Importance Large language models (LLMs) may facilitate the labor-intensive process of systematic reviews. However, the exact methods and reliability remain uncertain. Objective To explore the feasibility and reliability of using LLMs to assess risk of bias (ROB) in randomized clinical trials (RCTs). Design, Setting, and Participants A survey study was conducted between August 10, 2023, and October 30, 2023. Thirty RCTs were selected from published systematic reviews. Main Outcomes and Measures A structured prompt was developed to guide ChatGPT (LLM 1) and Claude (LLM 2) in assessing the ROB in these RCTs using a modified version of the Cochrane ROB tool developed by the CLARITY group at McMaster University. Each RCT was assessed twice by both models, and the results were documented. The results were compared with an assessment by 3 experts, which was considered a criterion standard. Correct assessment rates, sensitivity, specificity, and F1 scores were calculated to reflect accuracy, both overall and for each domain of the Cochrane ROB tool; consistent assessment rates and Cohen kappa were calculated to gauge consistency; and assessment time was calculated to measure efficiency. Performance between the 2 models was compared using risk differences. Results Both models demonstrated high correct assessment rates. LLM 1 reached a mean correct assessment rate of 84.5% (95% CI, 81.5%-87.3%), and LLM 2 reached a significantly higher rate of 89.5% (95% CI, 87.0%-91.8%). The risk difference between the 2 models was 0.05 (95% CI, 0.01-0.09). In most domains, domain-specific correct rates were around 80% to 90%; however, sensitivity below 0.80 was observed in domains 1 (random sequence generation), 2 (allocation concealment), and 6 (other concerns). Domains 4 (missing outcome data), 5 (selective outcome reporting), and 6 had F1 scores below 0.50. The consistent rates between the 2 assessments were 84.0% for LLM 1 and 87.3% for LLM 2. LLM 1's kappa exceeded 0.80 in 7 and LLM 2's in 8 domains. The mean (SD) time needed for assessment was 77 (16) seconds for LLM 1 and 53 (12) seconds for LLM 2. Conclusions In this survey study of applying LLMs for ROB assessment, LLM 1 and LLM 2 demonstrated substantial accuracy and consistency in evaluating RCTs, suggesting their potential as supportive tools in systematic review processes.

COVID-19 is not only associated with substantial acute liver and kidney injuries, but also with an elevated risk of post-acute sequelae involving the kidney and liver system. We aimed to investigate whether COVID-19 exposure increases the long-term risk of kidney and liver disease, and what are the magnitudes of these associations. We searched PubMed, Embase, Web of Science, , and the Living Overview of the Evidence COVID-19 Repository for cohort studies estimating the association between COVID-19 and kidney and liver outcomes. Random-effects meta-analyses were performed to combine the results of the included studies. We assessed the certainty of the evidence using the Grading of Recommendations Assessment, Development and Evaluation approach. Fifteen cohort studies with more than 32 million participants were included in the systematic review COVID-19 was associated with a 35% greater risk of kidney diseases (10 more per 1000 persons; low certainty evidence) and 54% greater risk of liver disease (3 more per 1000 persons; low certainty evidence). The absolute increases due to COVID-19 for acute kidney injury, chronic kidney disease, and liver test abnormality were 3, 8, and 3 per 1000 persons, respectively. Subgroup analyses found no differences between different type of kidney and liver diseases. The findings provide further evidence for the association between COVID-19 and incident kidney and liver conditions. The absolute magnitude of the effect of COVID-19 on kidney and liver outcomes was, however, relatively small.

