Machine learning for identifying Randomized Controlled Trials: An evaluation and practitioner's guide

A survey on clinical natural language processing in the United Kingdom from 2007 to 2022

Much of the knowledge and information needed for enabling high-quality clinical research is stored in free-text format. Natural language processing (NLP) has been used to extract information from these sources at scale for several decades. This paper aims to present a comprehensive review of clinical NLP for the past 15 years in the UK to identify the community, depict its evolution, analyse methodologies and applications, and identify the main barriers. We collect a dataset of clinical NLP projects (n = 94; £ = 41.97 m) funded by UK funders or the European Union's funding programmes. Additionally, we extract details on 9 funders, 137 organisations, 139 persons and 431 research papers. Networks are created from timestamped data interlinking all entities, and network analysis is subsequently applied to generate insights. 431 publications are identified as part of a literature review, of which 107 are eligible for final analysis. Results show, not surprisingly, clinical NLP in the UK has increased substantially in the last 15 years: the total budget in the period of 2019-2022 was 80 times that of 2007-2010. However, the effort is required to deepen areas such as disease (sub-)phenotyping and broaden application domains. There is also a need to improve links between academia and industry and enable deployments in real-world settings for the realisation of clinical NLP's great potential in care delivery. The major barriers include research and development access to hospital data, lack of capable computational resources in the right places, the scarcity of labelled data and barriers to sharing of pretrained models.

Vaccines for the common cold

BACKGROUND

The common cold is a spontaneously remitting infection of the upper respiratory tract, characterised by a runny nose, nasal congestion, sneezing, cough, malaise, sore throat, and fever (usually < 37.8 ºC). Whilst the common cold is generally not harmful, it is a cause of economic burden due to school and work absenteeism. In the United States, economic loss due to the common cold is estimated at more than USD 40 billion per year, including an estimate of 70 million workdays missed by employees, 189 million school days missed by children, and 126 million workdays missed by parents caring for children with a cold. Additionally, data from Europe show that the total cost per episode may be up to EUR 1102. There is also a large expenditure due to inappropriate antimicrobial prescription. Vaccine development for the common cold has been difficult due to antigenic variability of the common cold viruses; even bacteria can act as infective agents. Uncertainty remains regarding the efficacy and safety of interventions for preventing the common cold in healthy people, thus we performed an update of this Cochrane Review, which was first published in 2011 and updated in 2013 and 2017.

OBJECTIVES

To assess the clinical effectiveness and safety of vaccines for preventing the common cold in healthy people.

SEARCH METHODS

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (April 2022), MEDLINE (1948 to April 2022), Embase (1974 to April 2022), CINAHL (1981 to April 2022), and LILACS (1982 to April 2022). We also searched three trials registers for ongoing studies, and four websites for additional trials (April 2022). We did not impose any language or date restrictions.

SELECTION CRITERIA

Randomised controlled trials (RCTs) of any virus vaccine compared with placebo to prevent the common cold in healthy people.

DATA COLLECTION AND ANALYSIS

We used Cochrane's Screen4Me workflow to assess the initial search results. Four review authors independently performed title and abstract screening to identify potentially relevant studies. We retrieved the full-text articles for those studies deemed potentially relevant, and the review authors independently screened the full-text reports for inclusion in the review, recording reasons for exclusion of the excluded studies. Any disagreements were resolved by discussion or by consulting a third review author when needed. Two review authors independently collected data on a data extraction form, resolving any disagreements by consensus or by involving a third review author. We double-checked data transferred into Review Manager 5 software. Three review authors independently assessed risk of bias using RoB 1 tool as outlined in the Cochrane Handbook for Systematic Reviews of Interventions. We carried out statistical analysis using Review Manager 5. We did not conduct a meta-analysis, and we did not assess publication bias. We used GRADEpro GDT software to assess the certainty of the evidence and to create a summary of findings table.  MAIN RESULTS: We did not identify any new RCTs for inclusion in this update. This review includes one RCT conducted in 1965 with an overall high risk of bias. The RCT included 2307 healthy young men in a military facility, all of whom were included in the analyses, and compared the effect of three adenovirus vaccines (live, inactivated type 4, and inactivated type 4 and 7) against a placebo (injection of physiological saline or gelatin capsule). There were 13 (1.14%) events in 1139 participants in the vaccine group, and 14 (1.19%) events in 1168 participants in the placebo group. Overall, we do not know if there is a difference between the adenovirus vaccine and placebo in reducing the incidence of the common cold (risk ratio 0.95, 95% confidence interval 0.45 to 2.02; very low-certainty evidence). Furthermore, no difference in adverse events when comparing live vaccine preparation with placebo was reported. We downgraded the certainty of the evidence to very low due to unclear risk of bias, indirectness because the population of this study was only young men, and imprecision because confidence intervals were wide and the number of events was low. The included study did not assess vaccine-related or all-cause mortality.  AUTHORS' CONCLUSIONS: This Cochrane Review was based on one study with very low-certainty evidence, which showed that there may be no difference between the adenovirus vaccine and placebo in reducing the incidence of the common cold. We identified a need for well-designed, adequately powered RCTs to investigate vaccines for the common cold in healthy people. Future trials on interventions for preventing the common cold should assess a variety of virus vaccines for this condition, and should measure such outcomes as common cold incidence, vaccine safety, and mortality (all-cause and related to the vaccine).

Effectiveness and tolerability of dual and triple combination inhaler therapies compared with each other and varying doses of inhaled corticosteroids in adolescents and adults with asthma: a systematic review and network meta-analysis

BACKGROUND

Current guidelines recommend a higher-dose inhaled corticosteroids (ICS) or adding a long-acting muscarinic antagonist (LAMA) when asthma is not controlled with medium-dose (MD) ICS/long-acting beta2-agonist (LABA) combination therapy.

OBJECTIVES

To assess the effectiveness and safety of dual (ICS/LABA) and triple therapies (ICS/LABA/LAMA) compared with each other and with varying doses of ICS in adolescents and adults with uncontrolled asthma.

SEARCH METHODS

We searched multiple databases for pre-registered randomised controlled trials (RCTs) of at least 12 weeks of study duration from 2008 to 18 February 2022.

SELECTION CRITERIA

We searched studies, including adolescents and adults with uncontrolled asthma who had been treated with, or were eligible for, MD-ICS/LABA, comparing dual and triple therapies. We excluded cluster- and cross-over RCTs.

DATA COLLECTION AND ANALYSIS

We conducted a systematic review and network meta-analysis according to the previously published protocol. We used Cochrane's Screen4ME workflow to assess search results and Grading of Recommendations Assessment, Development and Evaluation (GRADE) to assess the certainty of evidence. The primary outcome was steroid-requiring asthma exacerbations and asthma-related hospitalisations (moderate to severe and severe exacerbations).

MAIN RESULTS

We included 17,161 patients with uncontrolled asthma from 17 studies (median duration 26 weeks; mean age 49.1 years; male 40%; white 81%; mean forced expiratory volume in 1 second (MEF 1)1.9 litres and 61% predicted). The quality of included studies was generally good except for some outcomes in a few studies due to high attrition rates. Medium-dose (MD) and high-dose (HD) triple therapies reduce steroid-requiring asthma exacerbations (hazard ratio (HR) 0.84 [95% credible interval (CrI) 0.71 to 0.99] and 0.69 [0.58 to 0.82], respectively) (high-certainty evidence), but not asthma-related hospitalisations, compared to MD-ICS/LABA. High-dose triple therapy likely reduces steroid-requiring asthma exacerbations compared to MD triple therapy (HR 0.83 [95% CrI 0.69 to 0.996], [moderate certainty]). Subgroup analyses suggest the reduction in steroid-requiring exacerbations associated with triple therapies may be only for those with a history of asthma exacerbations in the previous year but not for those without. High-dose triple therapy, but not MD triple, results in a reduction in all-cause adverse events (AEs) and likely reduces dropouts due to AEs compared to MD-ICS/LABA (odds ratio (OR) 0.79 [95% CrI 0.69 to 0.90], [high certainty] and 0.50 [95% CrI 0.30 to 0.84], [moderate certainty], respectively). Triple therapy results in little to no difference in all-cause or asthma-related serious adverse events (SAEs) compared to dual therapy (high certainty). The evidence suggests triple therapy results in little or no clinically important difference in symptoms or quality of life compared to dual therapy considering the minimal clinically important differences (MCIDs) and HD-ICS/LABA is unlikely to result in any significant benefit or harm compared to MD-ICS/LABA.

AUTHORS' CONCLUSIONS

Medium-dose and HD triple therapies reduce steroid-requiring asthma exacerbations, but not asthma-related hospitalisations, compared to MD-ICS/LABA especially in those with a history of asthma exacerbations in the previous year. High-dose triple therapy is likely superior to MD triple therapy in reducing steroid-requiring asthma exacerbations. Triple therapy is unlikely to result in clinically meaningful improvement in symptoms or quality of life compared to dual therapy considering the MCIDs. High-dose triple therapy, but not MD triple, results in a reduction in all-cause AEs and likely reduces dropouts due to AEs compared to MD-ICS/LABA. Triple therapy results in little to no difference in all-cause or asthma-related SAEs compared to dual therapy. HD-ICS/LABA is unlikely to result in any significant benefit or harm compared to MD-ICS/LABA, although long-term safety of higher rather than MD- ICS remains to be demonstrated given the median duration of included studies was six months. The above findings may assist deciding on a treatment option when asthma is not controlled with MD-ICS/LABA.

Development and validation of paired MEDLINE and Embase search filters for cost-utility studies

BACKGROUND

Search filters are standardised sets of search terms, with validated performance, that are designed to retrieve studies with specific characteristics. A cost-utility analysis (CUA) is the preferred type of economic evaluation to underpin decision-making at the National Institute for Health and Care Excellence (NICE). Until now, when searching for economic evidence for NICE guidelines, we have used a broad set of health economic-related search terms, even when the reviewer's interest is confined to CUAs alone.

METHODS

We developed search filters to retrieve CUAs from MEDLINE and Embase. Our aim was to achieve recall of 90% or better across both databases while reducing the overall yield compared with our existing broad economic filter. We used the relative recall method along with topic expert input to derive and validate 3 pairs of filters, assessed by their ability to identify a gold-standard set of CUAs that had been used in published NICE guidelines. We developed and validated MEDLINE and Embase filters in pairs (testing whether, when used together, they find target studies in at least 1 database), as this is how they are used in practice. We examined the proxy-precision of our new filters by comparing their overall yield with our previous approach using publications indexed in a randomly selected year (2010).

RESULTS

All 3 filter-pairs exceeded our target recall and led to substantial improvements in search proxy-precision. Our paired 'sensitive' filters achieved 100% recall (95% CI 99.0 to 100%) in the validation set. Our paired 'precise' filters also had very good recall (97.6% [95%CI: 95.4 to 98.9%]). We estimate that, compared with our previous search strategy, using the paired 'sensitive' filters would reduce reviewer screening burden by a factor of 5 and the 'precise' versions would do so by a factor of more than 20.

CONCLUSIONS

Each of the 3 paired cost-utility filters enable the identification of almost all CUAs from MEDLINE and Embase from the validation set, with substantial savings in screening workload compared to our previous search practice. We would encourage other researchers who regularly use multiple databases to consider validating search filters in combination as this will better reflect how they use databases in their everyday work.

Increasing comprehensiveness and reducing workload in a systematic review of complex interventions using automated machine learning

BACKGROUND

As part of our ongoing systematic review of complex interventions for the primary prevention of cardiovascular diseases, we have developed and evaluated automated machine-learning classifiers for title and abstract screening. The aim was to develop a high-performing algorithm comparable to human screening.

METHODS

We followed a three-phase process to develop and test an automated machine learning-based classifier for screening potential studies on interventions for primary prevention of cardiovascular disease. We labelled a total of 16,611 articles during the first phase of the project. In the second phase, we used the labelled articles to develop a machine learning-based classifier. After that, we examined the performance of the classifiers in correctly labelling the papers. We evaluated the performance of the five deep-learning models [i.e. parallel convolutional neural network ( CNN ), stacked CNN , parallel-stacked CNN , recurrent neural network ( RNN ) and CNN-RNN]. The models were evaluated using recall, precision and work saved over sampling at no less than 95% recall.

