Background
Network meta‐analysis (NMA) is a powerful analysis method used to identify the best treatments for a condition and is used extensively by health care decision makers. Although software ...routines exist for conducting NMA, they require considerable statistical programming expertise to use, which limits the number of researchers able to conduct such analyses.
Objectives
To develop a web‐based tool allowing users with only standard internet browser software to be able to conduct NMAs using an intuitive “point and click” interface and present the results using visual plots.
Methods
Using the existing netmeta and Shiny packages for R to conduct the analyses, and to develop the user interface, we created the MetaInsight tool which is freely available to use via the web.
Results
A package was created for conducting NMA which satisfied our objectives, and this is described, and its application demonstrated, using an illustrative example.
Conclusions
We believe that many researchers will find our package helpful for facilitating NMA as well as allowing decision makers to scrutinize presented results visually and in real time. This will impact on the relevance of statistical analyses for health care decision making and sustainably increase capacity by empowering informed nonspecialists to be able to conduct more clinically relevant reviews. It is also hoped that others will be inspired to create similar tools for other advanced specialist analyses methods using the freely available technologies we have adopted.
Network meta-analyses (NMA) have extensively been used to compare the effectiveness of multiple interventions for health care policy and decision-making. However, methods for evaluating the ...performance of multiple diagnostic tests are less established. In a decision-making context, we are often interested in comparing and ranking the performance of multiple diagnostic tests, at varying levels of test thresholds, in one simultaneous analysis.
Motivated by an example of cognitive impairment diagnosis following stroke, we synthesized data from 13 studies assessing the efficiency of two diagnostic tests: Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA), at two test thresholds: MMSE <25/30 and <27/30, and MoCA <22/30 and <26/30. Using Markov chain Monte Carlo (MCMC) methods, we fitted a bivariate network meta-analysis model incorporating constraints on increasing test threshold, and accounting for the correlations between multiple test accuracy measures from the same study.
We developed and successfully fitted a model comparing multiple tests/threshold combinations while imposing threshold constraints. Using this model, we found that MoCA at threshold <26/30 appeared to have the best true positive rate, whereas MMSE at threshold <25/30 appeared to have the best true negative rate.
The combined analysis of multiple tests at multiple thresholds allowed for more rigorous comparisons between competing diagnostics tests for decision making.
Abstract
Background
The aim of this study was to describe outcomes in hospitalised older people with different levels of frailty and COVID-19 infection.
Methods
We undertook a single-centre, ...retrospective cohort study examining COVID-19-related mortality using electronic health records, for older people (65 and over) with frailty, hospitalised with or without COVID-19 infection. Baseline covariates included demographics, early warning scores, Charlson Comorbidity Indices and frailty (Clinical Frailty Scale, CFS), linked to COVID-19 status.
Findings
We analysed outcomes on 1,071 patients with COVID-19 test results (285 (27%) were positive for COVID-19). The mean age at ED arrival was 79.7 and 49.4% were female. All-cause mortality (by 30 days) rose from 9 (not frail) to 33% (severely frail) in the COVID-negative cohort but was around 60% for all frailty categories in the COVID-positive cohort. In adjusted analyses, the hazard ratio for death in those with COVID-19 compared to those without COVID-19 was 7.3 (95% CI: 3.00, 18.0) with age, comorbidities and illness severity making small additional contributions.
Interpretation
In this study, frailty measured using the CFS appeared to make little incremental contribution to the hazard of dying in older people hospitalised with COVID-19 infection; illness severity and comorbidity had a modest association with the overall adjusted hazard of death, whereas confirmed COVID-19 infection dominated, with a sevenfold hazard for death.
Current UK vaccination policy is to offer future COVID-19 booster doses to individuals at high risk of serious illness from COVID-19, but it is still uncertain which groups of the population could ...benefit most. In response to an urgent request from the UK Joint Committee on Vaccination and Immunisation, we aimed to identify risk factors for severe COVID-19 outcomes (ie, COVID-19-related hospitalisation or death) in individuals who had completed their primary COVID-19 vaccination schedule and had received the first booster vaccine.
We constructed prospective cohorts across all four UK nations through linkages of primary care, RT-PCR testing, vaccination, hospitalisation, and mortality data on 30 million people. We included individuals who received primary vaccine doses of BNT162b2 (tozinameran; Pfizer–BioNTech) or ChAdOx1 nCoV-19 (Oxford–AstraZeneca) vaccines in our initial analyses. We then restricted analyses to those given a BNT162b2 or mRNA-1273 (elasomeran; Moderna) booster and had a severe COVID-19 outcome between Dec 20, 2021, and Feb 28, 2022 (when the omicron (B.1.1.529) variant was dominant). We fitted time-dependent Poisson regression models and calculated adjusted rate ratios (aRRs) and 95% CIs for the associations between risk factors and COVID-19-related hospitalisation or death. We adjusted for a range of potential covariates, including age, sex, comorbidities, and previous SARS-CoV-2 infection. Stratified analyses were conducted by vaccine type. We then did pooled analyses across UK nations using fixed-effect meta-analyses.