Objective We explored the comparative effectiveness of available therapies for chronic pain associated with temporomandibular disorders (TMD). DESIGN Systematic review and network meta-analysis of randomised clinical trials (RCTs).DATA SOURcES MEDLINE, EMBASE, CINAHL, CENTRAL, and SCOPUS were searched to May 2021, and again in January 2023.STUDY SELEcTION Interventional RCTs that enrolled patients presenting with chronic pain associated with TMD.DATA EXTRAcTION AND SYNTHESIS Pairs of reviewers independently identified eligible studies, extracted data, and assessed risk of bias. We captured all reported patient-important outcomes, including pain relief, physical functioning, emotional functioning, role functioning, social functioning, sleep quality, and adverse events. We conducted frequentist network meta-analyses to summarise the evidence and used the GRADE approach to rate the certainty of evidence and categorise interventions from most to least beneficial.RESULTS 233 trials proved eligible for review, of which 153-enrolling 8713 participants and exploring 59 interventions or combinations of interventions-were included in network meta-analyses. All subsequent effects refer to comparisons with placebo or sham procedures. Effects on pain for eight interventions were supported by high to moderate certainty evidence. The three therapies probably most effective for pain relief were cognitive behavioural therapy (CBT) augmented with biofeedback or relaxation therapy (risk difference (RD) for achieving the minimally important difference (MID) in pain relief of 1 cm on a 10 cm visual analogue scale: 36% (95% CI 33 to 39)), therapist-assisted jaw mobilisation (RD 36% (95% CI 31 to 40)), and manual trigger point therapy (RD 32% (29 to 34)). Five interventions were less effective, yet more effective than placebo, showing RDs ranging between 23% and 30%: CBT, supervised postural exercise, supervised jaw exercise and stretching, supervised jaw exercise and stretching with manual trigger point therapy, and usual care (such as home exercises, self stretching, reassurance). Moderate certainty evidence showed four interventions probably improved physical functioning: supervised jaw exercise and stretching (RD for achieving the MID of 5 points on the short form-36 physical component summary score: 43% (95% CI 33 to 51)), manipulation (RD 43% (25 to 56)), acupuncture (RD 42% (33 to 50)), and supervised jaw exercise and mobilisation (RD 36% (19 to 51)). The evidence for pain relief or physical functioning among other interventions, and all evidence for adverse events, was low or very low certainty.cONcLUSION When restricted to moderate or high certainty evidence, interventions that promote coping and encourage movement and activity were found to be most effective for reducing chronic TMD pain.REGISTRATION PROSPERO (CRD42021258567)

Background: A variety of nonpharmacological interventions that improve the quality of life of patients with advanced cancer have been difficult for medical staff to select through randomized controlled trials or traditional meta-analyses. Thus, a network meta-analysis is necessary. Objective: This study used network meta-analysis to analyze the effect of 13 different nonpharmacological interventions on improving the living quality of patients with advanced cancer. Methods: Five English databases were searched up to January 2019. The search strategy only included terms relating to or describing the intervention. Results: The study included 13 different nonpharmacological interventions. The overall efficacy was summarized through a holistic study of quality of life. The study found that the combined effect sizes of 13 nonpharmacological interventions crossed the invalid line (weighted mean difference, -13 [95% confidence interval, -33 to 8.5] to 1.7 [95% confidence interval, -18 to 22]), indicating that none of the intervention was significantly different from each other. By evaluating the heterogeneity of this outcome, no significant evidence of heterogeneity ( P > .05) was observed. Probability ranking according to the surface under the cumulative ranking curve showed that there was a great possibility for the CanWalk intervention and structured multidisciplinary intervention to improve outcomes for cancer patients. Conclusions: Thirteen nonpharmacological interventions did not significantly impact quality of life. Regarding the probability rank, CanWalk intervention may be the most promising way that advanced cancer patients can help themselves to a better life. Because of the limitations of the current studies, the conclusion needs further evidence. Implications for practice: Nurses should consider recommending moderate physical activity for patients with advanced cancer.