RESULTS

We labelled a total of 16,611 articles, of which 676 (4.0%) were tagged as 'relevant' and 15,935 (96%) were tagged as 'irrelevant'. The recall ranged from 51.9% to 96.6%. The precision ranged from 64.6% to 99.1%. The work saved over sampling ranged from 8.9% to as high as 92.1%. The best-performing model was parallel CNN , yielding a 96.4% recall, as well as 99.1% precision, and a potential workload reduction of 89.9%.

FUTURE WORK AND LIMITATIONS

We used words from the title and the abstract only. More work needs to be done to look into possible changes in performance, such as adding features such as full document text. The approach might also not be able to be used for other complex systematic reviews on different topics.

CONCLUSION

Our study shows that machine learning has the potential to significantly aid the labour-intensive screening of abstracts in systematic reviews of complex interventions. Future research should concentrate on enhancing the classifier system and determining how it can be integrated into the systematic review workflow.

FUNDING

This project was funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme and will be published in Health Technology Assessment. See the NIHR Journals Library website for further project information.

High versus low blood pressure targets for cardiac surgery while on cardiopulmonary bypass

BACKGROUND

Cardiac surgery is performed worldwide. Most types of cardiac surgery are performed using cardiopulmonary bypass (CPB). Cardiac surgery performed with CPB is associated with morbidities. CPB needs an extracorporeal circulation that replaces the heart and lungs, and performs circulation, ventilation, and oxygenation of the blood. The lower limit of mean blood pressure to maintain blood flow to vital organs increases in people with chronic hypertension. Because people undergoing cardiac surgery commonly have chronic hypertension, we hypothesised that maintaining a relatively high blood pressure improves desirable outcomes among the people undergoing cardiac surgery with CPB.

OBJECTIVES

To evaluate the benefits and harms of higher versus lower blood pressure targets during cardiac surgery with CPB.

SEARCH METHODS

We used standard, extensive Cochrane search methods. The latest search of databases was November 2021 and trials registries in January 2020.

SELECTION CRITERIA

We included randomised controlled trials (RCTs) comparing a higher blood pressure target (mean arterial pressure 65 mmHg or greater) with a lower blood pressure target (mean arterial pressure less than 65 mmHg) in adults undergoing cardiac surgery with CPB.

DATA COLLECTION AND ANALYSIS

We used standard Cochrane methods. Primary outcomes were 1. acute kidney injury, 2. cognitive deterioration, and 3. all-cause mortality. Secondary outcomes were 4. quality of life, 5. acute ischaemic stroke, 6. haemorrhagic stroke, 7. length of hospital stay, 8. renal replacement therapy, 9. delirium, 10. perioperative transfusion of blood products, and 11. perioperative myocardial infarction. We used GRADE to assess certainty of evidence.

MAIN RESULTS

We included three RCTs with 737 people compared a higher blood pressure target with a lower blood pressure target during cardiac surgery with CPB. A high blood pressure target may result in little to no difference in acute kidney injury (risk ratio (RR) 1.30, 95% confidence interval (CI) 0.81 to 2.08; I² = 72%; 2 studies, 487 participants; low-certainty evidence), cognitive deterioration (RR 0.82, 95% CI 0.45 to 1.50; I² = 0%; 2 studies, 389 participants; low-certainty evidence), and all-cause mortality (RR 1.33, 95% CI 0.30 to 5.90; I² = 49%; 3 studies, 737 participants; low-certainty evidence). No study reported haemorrhagic stroke. Although a high blood pressure target may increase the length of hospital stay slightly, we found no differences between a higher and a lower blood pressure target for the other secondary outcomes. We also identified one ongoing RCT which is comparing a higher versus a lower blood pressure target among the people who undergo cardiac surgery with CPB.

AUTHORS' CONCLUSIONS

A high blood pressure target may result in little to no difference in patient outcomes including acute kidney injury and mortality. Given the wide CIs, further studies are needed to confirm the efficacy of a higher blood pressure target among those who undergo cardiac surgery with CPB.

Pharmacological interventions for pain and sedation management in newborn infants undergoing therapeutic hypothermia

BACKGROUND

Newborn infants affected by hypoxic-ischemic encephalopathy (HIE) undergo therapeutic hypothermia. As this treatment seems to be associated with pain, and intensive and invasive care is needed, pharmacological interventions are often used. Moreover, painful procedures in the newborn period can affect pain responses later in life, impair brain development, and possibly have a long-term negative impact on neurodevelopment and quality of life.

OBJECTIVES

To determine the effects of pharmacological interventions for pain and sedation management in newborn infants undergoing therapeutic hypothermia. Primary outcomes were analgesia and sedation, and all-cause mortality to discharge.

SEARCH METHODS

We searched CENTRAL, PubMed, CINAHL (Cumulative Index to Nursing and Allied Health Literature), and the trial register ISRCTN in August 2021. We also checked the reference lists of relevant articles to identify additional studies.

SELECTION CRITERIA

We included randomized controlled trials (RCT), quasi-RCTs and cluster-randomized trials comparing drugs used for the management of pain or sedation, or both, during therapeutic hypothermia: any opioids (e.g. morphine, fentanyl), alpha-2 agonists (e.g. clonidine, dexmedetomidine), N-Methyl-D-aspartate (NMDA) receptor antagonist (e.g. ketamine), other analgesics (e.g. paracetamol), and sedatives (e.g. benzodiazepines such as midazolam) versus another drug, placebo, no intervention, or non-pharmacological interventions.  Primary outcomes were analgesia and sedation, and all-cause mortality to discharge.

DATA COLLECTION AND ANALYSIS

Two review authors independently assessed studies identified by the search strategy for inclusion. We planned to use the GRADE approach to assess the certainty of evidence. We planned to assess the methodological quality of included trials using Cochrane Effective Practice and Organisation of Care Group (EPOC) criteria (assessing randomization, blinding, loss to follow-up, and handling of outcome data). We planned to evaluate treatment effects using a fixed-effect model with risk ratio (RR) for categorical data and mean, standard deviation (SD), and mean difference (MD) for continuous data.  MAIN RESULTS: We did not find any completed studies for inclusion. Amongst the four excluded studies, topiramate and atropine were used in two and one trial, respectively; one study used dexmedetomidine and was initially reported in 2019 to be a randomized trial. However, it was an observational study (correction in 2021). We identified one ongoing study comparing dexmedetomidine to morphine.

AUTHORS' CONCLUSIONS

We found no studies that met our inclusion criteria and hence there is no evidence to recommend or refute the use of pharmacological interventions for pain and sedation management in newborn infants undergoing therapeutic hypothermia.

Machine learning algorithms to identify cluster randomized trials from MEDLINE and EMBASE

BACKGROUND

Cluster randomized trials (CRTs) are becoming an increasingly important design. However, authors of CRTs do not always adhere to requirements to explicitly identify the design as cluster randomized in titles and abstracts, making retrieval from bibliographic databases difficult. Machine learning algorithms may improve their identification and retrieval. Therefore, we aimed to develop machine learning algorithms that accurately determine whether a bibliographic citation is a CRT report.

METHODS

We trained, internally validated, and externally validated two convolutional neural networks and one support vector machine (SVM) algorithm to predict whether a citation is a CRT report or not. We exclusively used the information in an article citation, including the title, abstract, keywords, and subject headings. The algorithms' output was a probability from 0 to 1. We assessed algorithm performance using the area under the receiver operating characteristic (AUC) curves. Each algorithm's performance was evaluated individually and together as an ensemble. We randomly selected 5000 from 87,633 citations to train and internally validate our algorithms. Of the 5000 selected citations, 589 (12%) were confirmed CRT reports. We then externally validated our algorithms on an independent set of 1916 randomized trial citations, with 665 (35%) confirmed CRT reports.

RESULTS

In internal validation, the ensemble algorithm discriminated best for identifying CRT reports with an AUC of 98.6% (95% confidence interval: 97.8%, 99.4%), sensitivity of 97.7% (94.3%, 100%), and specificity of 85.0% (81.8%, 88.1%). In external validation, the ensemble algorithm had an AUC of 97.8% (97.0%, 98.5%), sensitivity of 97.6% (96.4%, 98.6%), and specificity of 78.2% (75.9%, 80.4%)). All three individual algorithms performed well, but less so than the ensemble.

CONCLUSIONS

We successfully developed high-performance algorithms that identified whether a citation was a CRT report with high sensitivity and moderately high specificity. We provide open-source software to facilitate the use of our algorithms in practice.

Biologics for chronic severe asthma: a network meta‐analysis

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To evaluate the benefits and harms of biological agents targeting type‐2 inflammation (benralizumab, dupilumab, mepolizumab, omalizumab, reslizumab, tezepelumab) in people with severe asthma, with a network meta‐analysis and to rank agents by effectiveness.

Consistency of covid-19 trial preprints with published reports and impact for decision making: retrospective review

Objective

To assess the trustworthiness (ie, complete and consistent reporting of key methods and results between preprint and published trial reports) and impact (ie, effects of preprints on meta-analytic estimates and the certainty of evidence) of preprint trial reports during the covid-19 pandemic.

Design

Retrospective review.

Data sources

World Health Organization covid-19 database and the Living Overview of the Evidence (L-OVE) covid-19 platform by the Epistemonikos Foundation (up to 3 August 2021).

Main outcome measures

Comparison of characteristics of covid-19 trials with and without preprints, estimates of time to publication of covid-19 preprints, and description of differences in reporting of key methods and results between preprints and their later publications. For the effects of eight treatments on mortality and mechanical ventilation, the study comprised meta-analyses including preprints and excluding preprints at one, three, and six months after the first trial addressing the treatment became available either as a preprint or publication (120 meta-analyses in total, 60 of which included preprints and 60 of which excluded preprints) and assessed the certainty of evidence using the GRADE framework.

Results

Of 356 trials included in the study, 101 were only available as preprints, 181 as journal publications, and 74 as preprints first and subsequently published in journals. The median time to publication of preprints was about six months. Key methods and results showed few important differences between trial preprints and their subsequent published reports. Apart from two (3.3%) of 60 comparisons, point estimates were consistent between meta-analyses including preprints versus those excluding preprints as to whether they indicated benefit, no appreciable effect, or harm. For nine (15%) of 60 comparisons, the rating of the certainty of evidence was different when preprints were included versus being excluded-the certainty of evidence including preprints was higher in four comparisons and lower in five comparisons.

Conclusion

No compelling evidence indicates that preprints provide results that are inconsistent with published papers. Preprints remain the only source of findings of many trials for several months-an unsuitable length of time in a health emergency that is not conducive to treating patients with timely evidence. The inclusion of preprints could affect the results of meta-analyses and the certainty of evidence. Evidence users should be encouraged to consider data from preprints.

Increased versus stable doses of inhaled corticosteroids for exacerbations of chronic asthma in adults and children

BACKGROUND

People with asthma may experience exacerbations, or 'attacks', during which their symptoms worsen and additional treatment is required. Written action plans sometimes advocate a short-term increase in the dose of inhaled corticosteroids (ICS) at the first sign of an exacerbation to reduce the severity of the attack and to prevent the need for oral steroids or hospital admission.

OBJECTIVES

To compare the clinical effectiveness and safety of increased versus stable doses of ICS as part of a patient-initiated action plan for the home management of exacerbations in children and adults with persistent asthma.

SEARCH METHODS

We searched the Cochrane Airways Group Specialised Register, which is derived from searches of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and CINAHL (Cumulative Index to Nursing and Allied Health Literature), and handsearched abstracts to 20 December 2021. We also searched major trial registries for ongoing trials.

SELECTION CRITERIA

We included parallel and cross-over randomised controlled trials (RCTs) that allocated people with persistent asthma to take a blinded inhaler in the event of an exacerbation which either increased their daily dose of ICS or kept it stable (placebo).

