAIDs (n=10), which was over a short 2-week treatment span.34 The results were also below the MCID for the VAS pain scale, therefore unlikely to be clinically relevant.21 34 There were no two studies that employed identical interventions, comparators, US-guided techniques, treatment duration, and follow-up. As such, no pooling of data was possible. Frequent limited sample sizes and small point estimates warrant extreme caution with interpretation of results. Nevertheless, US-guided intervention research is rapidly emerging for MPS. Our previous systematic review which conducted its search in 2016 yielded only two studies.20 Additionally, most findings support the use of US-guided interventions, especially when compared with blinded injection techniques. Since all interventional techniques are, in theory, dependent on accurate identification of MTrPs, the positive results from US-guided interventions directed at MTrPs are as expected.18 32–40 By improving accuracy of interventional procedures, patients require less medication or intervention attempts, while maximizing the benefits. However, it is important to acknowledge that accurate MTrP localization is dependent on practitioner experience, type of US modality, and affected muscle group. Elastography has been found to have greater yield in MTrP localization compared with Bright-mode sonography, although the former is limited by increased technical complexity.14 16 17 Additionally, echogenicity may vary depending on muscle group.18 19 Considering that MPS is a clinical diagnosis, we recommend US-guidance as an adjunct to palpation of MTrPs, and not as a stand-alone diagnostic tool. Doing so may not only improve accuracy but also reduce adverse events such as hematoma and pneumothorax through visualization of blood vessels and lung pleura, respectively.
Overall risk of bias assessments.
Risk of bias of included studies.
There is also new evidence to suggest that US-guided injections into MTrPs versus non-MTrPs yield no differences in immediate pain outcomes, which may imply that accurate localization of MTrPs may be unnecessary.31 This finding may also signify that the MTrP which has been thought to be the cause of the pain in MPS may not be and that the other parts of muscle may also be affected. Further research is required to validate such theories.
Study limitations
Our review was limited by small sample sizes, clinical heterogeneity, search parameters, and ratings on clinical relevance. First, all studies varied considerably with respect to intervention, comparison, and target intervention sites. Additionally, the US-guided techniques varied considerably between studies. Limited sample sizes and considerable clinical heterogeneity of results must be overcome with future high-powered, low risk of bias RCTs. Standardization of US-guided techniques would also allow for better reproducibility and more accurate comparisons to be made between trials. Furthermore, some relevant non-English studies may have been missed due to our language restriction, despite systematically searching five databases, repeating the search, and hand-searching reference and citation lists. Lastly, there are limitations that may have affected the rating of clinical relevance. There is a lack of consensus regarding MCIDs for the outcome metrics. Moreover, we were sometimes unable to find MCIDs that were determined from MPS populations. As a result, we based our judgment on MCIDs calculated from other patient populations, for example chronic neck pain.
Conclusion
The value of US-guided interventions remains unclear for treatment of MPS. Although there is some evidence that pain and functional outcomes are improved with the addition of US-guidance, issues with clinical relevance, limited sample sizes, and small point estimates warrant more high-quality research to validate and build on current findings.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors DD and DK were responsible for the study conception and design. DD and KJQC were responsible for data extraction and validation, and data analysis and interpretation. All authors provided a critical review, drafted the manuscript, and approved the final manuscript. DK is the guarantor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.