IMPORTANCE Respiratory syncytial virus (RSV) is the leading cause of acute lower respiratory infection children younger than 5 years; effective prevention strategies are urgently needed. OBJECTIVE To compare the efficacy and safety of monoclonal antibodies for the prevention of RSV infection in infants and children. DATA SOURCES In this systematic review and network meta-analysis, PubMed, Embase, CENTRAL, and ClinicalTrials.gov were searched from database inception to March 2022. STUDY SELECTION Randomized clinical trials that enrolled infants at high risk of RSV infection to receive a monoclonal antibody or placebowere included. Keywords and extensive vocabulary related to monoclonal antibodies, RSV, and randomized clinical trials were searched. DATA EXTRACTION AND SYNTHESIS The Preferred Reporting Items for Systematic Reviews and Meta-analyses reporting guideline was used. Teams of 2 reviewers independently performed literature screening, data extraction, and risk of bias assessment. The Grading of Recommendations, Assessments, Developments, and Evaluation approach was used to rate the certainty of evidence. A random-effects model network meta-analysis was conducted using a consistency model under the frequentist framework. MAIN OUTCOMES AND MEASURES The main outcomes were all-cause mortality, RSV-related hospitalization, RSV-related infection, drug-related adverse events, intensive care unit admission, supplemental oxygen use, and mechanical ventilation use. RESULTS Fifteen randomized clinical trials involving 18 395 participants were eligible; 14 were synthesized, with 18 042 total participants (median age at study entry, 3.99 months [IQR, 3.25-6.58 months]; median proportion of males, 52.37%[IQR, 50.49%-53.85%]). Compared with placebo, with moderate- to high-certainty evidence, nirsevimab, palivizumab, and motavizumab were associated with significantly reduced RSV-related infections per 1000 participants (nirsevimab: -123 [95% CI, -138 to -100]; palivizumab: -108 [95% CI, -127 to -82]; motavizumab: -136 [95% CI, -146 to -125]) and RSV-related hospitalizations per 1000 participants (nirsevimab: -54 [95% CI, -64 to -38; palivizumab: -39 [95% CI, -48 to -28]; motavizumab: -48 [95% CI, - 58 to -33]). With moderate-certainty evidence, both motavizumab and palivizumab were associated with significant reductions in intensive care unit admissions per 1000 participants (-8 [95% CI, -9 to -4] and -5 [95% CI, -7 to 0], respectively) and supplemental oxygen use per 1000 participants (-59 [95% CI, -63 to -54] and -55 [95% CI, -61 to -41], respectively), and nirsevimab was associated with significantly reduced supplemental oxygen use per 1000 participants (-59 [95% CI, -65 to -40]). No significant differences were found in all-cause mortality and drug-related adverse events. Suptavumab did not show any significant benefits for the outcomes of interest. CONCLUSIONS AND RELEVANCE In this study, motavizumab, nirsevimab, and palivizumab were associated with substantial benefits in the prevention of RSV infection, without a significant increase in adverse events compared with placebo. However, more research is needed to confirm the present conclusions, especially for safety and cost-effectiveness.

BackgroundPharmacological treatment is common in practice and widely used for the management of insomnia. However, evidence comparing the relative effectiveness, safety, and certainty of evidence among drug classes and individual drugs for insomnia are still lacking. This study aimed to determine the relative effectiveness, safety, and tolerability of drugs for insomnia.MethodsIn this systematic review and network meta-analysis we systematically searched PubMed, Embase, Cochrane Central Register of Controlled Trials, PsycINFO, and ClinicalTrials.gov, from inception to January 10, 2022 to identify randomized controlled trials that compared insomnia drugs with placebo or an active comparator in adults with insomnia. We conducted random-effects frequentist network meta-analyses to summarize the evidence, and used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach to assess the certainty, categorize interventionsand present the findings.ResultsA total of 148 articles met our eligibility criteria; these included 153 trials which enrolled 46,412 participants and assessed 36 individual drugs from eight drug classes. Compared with placebo, both subjectively and objectively measured total sleep time were significantly improved with non-benzodiazepine (subjective: mean difference [MD] 25.07, 95% confidence interval [CI] 15.49-34.64, low certainty; objective: MD 22.34, 95% CI 7.64-37.05, high certainty), antidepressants (subjective: MD 54.40, 95% CI 34.96-75.83, low certainty; objective: MD 35.64, 95% CI 