DATA COLLECTION AND ANALYSIS

Two review authors independently selected trials, assessed quality, and extracted data. We reassessed risk of bias for all studies at the result level using the revised risk of bias tool for RCTs (Risk of Bias 2), and employed the GRADE approach to assess our confidence in the synthesised effect estimates. The primary outcome was treatment failure, defined as the need for rescue oral steroids in the randomised population. Secondary outcomes were treatment failure in the subset who initiated the study inhaler (treated population), unscheduled physician visits, unscheduled acute care, emergency department or hospital visits, serious and non-serious adverse events, and duration of exacerbation.

MAIN RESULTS

This review update added a new study that increased the number of people in the primary analysis from 1520 to 1774, and incorporates the most up-to-date methods to assess the likely impact of bias within the meta-analyses. The updated review now includes nine RCTs (1923 participants; seven parallel and two cross-over) conducted in Europe, North America, and Australasia and published between 1998 and 2018. Five studies evaluated adult populations (n = 1247; ≥ 15 years), and four studies evaluated child or adolescent populations (n = 676; < 15 years). All study participants had mild to moderate asthma. Studies varied in the dose of maintenance ICS, age, fold increase of ICS in the event of an exacerbation, criteria for initiating the study inhaler, and allowed medications. Approximately 50% of randomised participants initiated the study inhaler (range 23% to 100%), and the included studies reported treatment failure in a variety of ways, meaning assumptions were required to permit the combining of data. Participants randomised to increase their ICS dose at the first signs of an exacerbation had similar odds of needing rescue oral corticosteroids to those randomised to a placebo inhaler (odds ratio (OR) 0.97, 95% confidence interval (CI) 0.76 to 1.25; 8 studies; 1774 participants; I2 = 0%; moderate quality evidence). We could draw no firm conclusions from subgroup analyses conducted to investigate the impact of age, time to treatment initiation, baseline dose, smoking history, and fold increase of ICS on the primary outcome. Results for the same outcome in the subset of participants who initiated the study inhaler were unchanged from the previous version, which provides a different point estimate with very low confidence due to heterogeneity, imprecision, and risk of bias (OR 0.84, 95% CI 0.54 to 1.30; 7 studies; 766 participants; I2 = 42%; random-effects model). Confidence was reduced due to risk of bias and assumptions that had to be made to include study data in the intention-to-treat and treated-population analyses. Sensitivity analyses that tested the impact of assumptions made for synthesis and to exclude cross-over studies, studies at overall high risk of bias, and those with commercial funding did not change our conclusions. Pooled effects for unscheduled physician visits, unscheduled acute care, emergency department or hospital visits, and duration of exacerbation made it very difficult to determine where the true effect may lie, and confidence was reduced by risk of bias. Point estimates for both serious and non-serious adverse events favoured keeping ICS stable, but imprecision and risk of bias due to missing data and outcome measurement and reporting reduced our confidence in the effects (serious adverse events: OR 1.69, 95% CI 0.77 to 3.71; 2 studies; 394 participants; I² = 0%; non-serious adverse events: OR 2.15, 95% CI 0.68 to 6.73; 2 studies; 142 participants; I² = 0%).

AUTHORS' CONCLUSIONS

Evidence from double-blind trials of adults and children with mild to moderate asthma suggests there is unlikely to be an important reduction in the need for oral steroids from increasing a patient's ICS dose at the first sign of an exacerbation. Other clinically important benefits and potential harms of increased doses of ICS compared with keeping the dose stable cannot be ruled out due to wide confidence intervals, risk of bias in the trials, and assumptions that had to be made for synthesis. Included studies conducted between 1998 and 2018 reflect evolving clinical practice and study methods, and the data do not support thorough investigation of effect modifiers such as baseline dose, fold increase, asthma severity and timing. The review does not include recent evidence from pragmatic, unblinded studies showing benefits of larger dose increases in those with poorly controlled asthma. A systematic review is warranted to examine the differences between the blinded and unblinded trials using robust methods for assessing risk of bias to present the most complete view of the evidence for decision makers.

Do physical activity interventions combining self-monitoring with other components provide an additional benefit compared with self-monitoring alone? A systematic review and meta-analysis

Objective

To determine the net effect of different physical activity intervention components on step counts in addition to self-monitoring.

Design

A systematic review with meta-analysis and meta-regression.

Data sources

Five databases (PubMed, Scopus, Web of Science, ProQuest and Discus) were searched from inception to May 2022. The database search was complemented with backward and forward citation searches and search of the references from relevant systematic reviews.

Eligibility criteria

Randomised controlled trials comparing an intervention using self-monitoring (active control arm) with an intervention comprising the same treatment PLUS any additional component (intervention arm).

Data extraction and synthesis

The effect measures were mean differences in daily step count. Meta-analyses were performed using random-effects models, and effect moderators were explored using univariate and multivariate meta-regression models.

Results

Eighty-five studies with 12 057 participants were identified, with 75 studies included in the meta-analysis at postintervention and 24 at follow-up. At postintervention, the mean difference between the intervention and active control arms was 926 steps/day (95% CI 651 to 1201). At a follow-up, the mean difference was 413 steps/day (95% CI 210 to 615). Interventions with a prescribed goal and involving human counselling, particularly via phone/video calls, were associated with a greater mean difference in the daily step count than interventions with added print materials, websites, smartphone apps or incentives.

Conclusion

Physical activity interventions that combine self-monitoring with other components provide an additional modest yet sustained increase in step count compared with self-monitoring alone. Some forms of counselling, particularly remote phone/video counselling, outperformed other intervention components, such as websites and smartphone apps.

PROSPERO registered number

CRD42020199482.

Interventions for the treatment of persistent post-COVID-19 olfactory dysfunction

BACKGROUND

Olfactory dysfunction is a common consequence of COVID-19 infection and persistent symptoms can have a profound impact on quality of life. At present there is little guidance on how best to treat this condition. A variety of interventions have been suggested to promote recovery, including medication and olfactory training. However, it is uncertain whether any intervention is of benefit. This is an update of the 2021 review with one additional study added.  OBJECTIVES: 1) To evaluate the benefits and harms of any intervention versus no treatment for people with persisting olfactory dysfunction due to COVID-19 infection.  2) To keep the evidence up-to-date, using a living systematic review approach.  SEARCH METHODS: The Cochrane ENT Information Specialist searched the Cochrane ENT Register; Central Register of Controlled Trials (CENTRAL); Ovid MEDLINE; Ovid Embase; Web of Science; ClinicalTrials.gov; ICTRP and additional sources for published and unpublished trials. The date of the latest search was 20 October 2021.   SELECTION CRITERIA: We included randomised controlled trials (RCTs) in people with COVID-19 related olfactory disturbance that had persisted for at least four weeks. We included any intervention compared to no treatment or placebo.  DATA COLLECTION AND ANALYSIS: We used standard Cochrane methods. Our primary outcomes were the recovery of sense of smell, disease-related quality of life and serious adverse effects. Secondary outcomes were the change in sense of smell, general quality of life, prevalence of parosmia and other adverse effects (including nosebleeds/bloody discharge). We used GRADE to assess the certainty of the evidence for each outcome.

MAIN RESULTS

We included two studies with 30 participants. The studies evaluated the following interventions: systemic corticosteroids plus intranasal corticosteroid/mucolytic/decongestant and palmitoylethanolamide plus luteolin.  Systemic corticosteroids plus intranasal corticosteroid/mucolytic/decongestant compared to no intervention We included a single RCT with 18 participants who had anosmia for at least 30 days following COVID-19 infection. Participants received a 15-day course of oral corticosteroids combined with nasal irrigation (consisting of an intranasal corticosteroid/mucolytic/decongestant solution) or no intervention. Psychophysical testing was used to assess olfactory function at 40 days. This is a single, small study and for all outcomes the certainty of evidence was very low. We are unable to draw meaningful conclusions from the numerical results. Palmitoylethanolamide plus luteolin compared to no intervention We included a single RCT with 12 participants who had anosmia or hyposmia for at least 90 days following COVID-19 infection. Participants received a 30-day course of palmitoylethanolamide and luteolin or no intervention. Psychophysical testing was used to assess olfactory function at 30 days. This is a single, small study and for all outcomes the certainty of evidence was very low. We are unable to draw meaningful conclusions from the numerical results.

AUTHORS' CONCLUSIONS

There is very limited evidence available on the efficacy and harms of treatments for persistent olfactory dysfunction following COVID-19 infection. However, we have identified a number of ongoing trials in this area. As this is a living systematic review we will update the data regularly, as new results become available.

Interventions for the prevention of persistent post-COVID-19 olfactory dysfunction

BACKGROUND

Loss of olfactory function is well recognised as a symptom of COVID-19 infection, and the pandemic has resulted in a large number of individuals with abnormalities in their sense of smell. For many, the condition is temporary and resolves within two to four weeks. However, in a significant minority the symptoms persist. At present, it is not known whether early intervention with any form of treatment (such as medication or olfactory training) can promote recovery and prevent persisting olfactory disturbance. This is an update of the 2021 review with four studies added.

OBJECTIVES

1) To evaluate the benefits and harms of any intervention versus no treatment for people with acute olfactory dysfunction due to COVID-19 infection.  2) To keep the evidence up-to-date, using a living systematic review approach.  SEARCH METHODS: The Cochrane ENT Information Specialist searched the Cochrane ENT Register; Central Register of Controlled Trials (CENTRAL); Ovid MEDLINE; Ovid Embase; Web of Science; ClinicalTrials.gov; ICTRP and additional sources for published and unpublished trials. The date of the latest search was 20 October 2021.

SELECTION CRITERIA

We included randomised controlled trials (RCTs) in people with COVID-19 related olfactory disturbance, which had been present for less than four weeks. We included any intervention compared to no treatment or placebo.  DATA COLLECTION AND ANALYSIS: We used standard Cochrane methods. Our primary outcomes were the presence of normal olfactory function, serious adverse effects and change in sense of smell. Secondary outcomes were the prevalence of parosmia, change in sense of taste, disease-related quality of life and other adverse effects (including nosebleeds/bloody discharge). We used GRADE to assess the certainty of the evidence for each outcome.  MAIN RESULTS: We included five studies with 691 participants. The studies evaluated the following interventions: intranasal corticosteroid sprays, intranasal corticosteroid drops, intranasal hypertonic saline and zinc sulphate.  Intranasal corticosteroid spray compared to no intervention/placebo We included three studies with 288 participants who had olfactory dysfunction for less than four weeks following COVID-19. Presence of normal olfactory function The evidence is very uncertain about the effect of intranasal corticosteroid spray on both self-rated recovery of olfactory function and recovery of olfactory function using psychophysical tests at up to four weeks follow-up (self-rated: risk ratio (RR) 1.19, 95% confidence interval (CI) 0.85 to 1.68; 1 study; 100 participants; psychophysical testing: RR 2.3, 95% CI 1.16 to 4.63; 1 study; 77 participants; very low-certainty evidence).  Change in sense of smell The evidence is also very uncertain about the effect of intranasal corticosteroid spray on self-rated change in the sense of smell (at less than 4 weeks: mean difference (MD) 0.5 points lower, 95% CI 1.38 lower to 0.38 higher; 1 study; 77 participants; at > 4 weeks to 3 months: MD 2.4 points higher, 95% CI 1.32 higher to 3.48 higher; 1 study; 100 participants; very low-certainty evidence, rated on a scale of 1 to 10, higher scores mean better olfactory function). Intranasal corticosteroids may make little or no difference to the change in sense of smell when assessed with psychophysical testing (MD 0.2 points, 95% CI 2.06 points lower to 2.06 points higher; 1 study; 77 participants; low-certainty evidence, 0- to 24-point scale, higher scores mean better olfactory function).  Serious adverse effects The authors of one study reported no adverse effects, but their intention to collect these data was not pre-specified so we are uncertain if these were systematically sought and identified. The remaining two studies did not report on adverse effects.  Intranasal corticosteroid drops compared to no intervention/placebo We included one study with 248 participants who had olfactory dysfunction for ≤ 15 days following COVID-19. Presence of normal olfactory function Intranasal corticosteroid drops may make little or no difference to self-rated recovery at > 4 weeks to 3 months (RR 1.00, 95% CI 0.89 to 1.11; 1 study; 248 participants; low-certainty evidence). No other outcomes were assessed by this study.  Data on the use of hypertonic saline nasal irrigation and the use of zinc sulphate to prevent persistent olfactory dysfunction are included in the full text of the review.