13.05-58.24, high certainty), and orexin receptor antagonists (subjective: MD 21.62, 95% CI 0.84-42.40, high certainty; objective: MD 31.81, 95% CI 2.66-60.95, high certainty); of which doxepin, almorexant, suvorexant, and lemborexant were among the relatively effective drugs with relatively good tolerability and lower risks of any adverse events (AEs). Both subjectively and objectively measured sleep onset latency were significantly shortened with non-benzodiazepines (subjective: MD - 10.12, 95% CI - 13.84 to - 6.40, moderate certainty; objective: MD - 12.11, 95% CI - 19.31 to - 4.90, moderate certainty) and melatonin receptor agonists (subjective: MD - 7.73, 95% CI - 15.21 to - 0.26, high certainty; objective: MD - 7.04, 95% CI - 12.12 to - 1.95, moderate certainty); in particular, zopiclone was among the most effective drugs with a lower risk of any AEs but worse tolerability. Non-benzodiazepines could significantly decrease both subjective and objective measured wake time after sleep onset (subjective: MD - 16.67, 95% CI - 21.79 to - 11.56, moderate certainty; objective: MD - 13.92, 95% CI - 22.71 to - 5.14, moderate certainty).ConclusionsNon-benzodiazepines probably improve total sleep time, sleep onset latency, and wake time after sleep onset. Other insomnia drug classes and individual drugs also showed potential benefits in improving insomnia symptoms. However, the choice of insomnia drugs should be based on the phenotype of insomnia presented, as well as each drug's safety and tolerability.Protocol registration PROSPERO (CRD42019138790).

Background Pulmonary embolism (PE) is a potentially life-threatening condition in which a clot can migrate from the deep veins, most commonly in the leg, to the lungs. Conventional treatment of PE used unfractionated heparin (UFH), low molecular weight heparin (LMWH), fondaparinux, and vitamin K antagonists (VKAs). Recently, two forms of direct oral anticoagulants (DOACs) have been developed: oral direct thrombin inhibitors (DTIs) and oral factor Xa inhibitors. DOACs have characteristics that may be favourable to conventional treatment, including oral administration, a predictable effect, no need for frequent monitoring or re-dosing, and few known drug interactions. This review reports the efficacy and safety of these drugs in the long-term treatment of PE (minimum duration of three months). This is an update of a Cochrane Review first published in 2015. Objectives To assess the efficacy and safety of oral DTIs and oral factor Xa inhibitors versus conventional anticoagulants for the long-term treatment of PE. Search methods The Cochrane Vascular Information Specialist searched the Cochrane Vascular Specialised Register, CENTRAL, MEDLINE, Embase and CINAHL databases, the World Health Organization International Clinical Trials Registry Platform and the ClinicalTrials.gov trials registers to 2 March 2022. We checked the reference lists of relevant articles for additional studies. Selection criteria We included randomised controlled trials (RCTs) in which people with a PE confirmed by standard imaging techniques were allocated to receive an oral DTI or an oral factor Xa inhibitor compared with a conventional anticoagulant or compared with each other for the long-term treatment of PE (minimum duration three months). Data collection and analysis We used standard Cochrane methods. Our primary outcomes were recurrent PE, recurrent venous thromboembolism (VTE), and deep vein thrombosis (DVT). Secondary outcomes were all-cause mortality, major bleeding, and health-related quality of life. We used GRADE to assess the certainty of evidence for each outcome. Main results We identified five additional RCTs with 1484 participants for this update. Together with the previously included trials, we have included ten RCTs with a total of 13,073 participants. Two studies investigated an oral DTI (dabigatran) and eight studies investigated oral factor Xa inhibitors (three rivaroxaban, three apixaban, and two edoxaban). The studies were of good methodological quality overall. Meta-analysis showed no clear difference in the efficacy and safety of oral DTI compared with conventional anticoagulation in preventing recurrent PE (odds ratio (OR) 1.02, 95% confidence interval (CI) 0.50 to 2.04; 2 studies, 1602 participants; moderate-certainty evidence), recurrent VTE (OR 0.93, 95% CI 0.52 to 1.66; 2 studies, 1602 participants; moderate-certainty evidence), DVT (OR 0.79, 95% CI 0.29 to 2.13; 2 studies, 1602 participants; moderate-certainty evidence), and major bleeding (OR 0.50, 95% CI 0.15 to 1.68; 2 studies, 1527 participants; moderate-certainty evidence). We downgraded the certainty of evidence by one level for imprecision due to the low number of events. There