AUTHORS' CONCLUSIONS

There is very limited evidence available on the efficacy and harms of treatments for preventing persistent olfactory dysfunction following COVID-19 infection. However, we have identified a number of ongoing trials in this area. As this is a living systematic review we will update the data regularly, as new results become available.

Insertion site of central venous catheter among hospitalized adult patients: A systematic review and network meta-analysis

Introduction

Central venous catheterization is a commonly performed procedure, accounting for approximately 8% of hospitalized patients. Based on the current literatures, the most acceptable site for central venous catheterization is inconclusive, considering various complications in hospitalized patients. Herein, we conducted a network meta-analysis to assess the clinically important complications among internal jugular, subclavian, femoral, and peripheral insertion.

Materials and methods

The Cochrane Central Register of Controlled Trials, MEDLINE, Web of Science, Ichushi databases, Clinicaltrials.gov, and International Clinical Trials Registry Platform were searched. Studies including adults aged ≥ 18 years and randomized control trials that compared two different insertion sites (internal jugular, subclavian, femoral, and peripheral vein) were selected. The primary outcomes were clinically important infectious, thrombotic, and mechanical complications.

Results

Among the 5,819 records initially identified, 13 trials (6,201 patients) were included for a network meta-analysis. For clinically important infectious complication, subclavian insertion decreased the complication risk, compared with internal jugular [risk ratio (RR), 0.30; 95% confidence interval (CI), 0.11-0.81; moderate certainty], and femoral insertion increased than subclavian insertion (RR 2.56; 95% CI, 1.02-6.44; moderate certainty). Peripheral insertion was also significantly associated with a lower risk compared with internal jugular (RR 0.06; 95% CI, 0.01-0.32; low certainty); subclavian (RR 0.21; 95% CI, 0.05-0.77; moderate certainty); and femoral insertion (RR 0.08; 95% CI, 0.02-0.40; low certainty). For clinically important thrombotic complication, we did not find significant differences between insertion sites. For clinically important mechanical complication, femoral insertion decreased the complication risk, compared with internal jugular (RR 0.42; 95% CI, 0.21-0.82; moderate certainty) and subclavian insertion (RR 0.33; 95% CI, 0.16-0.66; moderate certainty). Peripheral insertion was also associated with the lower complication risk compared with internal jugular (RR 0.39; 95% CI, 0.18-0.85; low certainty) and subclavian insertion (RR 0.31; 95% CI, 0.13-0.75; moderate certainty).

Conclusion

The insertion site of the central venous catheter, which is most likely to cause the fewest complications, should be selected. Our findings can provide the rationale for deciding the insertion site for a central venous catheter.

Systematic review registration

[www.protocols.io], identifier [61375].

Improvement of intervention information detection for automated clinical literature screening during systematic review

Systematic literature review (SLR) is a crucial method for clinicians and policymakers to make their decisions in a flood of new clinical studies. Because manual literature screening in SLR is a highly laborious task, its automation by natural language processing (NLP) has been welcomed. Although intervention is a key information for literature screening, NLP models for its detection in previous works have not shown adequate performance. In this work, we first design an algorithm for automated construction of high-quality intervention labels by utilizing information retrieved from a clinical trial database. We then design another algorithm for improving model's recall and F1 score by imposing adaptive weights on training instances in the loss function. The intervention detection model trained on the weighted datasets is tested with the Evidence-Based Medicine NLP (EBM-NLP) corpus, and shows 9.7% and 4.0% improvements respectively in recall and F1 score compared to the previous state-of-the-art model on the corpus. The proposed algorithms can boost automation of literature screening during SLR in the clinical domain.

Pulmonary rehabilitation versus usual care for adults with asthma

BACKGROUND

Asthma is a respiratory disease characterised by variable airflow limitation and the presence of respiratory symptoms including wheeze, chest tightness, cough and/or dyspnoea. Exercise training is beneficial for people with asthma; however, the response to conventional models of pulmonary rehabilitation is less clear.

OBJECTIVES

To evaluate, in adults with asthma, the effectiveness of pulmonary rehabilitation compared to usual care on exercise performance, asthma control, and quality of life (co-primary outcomes), incidence of severe asthma exacerbations/hospitalisations, mental health, muscle strength, physical activity levels, inflammatory biomarkers, and adverse events.

SEARCH METHODS

We identified studies from the Cochrane Airways Trials Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform, from their inception to May 2021, as well as the reference lists of all primary studies and review articles.

SELECTION CRITERIA

We included randomised controlled trials in which pulmonary rehabilitation was compared to usual care in adults with asthma. Pulmonary rehabilitation must have included a minimum of four weeks (or eight sessions) aerobic training and education or self-management. Co-interventions were permitted; however, exercise training alone was not.  DATA COLLECTION AND ANALYSIS: Following the use of Cochrane's Screen4Me workflow, two review authors independently screened and selected trials for inclusion, extracted study characteristics and outcome data, and assessed risk of bias using the Cochrane risk of bias tool. We contacted study authors to retrieve missing data. We calculated between-group effects via mean differences (MD) or standardised mean differences (SMD) using a random-effects model. We evaluated the certainty of evidence using GRADE methodology.

MAIN RESULTS

We included 10 studies involving 894 participants (range 24 to 412 participants (n = 2 studies involving n > 100, one contributing to meta-analysis), mean age range 27 to 54 years). We identified one ongoing study and three studies awaiting classification. One study was synthesised narratively, and another involved participants specifically with asthma-COPD overlap. Most programmes were outpatient-based, lasting from three to four weeks (inpatient) or eight to 12 weeks (outpatient). Education or self-management components included breathing retraining and relaxation, nutritional advice and psychological counselling. One programme was specifically tailored for people with severe asthma.  Pulmonary rehabilitation compared to usual care may increase maximal oxygen uptake (VO2 max) after programme completion, but the evidence is very uncertain for data derived using mL/kg/min (MD between groups of 3.63 mL/kg/min, 95% confidence interval (CI) 1.48 to 5.77; 3 studies; n = 129) and uncertain for data derived from % predicted VO2 max (MD 14.88%, 95% CI 9.66 to 20.1%; 2 studies; n = 60). The evidence is very uncertain about the effects of pulmonary rehabilitation compared to usual care on incremental shuttle walk test distance (MD between groups 74.0 metres, 95% CI 26.4 to 121.4; 1 study; n = 30). Pulmonary rehabilitation may have little to no effect on VO2 max at longer-term follow up (9 to 12 months), but the evidence is very uncertain (MD -0.69 mL/kg/min, 95% CI -4.79 to 3.42; I2 = 49%; 3 studies; n = 66). Pulmonary rehabilitation likely improves functional exercise capacity as measured by 6-minute walk distance, with MD between groups after programme completion of 79.8 metres (95% CI 66.5 to 93.1; 5 studies; n = 529; moderate certainty evidence). This magnitude of mean change exceeds the minimally clinically important difference (MCID) threshold for people with chronic respiratory disease. The evidence is very uncertain about the longer-term effects one year after pulmonary rehabilitation for this outcome (MD 52.29 metres, 95% CI 0.7 to 103.88; 2 studies; n = 42). Pulmonary rehabilitation may result in a small improvement in asthma control compared to usual care as measured by Asthma Control Questionnaire (ACQ), with an MD between groups of -0.46 (95% CI -0.76 to -0.17; 2 studies; n = 93; low certainty evidence); however, data derived from the Asthma Control Test were very uncertain (MD between groups 3.34, 95% CI -2.32 to 9.01; 2 studies; n = 442). The ACQ finding approximates the MCID of 0.5 points. Pulmonary rehabilitation results in little to no difference in asthma control as measured by ACQ at nine to 12 months follow-up (MD 0.09, 95% CI -0.35 to 0.53; 2 studies; n = 48; low certainty evidence). Pulmonary rehabilitation likely results in a large improvement in quality of life as assessed by the St George's Respiratory Questionnaire (SGRQ) total score (MD -18.51, 95% CI -20.77 to -16.25; 2 studies; n = 440; moderate certainty evidence), with this magnitude of change exceeding the MCID. However, pulmonary rehabilitation may have little to no effect on Asthma Quality of Life Questionnaire (AQLQ) total scores, with the evidence being very uncertain (MD 0.87, 95% CI -0.13 to 1.86; 2 studies; n = 442). Longer-term follow-up data suggested improvements in quality of life may occur as measured by SGRQ (MD -13.4, 95% CI -15.93 to -10.88; 2 studies; n = 430) but not AQLQ (MD 0.58, 95% CI -0.23 to 1.38; 2 studies; n = 435); however, the evidence is very uncertain. One study reported no difference between groups in the proportion of participants who experienced an asthma exacerbation during the intervention period. Data from one study suggest adverse events attributable to the intervention are rare.  Overall risk of bias was most commonly impacted by performance bias attributed to a lack of participant blinding to knowledge of the intervention. This is inherently challenging to overcome in rehabilitation studies.  AUTHORS' CONCLUSIONS: Moderate certainty evidence shows that pulmonary rehabilitation is probably associated with clinically meaningful improvements in functional exercise capacity and quality of life upon programme completion in adults with asthma. The certainty of evidence relating to maximal exercise capacity was very low to low. Pulmonary rehabilitation appears to confer minimal effect on asthma control, although the certainty of evidence is very low to low. Unclear reporting of study methods and small sample sizes limits our certainty in the overall body of evidence, whilst heterogenous study designs and interventions likely contribute to inconsistent findings across clinical outcomes and studies. There remains considerable scope for future research.

A Systematic Review and Meta-Analysis of the Success of Resin Composite Restorations

Aim:

This review examines the studies on resin composites within the past decade and evaluates the materials’ application methods and success through Fédération Dentaire Internationale (FDI) scores.

Materials and Methods:

Search results from three databases (a total of 3295 studies) were transferred to Mendeley Desktop (Mendeley, London, UK) software and deduplicated (2638 studies). Eight studies were included in this meta-analysis after scanning the title, abstract, and full-text screening. Risk of bias analysis was performed using the Cochrane Handbook tool. Shapiro–Wilk, Kolmogorov–Smirnov, Kruskal–Wallis, and post hoc Tamhane’s T2 were used for statistical analysis.

Results:

Conventional composite’s surface staining FDI score was higher than flowable and silorane composites (no difference between these two). Also, conventional composite has the highest score in terms of staining. Conventional composite + two-step self-etch adhesive and bulk fill composite + two-step self-etch adhesive have lower FDI scores in marginal staining. In noncarious cervical lesions (NCCL), conventional composite + two-step total-etch adhesive has a higher score than flowable composite + two-step self-etch adhesive. Flowable composite + two-step self-etch adhesive and conventional composite + single-step self-etch adhesive have the highest FDI scores for marginal adaptation (no statistical difference between them). Conventional composite + two-step self-etch adhesive and bulk fill composite + two-step self-etch have the lowest scores (no statistical difference between them). For recurrence of caries, all groups have Score 1, the best score, and there is no significant difference between them.

Conclusions:

Each restorative material and adhesive system has different advantages, and the practitioner must choose the most appropriate.

Artificial intelligence in COVID-19 evidence syntheses was underutilized, but impactful: a methodological study

OBJECTIVES

A rapidly developing scenario like a pandemic requires the prompt production of high-quality systematic reviews, which can be automated using artificial intelligence (AI) techniques. We evaluated the application of AI tools in COVID-19 evidence syntheses.

STUDY DESIGN

After prospective registration of the review protocol, we automated the download of all open-access COVID-19 systematic reviews in the COVID-19 Living Overview of Evidence database, indexed them for AI-related keywords, and located those that used AI tools. We compared their journals' JCR Impact Factor, citations per month, screening workloads, completion times (from pre-registration to preprint or submission to a journal) and AMSTAR-2 methodology assessments (maximum score 13 points) with a set of publication date matched control reviews without AI.