was also no clear difference between the oral factor Xa inhibitors and conventional anticoagulation in the prevention of recurrent PE (OR 0.92, 95% CI 0.66 to 1.29; 3 studies, 8186 participants; moderate-certainty evidence), recurrent VTE (OR 0.83, 95% CI 0.66 to 1.03; 8 studies, 11,416 participants; moderate-certainty evidence), DVT (OR 0.77, 95% CI 0.48 to 1.25; 2 studies, 8151 participants; moderate-certainty evidence), all-cause mortality (OR 1.16, 95% CI 0.79 to 1.70; 1 study, 4817 participants; moderate-certainty evidence) and major bleeding (OR 0.71, 95% CI 0.36 to 1.41; 8 studies, 11,447 participants; low-certainty evidence); the heterogeneity for major bleeding was significant (I-2 = 79%). We downgraded the certainty of the evidence to moderate and low because of imprecision due to the low number of events and inconsistency due to clinical heterogeneity. None of the included studies measured health-related quality of life. Authors' conclusions Available evidence shows there is probably little or no difference between DOACs and conventional anticoagulation in the prevention of recurrent PE, recurrent VTE, DVT, all-cause mortality, and major bleeding. The certainty of evidence was moderate or low. Future large clinical trials are required to identify if individual drugs differ in effectiveness and bleeding risk, and to explore effect differences in subgroups, including people with cancer and obesity.

Background: Specific treatments targeting Ebola virus are crucial in managing Ebola virus disease. To support the development of clinical practice guidelines on medications for Ebola virus disease, we aimed to evaluate the efficacy and safety of therapies for patients with Ebola virus disease. Methods: In this systematic review and network meta-analysis, we searched MEDLINE, Embase, the Cochrane Central Register of Controlled Trials, Scopus, Global Health, African Index Medicus, World Health Organization Global Index Medicus, the Cumulative Index to Nursing and Allied Health Literature, ClinicalTrials.gov, Epistemonikos, bioRxiv, medRxiv, and SSRN without language restrictions for randomised controlled trials (RCTs) published between database inception and Jan 1, 2022, comparing at least one therapeutic agent for Ebola virus disease against standard care or another therapeutic agent for Ebola virus disease. Two reviewers assessed study eligibility and extracted summary data independently using a standardised form. Our outcomes of interest were mortality, adverse maternal outcomes, risk of onward transmission, duration of admission to a health-care facility, functional status after Ebola virus disease, serious adverse events from medication, adverse perinatal outcomes, time to symptom resolution, and time to viral clearance. We did frequentist network meta-analyses to estimate the effect of all interventions and applied the Grading of Recommendations Assessment, Development and Evaluation approach to rate the certainty of the evidence. We registered the protocol with PROSPERO, CRD42022296539. Findings: We identified 7840 records through database searches, of which two RCTs with a total of 753 patients proved eligible. Only data on mortality, the duration of admission, serious adverse events, and time to viral clearance were available for meta-analysis. Compared with standard care, REGN-EB3 (relative risk [RR] 0·40, 95% CI 0·18 to 0·89; moderate certainty) and mAb114 (0·42, 0·19 to 0·93; moderate certainty) probably reduce mortality. Whether ZMapp (0·60, 0·28 to 1·26; very low certainty) and remdesivir (0·64, 0·29 to 1·39; very low certainty) reduce mortality compared with standard care is uncertain. With high certainty, REGN-EB3 reduces mortality compared with ZMapp (0·67, 0·52 to 0·88) and remdesivir (0·63, 0·49 to 0·82). With high certainty, mAb114 also reduces mortality compared with ZMapp (0·71, 0·55 to 0·91) and remdesivir (0·66, 0·52 to 0·84). Compared with standard care, REGN-EB3, mAb114, ZMapp, and remdesivir might have little or no effect on the time to viral clearance (mean difference ranged from -0·25 days to -1·14 days; low certainty). ZMapp might reduce the duration of admission compared with standard care (mean difference -2·02 days, 95% CI -4·05 to 0·01; low certainty). Findings for all comparisons suggested that there might be little or no difference in the prevalence of serious adverse events, but certainty was low or very low in all comparisons but one. Interpretation: REGN-EB3 and mAb114 separately reduce mortality compared with ZMapp, remdesivir, or standard care in patients with Ebola virus disease. These findings suggest that health-care workers should prioritise the use of REGN-EB3 and mAb114 for patients with Ebola virus disease during future outbreaks.