RESULTS

Of the 3,999 COVID-19 reviews, 28 (0.7%, 95% CI 0.47-1.03%) made use of AI. On average, compared to controls (n = 64), AI reviews were published in journals with higher Impact Factors (median 8.9 vs. 3.5, P < 0.001), and screened more abstracts per author (302.2 vs. 140.3, P = 0.009) and per included study (189.0 vs. 365.8, P < 0.001) while inspecting less full texts per author (5.3 vs. 14.0, P = 0.005). No differences were found in citation counts (0.5 vs. 0.6, P = 0.600), inspected full texts per included study (3.8 vs. 3.4, P = 0.481), completion times (74.0 vs. 123.0, P = 0.205) or AMSTAR-2 (7.5 vs. 6.3, P = 0.119).

CONCLUSION

AI was an underutilized tool in COVID-19 systematic reviews. Its usage, compared to reviews without AI, was associated with more efficient screening of literature and higher publication impact. There is scope for the application of AI in automating systematic reviews.

Topical anti‐inflammatory treatments for eczema: network meta‐analysis

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To compare the efficacy and safety of topical anti‐inflammatory treatments for reducing eczema symptoms or signs or improving eczema‐related quality of life in children and adults with eczema, by undertaking a network meta‐analysis. To provide a clinically useful ranking of these treatments according to their efficacy and safety.

Anti-IL-5 therapies for asthma

BACKGROUND

This is the second update of previously published reviews in the Cochrane Library (2015, first update 2017). Interleukin-5 (IL-5) is the main cytokine involved in the proliferation, maturation, activation and survival of eosinophils, which cause airway inflammation and are a classic feature of asthma. Studies of monoclonal antibodies targeting IL-5 or its receptor (IL-5R) suggest they reduce asthma exacerbations, improve health-related quality of life (HRQoL) and lung function in appropriately selected patients, justifying their inclusion in the latest guidelines.

OBJECTIVES

To compare the effects of therapies targeting IL-5 signalling (anti-IL-5 or anti-IL-5Rα) with placebo on exacerbations, health-related quality-of-life (HRQoL) measures and lung function in adults and children with chronic asthma, and specifically in those with eosinophilic asthma refractory to existing treatments.

SEARCH METHODS

We searched CENTRAL, MEDLINE, Embase, and two trials registers, manufacturers' websites, and reference lists of included studies. The most recent search was 7 February 2022.

SELECTION CRITERIA

We included randomised controlled trials comparing mepolizumab, reslizumab and benralizumab versus placebo in adults and children with asthma.

DATA COLLECTION AND ANALYSIS

Two review authors independently extracted data and analysed outcomes using a random-effects model. We used standard methods expected by Cochrane.

MAIN RESULTS

Seventeen studies on about 7600 participants met the inclusion criteria. Six used mepolizumab, five used reslizumab, and six used benralizumab. One study using benralizumab was terminated early due to sponsor decision and contributed no data. The studies were predominantly on people with severe eosinophilic asthma, which was similarly but variably defined. One was in children aged 6 to 17 years; nine others included children over 12 years but did not report results by age group separately. We deemed the overall risk of bias to be low, with all studies contributing data of robust methodology. We considered the certainty of the evidence for all comparisons to be high overall using the GRADE scheme, except for intravenous (IV) mepolizumab and subcutaneous (SC) reslizumab because these are not currently licensed delivery routes. The anti-IL-5 treatments assessed reduced rates of 'clinically significant' asthma exacerbation (defined by treatment with systemic corticosteroids for three days or more) by approximately half in participants with severe eosinophilic asthma on standard care (at least medium-dose inhaled corticosteroids (ICS)) with poorly controlled disease (either two or more exacerbations in the preceding year or Asthma Control Questionnaire (ACQ) score of 1.5 or more), except for reslizumab SC. The rate ratios for these effects were 0.45 (95% confidence interval (CI) 0.36 to 0.55; high-certainty evidence) for mepolizumab SC, 0.53 (95% CI 0.44 to 0.64; moderate-certainty evidence) for mepolizumab IV, 0.43 (95% CI 0.33 to 0.55; high-certainty evidence) for reslizumab IV, and 0.59 (95% CI 0.52 to 0.66; high-certainty evidence) for benralizumab SC. Non-eosinophilic participants treated with benralizumab also showed a significant reduction in exacerbation rates, an effect not seen with reslizumab IV, albeit in only one study. No data were available for non-eosinophilic participants treated with mepolizumab. There were improvements in validated HRQoL scores with all anti-IL-5 agents in severe eosinophilic asthma. This met the minimum clinically important difference (MCID) for the broader St. George's Respiratory Questionnaire (SGRQ; 4-point change) for benralizumab only, but the improvement in the ACQ and Asthma Quality of Life Questionnaire (AQLQ), which focus on asthma symptoms, fell short of the MCID (0.5 point change for both ACQ and AQLQ) for all of the interventions. The evidence for an improvement in HRQoL scores in non-eosinophilic participants treated with benralizumab and reslizumab was weak, but the tests for subgroup difference were negative. All anti-IL-5 treatments produced small improvements in mean pre-bronchodilator forced expiratory flow in one second (FEV1) of between 0.08 L and 0.15 L in eosinophilic participants, which may not be sufficient to be detected by patients. There were no excess serious adverse events with any anti-IL-5 treatment; in fact, there was a reduction in such events with benralizumab, likely arising from fewer asthma-related hospital admissions. There was no difference compared to placebo in adverse events leading to discontinuation with mepolizumab or reslizumab, but significantly more discontinued benralizumab than placebo, although the absolute numbers were small (42/2026 (2.1%) benralizumab versus 11/1227 (0.9%) placebo). The implications for efficacy or adverse events are unclear.

AUTHORS' CONCLUSIONS

Overall this analysis supports the use of anti-IL-5 treatments as an adjunct to standard care in people with severe eosinophilic asthma and poor symptom control. These treatments roughly halve the rate of asthma exacerbations in this population. There is limited evidence for improved HRQoL scores and lung function, which may not meet clinically detectable levels. The studies did not report safety concerns for mepolizumab or reslizumab, or any excess serious adverse events with benralizumab, although there remains a question over adverse events significant enough to prompt discontinuation. Further research is needed on biomarkers for assessing treatment response, optimal duration and long-term effects of treatment, risk of relapse on withdrawal, non-eosinophilic patients, children (particularly under 12 years), comparing anti-IL-5 treatments to each other and, in patients meeting relevant eligibility criteria, to other biological (monoclonal antibody) therapies. For benralizumab, future studies should closely monitor rates of adverse events prompting discontinuation.

Online randomised trials with children: A scoping review protocol

Introduction: This scoping review will determine how online, randomised trials with children are conducted. The objectives of the review are: (a) to determine what methods and tools have been used to create and conduct online trials with children and (b) to identify the gaps in the knowledge in this field. Over the last decade, randomised trials employing online methods have gained traction. Decentralised methods lend themselves to certain types of trials and can offer advantages over traditional trial methods, potentially increasing participant reach and diversity and decreasing research waste.  However, decentralised trials that have all aspects of the trial exclusively online are not yet common, and those involving children even less so. This scoping review will describe and evaluate the methods used in these trials to understand how they may be effectively employed. Methods: Methods are informed by guidance from the Joanna Briggs Institute and the Preferred Reporting Items for Systematic Reviews and Meta-analyses extension for scoping reviews. The search strategy was developed in consultation with an information specialist for the following databases: MEDLINE, CENTRAL, CINAHL, and Embase. Grey literature searches will be completed with the consultation of experts in decentralised trials and digital health using internet searches and suitable trial registries. Once identified, included full-text studies' references will be manually searched for any trials that may have been missed. We will include randomised and quasi-randomised trials conducted exclusively online with participants under the age of 18 published in English. We will not limit by country of conduct or date of publication. Data will be collected using a data charting tool and presented in text, graphical, and tabular formats. Ethics and Dissemination: Ethical approval is not needed since all data sources used are publicly available. The review will be available as a preprint before publication in an open-access, peer-reviewed journal.

Systemic pharmacological treatments for chronic plaque psoriasis: a network meta-analysis

BACKGROUND

Psoriasis is an immune-mediated disease with either skin or joints manifestations, or both, and it has a major impact on quality of life. Although there is currently no cure for psoriasis, various treatment strategies allow sustained control of disease signs and symptoms. The relative benefit of these treatments remains unclear due to the limited number of trials comparing them directly head-to-head, which is why we chose to conduct a network meta-analysis.

OBJECTIVES

To compare the efficacy and safety of non-biological systemic agents, small molecules, and biologics for people with moderate-to-severe psoriasis using a network meta-analysis, and to provide a ranking of these treatments according to their efficacy and safety.

SEARCH METHODS

For this update of the living systematic review, we updated our searches of the following databases monthly to October 2021: the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and Embase.

SELECTION CRITERIA

Randomised controlled trials (RCTs) of systemic treatments in adults over 18 years with moderate-to-severe plaque psoriasis, at any stage of treatment, compared to placebo or another active agent. The primary outcomes were: proportion of participants who achieved clear or almost clear skin, that is, at least Psoriasis Area and Severity Index (PASI) 90; proportion of participants with serious adverse events (SAEs) at induction phase (8 to 24 weeks after randomisation).

DATA COLLECTION AND ANALYSIS

We conducted duplicate study selection, data extraction, risk of bias assessment and analyses. We synthesised data using pairwise and network meta-analysis (NMA) to compare treatments and rank them according to effectiveness (PASI 90 score) and acceptability (inverse of SAEs). We assessed the certainty of NMA evidence for the two primary outcomes and all comparisons using CINeMA, as very low, low, moderate, or high. We contacted study authors when data were unclear or missing. We used the surface under the cumulative ranking curve (SUCRA) to infer treatment hierarchy, from 0% (worst for effectiveness or safety) to 100% (best for effectiveness or safety).

MAIN RESULTS

This update includes an additional 19 studies, taking the total number of included studies to 167, and randomised participants to 58,912, 67.2% men, mainly recruited from hospitals. Average age was 44.5 years, mean PASI score at baseline was 20.4 (range: 9.5 to 39). Most studies were placebo-controlled (57%). We assessed a total of 20 treatments. Most (140) trials were multicentric (two to 231 centres). One-third of the studies (57/167) had high risk of bias; 23 unclear risk, and most (87) low risk. Most studies (127/167) declared funding by a pharmaceutical company, and 24 studies did not report a funding source. Network meta-analysis at class level showed that all interventions (non-biological systemic agents, small molecules, and biological treatments) showed a higher proportion of patients reaching PASI 90 than placebo. Anti-IL17 treatment showed a higher proportion of patients reaching PASI 90 compared to all the interventions, except anti-IL23. Biologic treatments anti-IL17, anti-IL12/23, anti-IL23 and anti-TNF alpha showed a higher proportion of patients reaching PASI 90 than the non-biological systemic agents. For reaching PASI 90, the most effective drugs when compared to placebo were (SUCRA rank order, all high-certainty evidence): infliximab (risk ratio (RR) 50.19, 95% CI 20.92 to 120.45), bimekizumab (RR 30.27, 95% CI 25.45 to 36.01), ixekizumab (RR 30.19, 95% CI 25.38 to 35.93), risankizumab (RR 28.75, 95% CI 24.03 to 34.39). Clinical effectiveness of these drugs was similar when compared against each other. Bimekizumab, ixekizumab and risankizumab showed a higher proportion of patients reaching PASI 90 than other anti-IL17 drugs (secukinumab and brodalumab) and guselkumab. Infliximab, anti-IL17 drugs (bimekizumab, ixekizumab, secukinumab and brodalumab) and anti-IL23 drugs (risankizumab and guselkumab) except tildrakizumab showed a higher proportion of patients reaching PASI 90 than ustekinumab and three anti-TNF alpha agents (adalimumab, certolizumab and etanercept). Ustekinumab was superior to certolizumab; adalimumab and ustekinumab were superior to etanercept. No significant difference was shown between apremilast and two non-biological drugs: ciclosporin and methotrexate. We found no significant difference between any of the interventions and the placebo for the risk of SAEs. The risk of SAEs was significantly lower for participants on methotrexate compared with most of the interventions. Nevertheless, the SAE analyses were based on a very low number of events with low- to moderate-certainty for all the comparisons (except methotrexate versus placebo, which was high-certainty). The findings therefore have to be viewed with caution. For other efficacy outcomes (PASI 75 and Physician Global Assessment (PGA) 0/1), the results were similar to the results for PASI 90. Information on quality of life was often poorly reported and was absent for several of the interventions.

AUTHORS' CONCLUSIONS

Our review shows that, compared to placebo, the biologics infliximab, bimekizumab, ixekizumab, and risankizumab were the most effective treatments for achieving PASI 90 in people with moderate-to-severe psoriasis on the basis of high-certainty evidence. This NMA evidence is limited to induction therapy (outcomes measured from 8 to 24 weeks after randomisation), and is not sufficient for evaluating longer-term outcomes in this chronic disease. Moreover, we found low numbers of studies for some of the interventions, and the young age (mean 44.5 years) and high level of disease severity (PASI 20.4 at baseline) may not be typical of patients seen in daily clinical practice. We found no significant difference in the assessed interventions and placebo in terms of SAEs, and the safety evidence for most interventions was low to moderate quality. More randomised trials directly comparing active agents are needed, and these should include systematic subgroup analyses (sex, age, ethnicity, comorbidities, psoriatic arthritis). To provide long-term information on the safety of treatments included in this review, an evaluation of non-randomised studies and postmarketing reports from regulatory agencies is needed. Editorial note: This is a living systematic review. Living systematic reviews offer a new approach to review updating, in which the review is continually updated, incorporating relevant new evidence as it becomes available. Please refer to the Cochrane Database of Systematic Reviews for the current status of this review.

The impact of antibiotics on clinical response over time in uncomplicated cellulitis: a systematic review and meta-analysis

PURPOSE

Antibiotic treatment of uncomplicated cellulitis is highly variable with respect to agent, dose, and route of administration. As there is uncertainty about optimal/appropriate time to reassess, we aimed to assess time to clinical response.

METHODS

We conducted a systematic review of randomized controlled trials reporting clinical response of uncomplicated cellulitis to antibiotic treatment over multiple timepoints. PubMed, Embase, CENTRAL, WHO ICTRP, and clinicaltrials.gov were searched from inception to June 2021 without language restrictions. The primary outcome was time to clinical response. Other outcomes were components of clinical response (pain, severity score, redness, edema measured at ≥ 2 timepoints) and the proportion of patients with treatment failure. We performed a pooled estimate of the average time to clinical response together with 95% confidence intervals using a random effects model.

RESULTS

We included 32 randomized controlled trials (n = 13,576 participants). The mean time to clinical response was 1.68 days (95%CI 1.48-1.88; I² = 76%). The response to treatment for specific components was as follows: ~ 50% reduction of pain and severity score by day 5, a ~ 33% reduction in area of redness by day 2-3, and a 30-50% reduction of proportion of patients with edema by day 2-4. Treatment failure was variably defined with an overall failure rate of 12% (95%CI 9-16%).

CONCLUSION

The best available data suggest the optimal time to clinical reassessment is between 2 and 4 days, but this must be interpreted with caution due to considerable heterogeneity and small number of included studies.

Music therapy for autistic people

BACKGROUND

Social interaction and social communication are among the central areas of difficulty for autistic people. Music therapy uses music experiences and the relationships that develop through them to enable communication and expression, thus attempting to address some of the core problems of autistic people. Music therapy has been applied in autism since the early 1950s, but its availability to autistic individuals varies across countries and settings. The application of music therapy requires specialised academic and clinical training which enables therapists to tailor the intervention to the specific needs of the individual. The present version of this review on music therapy for autistic people is an update of the previous Cochrane review update published in 2014 (following the original Cochrane review published in 2006).

OBJECTIVES

To review the effects of music therapy, or music therapy added to standard care, for autistic people.

SEARCH METHODS

In  August 2021, we searched  CENTRAL, MEDLINE, Embase, eleven other databases and two trials registers. We also ran citation searches, checked reference lists, and contacted study authors to identify additional studies.

SELECTION CRITERIA

All randomised controlled trials (RCTs), quasi-randomised trials and controlled clinical trials comparing music therapy (or music therapy alongside standard care) to 'placebo' therapy, no treatment, or standard care for people with a diagnosis of autism spectrum disorder were considered for inclusion.

DATA COLLECTION AND ANALYSIS

We used standard Cochrane methodological procedures. Four authors independently selected studies and extracted data from all included studies. We synthesised the results of included studies in meta-analyses. Four authors independently assessed risk of bias (RoB) of each included study using the original RoB tool as well as the certainty of evidence using GRADE.   MAIN RESULTS: We included 16 new studies in this update which brought the total number of included studies to 26 (1165 participants). These studies examined the short- and medium-term effect of music therapy (intervention duration: three days to eight months) for autistic people in individual or group settings. More than half of the studies were conducted in North America or Asia. Twenty-one studies included children aged from two to 12 years. Five studies included children and adolescents, and/or young adults. Severity levels, language skills, and cognition were widely variable across studies. Measured immediately post-intervention, music therapy compared with 'placebo' therapy or standard care was more likely to positively effect global improvement (risk ratio (RR) 1.22, 95% confidence interval (CI) 1.06 to 1.40; 8 studies, 583 participants; moderate-certainty evidence; number needed to treat for an additional beneficial outcome (NNTB) = 11 for low-risk population, 95% CI 6 to 39; NNTB = 6 for high-risk population, 95% CI 3 to 21) and to slightly increase quality of life (SMD 0.28, 95% CI 0.06 to 0.49; 3 RCTs, 340 participants; moderate-certainty evidence, small to medium effect size). In addition, music therapy probably results in a large reduction in total autism symptom severity (SMD -0.83, 95% CI -1.41 to -0.24; 9 studies, 575 participants; moderate-certainty evidence). No clear evidence of a difference between music therapy and comparison groups at immediately post-intervention was found for social interaction (SMD 0.26, 95% CI -0.05 to 0.57, 12 studies, 603 participants; low-certainty evidence); non-verbal communication (SMD 0.26, 95% CI -0.03 to 0.55; 7 RCTs, 192 participants; low-certainty evidence); and verbal communication (SMD 0.30, 95% CI -0.18 to 0.78; 8 studies, 276 participants; very low-certainty evidence). Two studies investigated adverse events with one (36 participants) reporting no adverse events; the other study found no differences between music therapy and standard care immediately post-intervention (RR 1.52, 95% CI 0.39 to 5.94; 1 study, 290 participants; moderate-certainty evidence).  AUTHORS' CONCLUSIONS: The findings of this updated review provide evidence that music therapy is probably associated with an increased chance of global improvement for autistic people, likely helps them to improve total autism severity and quality of life, and probably does not increase adverse events immediately after the intervention. The certainty of the evidence was rated as 'moderate' for these four outcomes, meaning that we are moderately confident in the effect estimate. No clear evidence of a difference was found for social interaction, non-verbal communication, and verbal communication measured immediately post-intervention. For these outcomes, the certainty of the evidence was rated as 'low' or 'very low', meaning that the true effect may be substantially different from these results. Compared with earlier versions of this review, the new studies included in this update helped to increase the certainty and applicability of this review's findings through larger sample sizes, extended age groups, longer periods of intervention and inclusion of follow-up assessments, and by predominantly using validated scales measuring generalised behaviour (i.e. behaviour outside of the therapy context). This new evidence is important for autistic individuals and their families as well as for policymakers, service providers and clinicians, to help in decisions around the types and amount of intervention that should be provided and in the planning of resources. The applicability of the findings is still limited to the age groups included in the studies, and no direct conclusions can be drawn about music therapy in autistic individuals above the young adult age. More research using rigorous designs, relevant outcome measures, and longer-term follow-up periods is needed to corroborate these findings and to examine whether the effects of music therapy are enduring.

Using artificial intelligence methods for systematic review in health sciences: A systematic review

The exponential increase in published articles makes a thorough and expedient review of literature increasingly challenging. This review delineated automated tools and platforms that employ artificial intelligence (AI) approaches and evaluated the reported benefits and challenges in using such methods. A search was conducted in 4 databases (Medline, Embase, CDSR, and Epistemonikos) up to April 2021 for systematic reviews and other related reviews implementing AI methods. To be included, the review must use any form of AI method, including machine learning, deep learning, neural network, or any other applications used to enable the full or semi-autonomous performance of one or more stages in the development of evidence synthesis. Twelve reviews were included, using nine different tools to implement 15 different AI methods. Eleven methods were used in the screening stages of the review (73%). The rest were divided: two in data extraction (13%) and two in risk of bias assessment (13%). The ambiguous benefits of the data extractions, combined with the reported advantages from 10 reviews, indicating that AI platforms have taken hold with varying success in evidence synthesis. However, the results are qualified by the reliance on the self-reporting of the review authors. Extensive human validation still appears required at this stage in implementing AI methods, though further evaluation is required to define the overall contribution of such platforms in enhancing efficiency and quality in evidence synthesis.

Topical and oral steroids for otitis media with effusion (OME) in children

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the effects (benefits and harms) of topical and oral steroids for otitis media with effusion (OME) in children.

Autoinflation for otitis media with effusion (OME) in children

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the effects (benefits and harms) of autoinflation for otitis media with effusion (OME) in children.

Diclofenac for acute postoperative pain in children

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the efficacy and safety of diclofenac (any dose) for acute postoperative pain management in children compared with placebo, other active comparators, or diclofenac administered by either different routes (e.g. oral, rectal, etc.) or strategies (e.g. as needed versus as scheduled).

Algorithm-based pain management for people with dementia in nursing homes

BACKGROUND

People with dementia in nursing homes often experience pain, but often do not receive adequate pain therapy. The experience of pain has a significant impact on quality of life in people with dementia, and is associated with negative health outcomes. Untreated pain is also considered to be one of the causes of challenging behaviour, such as agitation or aggression, in this population. One approach to reducing pain in people with dementia in nursing homes is an algorithm-based pain management strategy, i.e. the use of a structured protocol that involves pain assessment and a series of predefined treatment steps consisting of various non-pharmacological and pharmacological pain management interventions.

OBJECTIVES

To assess the effects of algorithm-based pain management interventions to reduce pain and challenging behaviour in people with dementia living in nursing homes. To describe the components of the interventions and the content of the algorithms.

SEARCH METHODS

We searched ALOIS, the Cochrane Dementia and Cognitive Improvement Group's register, MEDLINE, Embase, PsycINFO, CINAHL (Cumulative Index to Nursing and Allied Health Literature), Web of Science Core Collection (ISI Web of Science), LILACS (Latin American and Caribbean Health Science Information database), ClinicalTrials.gov and the World Health Organization's meta-register the International Clinical Trials Registry Portal on 30 June 2021.

SELECTION CRITERIA

We included randomised controlled trials investigating the effects of algorithm-based pain management interventions for people with dementia living in nursing homes. All interventions had to include an initial pain assessment, a treatment algorithm (a treatment plan consisting of at least two different non-pharmacological or pharmacological treatment steps to reduce pain), and criteria to assess the success of each treatment step. The control groups could receive usual care or an active control intervention. Primary outcomes for this review were pain-related outcomes, e.g. the number of participants with pain (self- or proxy-rated), challenging behaviour (we used a broad definition that could also include agitation or behavioural and psychological symptoms assessed with any validated instrument), and serious adverse events.

DATA COLLECTION AND ANALYSIS

Two authors independently selected the articles for inclusion, extracted data and assessed the risk of bias of all included studies. We reported results narratively as there were too few studies for a meta-analysis. We used GRADE methods to rate the certainty of the results.