IntroductionThere was limited evidence on the quality of reporting and methodological quality of prediction models using machine learning methods in preterm birth. This systematic review aimed to assess the reporting quality and risk of bias of a machine learning-based prediction model in preterm birth. Material and methodsWe conducted a systematic review, searching the PubMed, Embase, the Cochrane Library, China National Knowledge Infrastructure, China Biology Medicine disk, VIP Database, and WanFang Data from inception to September 27, 2021. Studies that developed (validated) a prediction model using machine learning methods in preterm birth were included. We used the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement and Prediction model Risk of Bias Assessment Tool (PROBAST) to evaluate the reporting quality and the risk of bias of included studies, respectively. Findings were summarized using descriptive statistics and visual plots. The protocol was registered in PROSPERO (no. CRD 42022301623). ResultsTwenty-nine studies met the inclusion criteria, with 24 development-only studies and 5 development-with-validation studies. Overall, TRIPOD adherence per study ranged from 17% to 79%, with a median adherence of 49%. The reporting of title, abstract, blinding of predictors, sample size justification, explanation of model, and model performance were mostly poor, with TRIPOD adherence ranging from 4% to 17%. For all included studies, 79% had a high overall risk of bias, and 21% had an unclear overall risk of bias. The analysis domain was most commonly rated as high risk of bias in included studies, mainly as a result of small effective sample size, selection of predictors based on univariable analysis, and lack of calibration evaluation. ConclusionsReporting and methodological quality of machine learning-based prediction models in preterm birth were poor. It is urgent to improve the design, conduct, and reporting of such studies to boost the application of machine learning-based prediction models in preterm birth in clinical practice.

Background: COVID-19 might be a risk factor for various chronic diseases. However, the association between COVID-19 and the risk of incident diabetes remains unclear. We aimed to meta-analyze evidence on the relative risk of incident diabetes in patients with COVID-19. Methods: In this systematic review and meta-analysis, the Embase, PubMed, CENTRAL, and Web of Science databases were searched from December 2019 to June 8, 2022. We included cohort studies that provided data on the number, proportion, or relative risk of diabetes after confirming the COVID-19 diagnosis. Two reviewers independently screened studies for eligibility, extracted data, and assessed risk of bias. We used a random-effects meta-analysis to pool the relative risk with corresponding 95 % confidence intervals. Prespecified subgroup and meta-regression analyses were conducted to explore the potential influencing factors. We converted the relative risk to the absolute risk difference to present the evidence. This study was registered in advance (PROSPERO CRD42022337841). Main findings: Ten articles involving 11 retrospective cohorts with a total of 47.1 million participants proved eligible. We found a 64 % greater risk (RR = 1.64, 95%CI: 1.51 to 1.79) of diabetes in patients with COVID-19 compared with non-COVID-19 controls, which could increase the number of diabetes events by 701 (558 more to 865 more) per 10,000 persons. We detected significant subgroup effects for type of diabetes and sex. Type 2 diabetes has a higher relative risk than type 1. Moreover, men may be at a higher risk of overall diabetes than women. Sensitivity analysis confirmed the robustness of the results. No evidence was found for publication bias. Conclusions: COVID-19 is strongly associated with the risk of incident diabetes, including both type 1 and type 2 diabetes. We should be aware of the risk of developing diabetes after COVID-19 and prepare for the associated health problems, given the large and growing number of people infected with COVID-19. However, the body of evidence still needs to be strengthened.