MAIN RESULTS

We included three cluster-randomised controlled trials with a total of 808 participants (mean age 82 to 89 years). In two studies, participants had severe cognitive impairment and in one study mild to moderate impairment. The algorithms used in the studies varied in the number of treatment steps. The comparator was pain education for nursing staff in two studies and usual care in one study. We judged the risk of detection bias to be high in one study. The risk of selection bias and performance bias was unclear in all studies. Self-rated pain (i.e. pain rated by participants themselves) was reported in two studies. In one study, all residents in the nursing homes were included, but fewer than half of the participants experienced pain at baseline, and the mean values of self-rated and proxy-rated pain at baseline and follow-up in both study groups were below the threshold of pain that may require treatment. We considered the evidence from this study to be very low-certainty and therefore are uncertain whether the algorithm-based pain management intervention had an effect on self-rated pain intensity compared with pain education (MD -0.27, 95% CI -0.49 to -0.05, 170 participants; Verbal Descriptor Scale, range 0 to 3). In the other study, all participants had mild to moderate pain at baseline. Here, we found low-certainty evidence that an algorithm-based pain management intervention may have little to no effect on self-rated pain intensity compared with pain education (MD 0.4, 95% CI -0.58 to 1.38, 246 participants; Iowa Pain Thermometer, range 0 to 12). Pain was rated by proxy in all three studies. Again, we considered the evidence from the study in which mean pain scores indicated no pain, or almost no pain, at baseline to be very low-certainty and were uncertain whether the algorithm-based pain management intervention had an effect on proxy-rated pain intensity compared with pain education. For participants with mild to moderate pain at baseline, we found low-certainty evidence that an algorithm-based pain management intervention may reduce proxy-rated pain intensity in comparison with usual care (MD -1.49, 95% CI -2.11 to -0.87, 1 study, 128 participants; Pain Assessment in Advanced Dementia Scale-Chinese version, range 0 to 10), but may not be more effective than pain education (MD -0.2, 95% CI -0.79 to 0.39, 1 study, 383 participants; Iowa Pain Thermometer, range 0 to 12). For challenging behaviour, we found very low-certainty evidence from one study in which mean pain scores indicated no pain, or almost no pain, at baseline. We were uncertain whether the algorithm-based pain management intervention had any more effect than education for nursing staff on challenging behaviour of participants (MD -0.21, 95% CI -1.88 to 1.46, 1 study, 170 participants; Cohen-Mansfield Agitation Inventory-Chinese version, range 7 to 203). None of the studies systematically assessed adverse effects or serious adverse effects and no study reported information about the occurrence of any adverse effect. None of the studies assessed any of the other outcomes of this review.

AUTHORS' CONCLUSIONS

There is no clear evidence for a benefit of an algorithm-based pain management intervention in comparison with pain education for reducing pain intensity or challenging behaviour in people with dementia in nursing homes. We found that the intervention may reduce proxy-rated pain compared with usual care. However, the certainty of evidence is low because of the small number of studies, small sample sizes, methodological limitations, and the clinical heterogeneity of the study populations (e.g. pain level and cognitive status). The results should be interpreted with caution. Future studies should also focus on the implementation of algorithms and their impact in clinical practice.

Evaluation of publication type tagging as a strategy to screen randomized controlled trial articles in preparing systematic reviews

Objectives

To produce a systematic review (SR), reviewers typically screen thousands of titles and abstracts of articles manually to find a small number which are read in full text to find relevant articles included in the final SR. Here, we evaluate a proposed automated probabilistic publication type screening strategy applied to the randomized controlled trial (RCT) articles (i.e., those which present clinical outcome results of RCT studies) included in a corpus of previously published Cochrane reviews.

Materials and Methods

We selected a random subset of 558 published Cochrane reviews that specified RCT study only inclusion criteria, containing 7113 included articles which could be matched to PubMed identifiers. These were processed by our automated RCT Tagger tool to estimate the probability that each article reports clinical outcomes of a RCT.

Results

Removing articles with low predictive scores P < 0.01 eliminated 288 included articles, of which only 22 were actually typical RCT articles, and only 18 were actually typical RCT articles that MEDLINE indexed as such. Based on our sample set, this screening strategy led to fewer than 0.05 relevant RCT articles being missed on average per Cochrane SR.

Discussion

This scenario, based on real SRs, demonstrates that automated tagging can identify RCT articles accurately while maintaining very high recall. However, we also found that even SRs whose inclusion criteria are restricted to RCT studies include not only clinical outcome articles per se, but a variety of ancillary article types as well.

Conclusions

This encourages further studies learning how best to incorporate automated tagging of additional publication types into SR triage workflows.

Pharmacological interventions for prophylaxis of vestibular migraine

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the benefits and harms of pharmacological treatments used for prophylaxis of vestibular migraine.

Adverse effects of remdesivir, hydroxychloroquine and lopinavir/ritonavir when used for COVID-19: systematic review and meta-analysis of randomised trials

BACKGROUND

To summarise specific adverse effects of remdesivir, hydroxychloroquine and lopinavir/ritonavir in patients with COVID-19.

METHODS

We searched 32 databases through 27 October 2020. We included randomised trials comparing any of the drugs of interest to placebo or standard care, or against each other. We conducted fixed-effects pairwise meta-analysis and assessed the certainty of evidence using the grading of recommendations assessment, development and evaluation approach.

RESULTS

We included 16 randomised trials which enrolled 8152 patients. For most interventions and outcomes the certainty of the evidence was very low to low except for gastrointestinal adverse effects from hydroxychloroquine, which was moderate certainty. Compared with standard care or placebo, low certainty evidence suggests that remdesivir may not have an important effect on acute kidney injury (risk difference (RD) 8 fewer per 1000, 95% CI 27 fewer to 21 more) or cognitive dysfunction/delirium (RD 3 more per 1000, 95% CI 12 fewer to 19 more). Low certainty evidence suggests that hydroxychloroquine may increase the risk of cardiac toxicity (RD 10 more per 1000, 95% CI 0 more to 30 more) and cognitive dysfunction/delirium (RD 33 more per 1000, 95% CI 18 fewer to 84 more), whereas moderate certainty evidence suggests hydroxychloroquine probably increases the risk of diarrhoea (RD 106 more per 1000, 95% CI 48 more to 175 more) and nausea and/or vomiting (RD 62 more per 1000, 95% CI 23 more to 110 more) compared with standard care or placebo. Low certainty evidence suggests lopinavir/ritonavir may increase the risk of diarrhoea (RD 168 more per 1000, 95% CI 58 more to 330 more) and nausea and/or vomiting (RD 160 more per 1000, 95% CI 100 more to 210 more) compared with standard care or placebo.

DISCUSSION

Hydroxychloroquine probably increases the risk of diarrhoea and nausea and/or vomiting and may increase the risk of cardiac toxicity and cognitive dysfunction/delirium. Lopinavir/ritonavir may increase the risk of diarrhoea and nausea and/or vomiting. Remdesivir may have no important effect on risk of acute kidney injury or cognitive dysfunction/delirium. These findings provide important information to support the development of evidence-based management strategies for patients with COVID-19.

Adverse effects of remdesivir, hydroxychloroquine and lopinavir/ritonavir when used for COVID-19: systematic review and meta-analysis of randomised trials

BACKGROUND

To summarise specific adverse effects of remdesivir, hydroxychloroquine and lopinavir/ritonavir in patients with COVID-19.

METHODS

We searched 32 databases through 27 October 2020. We included randomised trials comparing any of the drugs of interest to placebo or standard care, or against each other. We conducted fixed-effects pairwise meta-analysis and assessed the certainty of evidence using the grading of recommendations assessment, development and evaluation approach.

RESULTS

We included 16 randomised trials which enrolled 8152 patients. For most interventions and outcomes the certainty of the evidence was very low to low except for gastrointestinal adverse effects from hydroxychloroquine, which was moderate certainty. Compared with standard care or placebo, low certainty evidence suggests that remdesivir may not have an important effect on acute kidney injury (risk difference (RD) 8 fewer per 1000, 95% CI 27 fewer to 21 more) or cognitive dysfunction/delirium (RD 3 more per 1000, 95% CI 12 fewer to 19 more). Low certainty evidence suggests that hydroxychloroquine may increase the risk of cardiac toxicity (RD 10 more per 1000, 95% CI 0 more to 30 more) and cognitive dysfunction/delirium (RD 33 more per 1000, 95% CI 18 fewer to 84 more), whereas moderate certainty evidence suggests hydroxychloroquine probably increases the risk of diarrhoea (RD 106 more per 1000, 95% CI 48 more to 175 more) and nausea and/or vomiting (RD 62 more per 1000, 95% CI 23 more to 110 more) compared with standard care or placebo. Low certainty evidence suggests lopinavir/ritonavir may increase the risk of diarrhoea (RD 168 more per 1000, 95% CI 58 more to 330 more) and nausea and/or vomiting (RD 160 more per 1000, 95% CI 100 more to 210 more) compared with standard care or placebo.

DISCUSSION

Hydroxychloroquine probably increases the risk of diarrhoea and nausea and/or vomiting and may increase the risk of cardiac toxicity and cognitive dysfunction/delirium. Lopinavir/ritonavir may increase the risk of diarrhoea and nausea and/or vomiting. Remdesivir may have no important effect on risk of acute kidney injury or cognitive dysfunction/delirium. These findings provide important information to support the development of evidence-based management strategies for patients with COVID-19.

Use of tocilizumab and sarilumab alone or in combination with corticosteroids for covid-19: systematic review and network meta-analysis

Objective

To compare the effects of interleukin 6 receptor blockers, tocilizumab and sarilumab, with or without corticosteroids, on mortality in patients with covid-19.

Design

Systematic review and network meta-analysis.

Data sources

World Health Organization covid-19 database, a comprehensive multilingual source of global covid-19 literature, and two prospective meta-analyses (up to 9 June 2021).

Review methods

Trials in which people with suspected, probable, or confirmed covid-19 were randomised to interleukin 6 receptor blockers (with or without corticosteroids), corticosteroids, placebo, or standard care. The analysis used a bayesian framework and assessed the certainty of evidence using the GRADE approach. Results from the fixed effect meta-analysis were used for the primary analysis.

Results

Of 45 eligible trials (20 650 patients) identified, 36 (19 350 patients) could be included in the network meta-analysis. Of 36 trials, 27 were at high risk of bias, primarily due to lack of blinding. Tocilizumab, in combination with corticosteroids, suggested a reduction in the risk of death compared with corticosteroids alone (odds ratio 0.79, 95% credible interval 0.70 to 0.88; 35 fewer deaths per 1000 people, 95% credible interval 52 fewer to 18 fewer per 1000; moderate certainty of evidence), as did sarilumab in combination with corticosteroids, compared with corticosteroids alone (0.73, 0.58 to 0.92; 43 fewer per 1000, 73 fewer to 12 fewer; low certainty). Tocilizumab and sarilumab, each in combination with corticosteroids, appeared to have similar effects on mortality when compared with each other (1.07, 0.86 to 1.34; eight more per 1000, 20 fewer to 35 more; low certainty). The effects of tocilizumab (1.12, 0.91 to 1.38; 20 more per 1000, 16 fewer to 59 more; low certainty) and sarilumab (1.07, 0.81 to 1.40; 11 more per 1000, 38 fewer to 55 more; low certainty), when used alone, suggested an increase in the risk of death.

Conclusion

These findings suggest that in patients with severe or critical covid-19, tocilizumab, in combination with corticosteroids, probably reduces mortality, and that sarilumab, in combination with corticosteroids, might also reduce mortality. Tocilizumab and sarilumab, in combination with corticosteroids, could have similar effectiveness. Tocilizumab and sarilumab, when used alone, might not be beneficial.

Direct Oral Anticoagulants Compared With Dalteparin for Treatment of Cancer-Associated Thrombosis: A Living, Interactive Systematic Review and Network Meta-analysis

OBJECTIVE

To maintain living, interactive evidence (LIvE) on the benefits and harms of different treatment options in adults with cancer-associated thrombosis (CAT).

METHODS

We have used a novel LIvE synthesis framework to maintain this living, interactive systematic review since September 19, 2018. Randomized controlled trials evaluating the efficacy and safety of direct oral anticoagulants (DOACs) compared with low-molecular-weight heparin for CAT are included in this analysis. Details of LIvE synthesis framework are available at the website https://cat.network-meta-analysis.com.

RESULTS

The results are constantly updated as new information becomes available (https://cat.network-meta-analysis.com/CAT.html). The living, interactive systematic review currently includes 4 randomized controlled trials (N=2894). Direct comparisons show that DOACs significantly decrease recurrent venous thromboembolism (VTE) events compared with dalteparin (odds ratio [OR], 0.59; 95% CI, 0.41 to 0.86; I², 25%) without significantly increasing major bleeding (OR, 1.34; 95% CI, 0.83 to 2.18; I², 28%). Mixed treatment comparisons show that apixaban (OR, 0.41; 95% credible interval [CrI], 0.16 to 0.95) and rivaroxaban (OR, 0.58; 95% CrI, 0.37 to 0.90) significantly decrease VTE recurrent events compared with dalteparin. Edoxaban significantly increases major bleeding compared with dalteparin (OR, 1.73; 95% CrI, 1.04 to 3.16), and rivaroxaban significantly increases clinically relevant nonmajor bleeding compared with dalteparin and other DOACs. There are no significant differences between DOACs in terms of VTE recurrences and major bleeding.

CONCLUSION

DOACs should be considered a standard of care for the treatment of CAT except in patients with a high risk of bleeding. Current evidence favors the use of apixaban for the treatment of CAT among other DOACs.

REGISTRATION

Open Science Framework (https://osf.io/dth86).

Cuffed versus uncuffed endotracheal tubes for neonates

BACKGROUND

Endotracheal intubation is a commonly performed procedure in neonates, the risks of which are well-described. Some endotracheal tubes (ETT) are equipped with a cuff that can be inflated after insertion of the ETT in the airway to limit leak or aspiration. Cuffed ETTs have been shown in larger children and adults to reduce gas leak around the ETT, ETT exchange, accidental extubation, and exposure of healthcare workers to anesthetic gas during surgery. With improved understanding of neonatal airway anatomy and the widespread use of cuffed ETTs by anesthesiologists, the use of cuffed tubes is increasing in neonates.

OBJECTIVES

To assess the benefits and harms of cuffed ETTs (inflated or non-inflated) compared to uncuffed ETTs for respiratory support in neonates.

SEARCH METHODS

We searched CENTRAL, PubMed, and CINAHL on 20 August 2021; we also searched trial registers and checked reference lists to identify additional studies.

SELECTION CRITERIA

We included randomized controlled trials (RCTs), quasi-RCTs, and cluster-randomized trials comparing cuffed (inflated and non-inflated) versus uncuffed ETTs in newborns. We sought to compare 1. inflated, cuffed versus uncuffed ETT; 2. non-inflated, cuffed versus uncuffed ETT; and 3. inflated, cuffed versus non-inflated, cuffed ETT.

DATA COLLECTION AND ANALYSIS

We used the standard methods of Cochrane Neonatal. Two review authors independently assessed studies identified by the search strategy for inclusion, extracted data, and assessed risk of bias. We used the GRADE approach to assess the certainty of evidence.

MAIN RESULTS

We identified one eligible RCT for inclusion that compared the use of cuffed (inflated if ETT leak greater than 20% with cuff pressure 20 cm H₂O or less) versus uncuffed ETT. The author provided a spreadsheet with individual data. Among 76 infants in the original manuscript, 69 met the inclusion and exclusion criteria for this Cochrane Review. We found possible bias due to lack of blinding and other bias. We are very uncertain about frequency of postextubation stridor, because the confidence intervals (CI) of the risk ratio (RR) were very wide (RR 1.36, 95% CI 0.35 to 5.25; risk difference (RD) 0.03, -0.11 to 0.18; 1 study, 69 participants; very low-certainty evidence). No neonate was diagnosed with postextubation subglottic stenosis; however, endoscopy was not available to confirm the clinical diagnosis. We are very uncertain about reintubation for stridor or subglottic stenosis because the CIs of the RR were very wide (RR 0.27, 95% CI 0.01 to 6.49; RD -0.03, 95% CI -0.11 to 0.05; 1 study, 69 participants; very low-certainty evidence). No neonate had surgical intervention (e.g. endoscopic balloon dilation, cricoid split, tracheostomy) for stridor or subglottic stenosis (1 study, 69 participants). Neonates randomized to cuffed ETT may be less likely to have a reintubation for any reason (RR 0.06, 95% CI 0.01 to 0.45; RD -0.39, 95% CI -0.57 to -0.21; number needed to treat for an additional beneficial outcome 3, 95% CI 2 to 5; 1 study, 69 participants; very low-certainty evidence). We are very uncertain about accidental extubation because the CIs of the RR were wide (RR 0.82, 95% CI 0.12 to 5.46; RD -0.01, 95% CI -0.12 to 0.10; 1 study, 69 participants; very low-certainty evidence). We are very uncertain about all-cause mortality during initial hospitalization because the CIs of the RR were extremely wide (RR 2.46, 95% CI 0.10 to 58.39; RD 0.03, 95% CI -0.05 to 0.10; 1 study, 69 participants; very low-certainty evidence). There is one ongoing trial. We classified two studies as awaiting classification because outcome data were not reported separately for newborns and older infants.

AUTHORS' CONCLUSIONS

Evidence for comparing cuffed versus uncuffed ETTs in neonates is limited by a small number of babies in a single RCT with possible bias. There is very low certainty evidence for all outcomes of this review. CIs of the estimate for postextubation stridor were wide. No neonate had clinical evidence for subglottic stenosis; however, endoscopy results were not available to assess the anatomy. Additional RCTs are necessary to evaluate the benefits and harms of cuffed ETTs (inflated and non-inflated) in the neonatal population. These studies must include neonates and be conducted both for short-term use (in the setting of the operating room) and chronic use (in the setting of chronic lung disease) of cuffed ETTs.

High versus low-added sugar consumption for the primary prevention of cardiovascular disease

BACKGROUND

High intake of added sugar have been suggested to impact the risk for cardiovascular disease (CVD). Knowledge on the subject can contribute to preventing CVD.

OBJECTIVES

To assess the effects of a high versus low-added sugar consumption for primary prevention of CVD in the general population.

SEARCH METHODS

We searched Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, MEDLINE, Embase, Conference Proceedings Citation Index-Science (CPCI-S) on 2 July 2021. We also conducted a search of ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) Search Portal for ongoing or unpublished trials. The search was performed together with reference checking, citation searching and contact with study authors to identify additional studies. We imposed no restriction on language of publication or publication status.

SELECTION CRITERIA

We included randomised controlled trials (RCTs), including cross-over trials, that compared different levels of added sugar intake. Exclusion criteria were: participants aged below 18 years; diabetes mellitus (type 1 and 2); and previous CVD. Primary outcomes were incident cardiovascular events (coronary, carotid, cerebral and peripheral arterial disease) and all-cause mortality. Secondary outcomes were changes in systolic and diastolic blood pressure, total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, fasting plasma glucose and adverse events (gastrointestinal symptoms and impaired dental health).

DATA COLLECTION AND ANALYSIS

We used the standard methodological procedures expected by Cochrane.

MAIN RESULTS

We included 21 RCTs (1110 participants completing the interventions) examining the effects of different levels of added sugar intake with a mean duration of 14 weeks. The study participants were generally described as healthy and the mean age ranged from 22 to 57 years. No studies reported on cardiovascular events or all-cause mortality. There was minimal effect of low intake of added sugar on total cholesterol levels (MD 0.11, 95% CI 0.01 to 0.21; I² = 0%; 16 studies; 763 participants; low certainty of evidence) and triglycerides (MD 0.10, 95% CI 0.03 to 0.17; I² = 3%; 14 studies; 725 participants) but no evidence of effect on LDL-cholesterol and HDL-cholesterol. There was minimal effect on diastolic blood pressure (MD 1.52, 95% CI 0.67 to 2.37; I² = 0%; 13 studies; 873 participants) and on systolic blood pressure (MD 1.44, 95% 0.08 to 2.80; I² = 27%, 14 studies; 873 participants; low certainty of evidence), but no evidence of effect on fasting plasma glucose. Only one study reported on dental health, with no events. No other trials reported adverse events (impaired dental health or gastrointestinal symptoms). All results were judged as low-quality evidence according to GRADE. The risk of bias was generally unclear, five studies were classified at an overall low risk of bias (low risk in at least four domains, not including other bias).

AUTHORS' CONCLUSIONS

No trials investigating the effect of added sugar on cardiovascular events or all-cause mortality were identified in our searches. Evidence is uncertain whether low intake of added sugar has an effect on risk factors for CVD; the effect was small and the clinical relevance is, therefore, uncertain. Practical ways to achieve reductions in dietary added sugar includes following current dietary recommendations. Future trials should have longer follow-up time and report on all-cause mortality and cardiovascular events in order to clarify the effect of added sugar on these outcomes. Future trials should also aim for more direct interventions and preferably be more independent of industry funding.

Semi-automated Tools for Systematic Searches

Traditionally, literature identification for systematic reviews has relied on a two-step process: first, searching databases to identify potentially relevant citations, and then manually screening those citations. A number of tools have been developed to streamline and semi-automate this process, including tools to generate terms; to visualize and evaluate search queries; to trace citation linkages; to deduplicate, limit, or translate searches across databases; and to prioritize relevant abstracts for screening. Research is ongoing into tools that can unify searching and screening into a single step, and several protype tools have been developed. As this field grows, it is becoming increasingly important to develop and codify methods for evaluating the extent to which these tools fulfill their purpose.

Machine learning in systematic reviews: Comparing automated text clustering with Lingo3G and human researcher categorization in a rapid review

Systematic reviews are resource-intensive. The machine learning tools being developed mostly focus on the study identification process, but tools to assist in analysis and categorization are also needed. One possibility is to use unsupervised automatic text clustering, in which each study is automatically assigned to one or more meaningful clusters. Our main aim was to assess the usefulness of an automated clustering method, Lingo3G, in categorizing studies in a simplified rapid review, then compare performance (precision and recall) of this method compared to manual categorization. We randomly assigned all 128 studies in a review to be coded by a human researcher blinded to cluster assignment (mimicking two independent researchers) or by a human researcher non-blinded to cluster assignment (mimicking one researcher checking another's work). We compared time use, precision and recall of manual categorization versus automated clustering. Automated clustering and manual categorization organized studies by population and intervention/context. Automated clustering failed to identify two manually identified categories but identified one additional category not identified by the human researcher. We estimate that automated clustering has similar precision to both blinded and non-blinded researchers (e.g., 88% vs. 89%), but higher recall (e.g., 89% vs. 84%). Manual categorization required 49% more time than automated clustering. Using a specific clustering algorithm, automated clustering can be helpful with categorization of and identifying patterns across studies in simpler systematic reviews. We found that the clustering was sensitive enough to group studies according to linguistic differences that often corresponded to the manual categories.

Global Research on Coronaviruses: Metadata-Based Analysis for Public Health Policies

BACKGROUND

Within the context of the COVID-19 pandemic, this paper suggests a data science strategy for analyzing global research on coronaviruses. The application of reproducible research principles founded on text-as-data information, open science, the dissemination of scientific data, and easy access to scientific production may aid public health in the fight against the virus.

OBJECTIVE

The primary goal of this paper was to use global research on coronaviruses to identify critical elements that can help inform public health policy decisions. We present a data science framework to assist policy makers in implementing cutting-edge data science techniques for the purpose of developing evidence-based public health policies.

METHODS

We used the EpiBibR (epidemiology-based bibliography for R) package to gain access to coronavirus research documents worldwide (N=121,231) and their associated metadata. To analyze these data, we first employed a theoretical framework to group the findings into three categories: conceptual, intellectual, and social. Second, we mapped the results of our analysis in these three dimensions using machine learning techniques (ie, natural language processing) and social network analysis.

RESULTS

Our findings, firstly, were methodological in nature. They demonstrated the potential for the proposed data science framework to be applied to public health policies. Additionally, our findings indicated that the United States and China were the primary contributors to global coronavirus research during the study period. They also demonstrated that India and Europe were significant contributors, albeit in a secondary position. University collaborations in this domain were strong between the United States, Canada, and the United Kingdom, confirming the country-level findings.

CONCLUSIONS

Our findings argue for a data-driven approach to public health policy, particularly when efficient and relevant research is required. Text mining techniques can assist policy makers in calculating evidence-based indices and informing their decision-making process regarding specific actions necessary for effective health responses.