Between Dec 8, 2020, and Feb 28, 2022, 17 337 580 individuals completed their primary vaccine schedule and 14 698 030 individuals received a booster dose. Between Dec 20, 2021, and Feb 28, 2022, 59 510 (0·3%) of the primary vaccine group and 26 100 (0·2%) of those who received their booster had severe COVID-19 outcomes. The risk of severe COVID-19 outcomes reduced after receiving the booster (rate change: 8·8 events per 1000 person-years to 7·6 events per 1000 person-years). Older adults (≥80 years vs 18–49 years; aRR 3·60 95% CI 3·45–3·75), those with comorbidities (≥5 comorbidities vs none; 9·51 9·07–9·97), being male (male vs female; 1·23 1·20–1·26), and those with certain underlying health conditions—in particular, individuals receiving immunosuppressants (yes vs no; 5·80 5·53–6·09)—and those with chronic kidney disease (stage 5 vs no; 3·71 2·90–4·74) remained at high risk despite the initial booster. Individuals with a history of COVID-19 infection were at reduced risk (infected ≥9 months before booster dose vs no previous infection; aRR 0·41 95% CI 0·29–0·58).
Older people, those with multimorbidity, and those with specific underlying health conditions remain at increased risk of COVID-19 hospitalisation and death after the initial vaccine booster and should, therefore, be prioritised for additional boosters, including novel optimised versions, and the increasing array of COVID-19 therapeutics.
National Core Studies–Immunity, UK Research and Innovation (Medical Research Council), Health Data Research UK, the Scottish Government, and the University of Edinburgh.
Several countries restricted the administration of ChAdOx1 to older age groups in 2021 over safety concerns following case reports and observed versus expected analyses suggesting a possible ...association with cerebral venous sinus thrombosis (CVST). Large datasets are required to precisely estimate the association between Coronavirus Disease 2019 (COVID-19) vaccination and CVST due to the extreme rarity of this event. We aimed to accomplish this by combining national data from England, Scotland, and Wales.
We created data platforms consisting of linked primary care, secondary care, mortality, and virological testing data in each of England, Scotland, and Wales, with a combined cohort of 11,637,157 people and 6,808,293 person years of follow-up. The cohort start date was December 8, 2020, and the end date was June 30, 2021. The outcome measure we examined was incident CVST events recorded in either primary or secondary care records. We carried out a self-controlled case series (SCCS) analysis of this outcome following first dose vaccination with ChAdOx1 and BNT162b2. The observation period consisted of an initial 90-day reference period, followed by a 2-week prerisk period directly prior to vaccination, and a 4-week risk period following vaccination. Counts of CVST cases from each country were tallied, then expanded into a full dataset with 1 row for each individual and observation time period. There was a combined total of 201 incident CVST events in the cohorts (29.5 per million person years). There were 81 CVST events in the observation period among those who a received first dose of ChAdOx1 (approximately 16.34 per million doses) and 40 for those who received a first dose of BNT162b2 (approximately 12.60 per million doses). We fitted conditional Poisson models to estimate incidence rate ratios (IRRs). Vaccination with ChAdOx1 was associated with an elevated risk of incident CVST events in the 28 days following vaccination, IRR = 1.93 (95% confidence interval (CI) 1.20 to 3.11). We did not find an association between BNT162b2 and CVST in the 28 days following vaccination, IRR = 0.78 (95% CI 0.34 to 1.77). Our study had some limitations. The SCCS study design implicitly controls for variables that are constant over the observation period, but also assumes that outcome events are independent of exposure. This assumption may not be satisfied in the case of CVST, firstly because it is a serious adverse event, and secondly because the vaccination programme in the United Kingdom prioritised the clinically extremely vulnerable and those with underlying health conditions, which may have caused a selection effect for individuals more prone to CVST. Although we pooled data from several large datasets, there was still a low number of events, which may have caused imprecision in our estimates.
In this study, we observed a small elevated risk of CVST events following vaccination with ChAdOx1, but not BNT162b2. Our analysis pooled information from large datasets from England, Scotland, and Wales. This evidence may be useful in risk-benefit analyses of vaccine policies and in providing quantification of risks associated with vaccination to the general public.
SARS-CoV-2 infection in children and young people (CYP) can lead to life-threatening COVID-19, transmission within households and schools, and the development of long COVID. Using linked health and ...administrative data, we investigated vaccine uptake among 3,433,483 CYP aged 5-17 years across all UK nations between 4th August 2021 and 31st May 2022. We constructed national cohorts and undertook multi-state modelling and meta-analysis to identify associations between demographic variables and vaccine uptake. We found that uptake of the first COVID-19 vaccine among CYP was low across all four nations compared to other age groups and diminished with subsequent doses. Age and vaccination status of adults living in the same household were identified as important risk factors associated with vaccine uptake in CYP. For example, 5-11 year-olds were less likely to receive their first vaccine compared to 16-17 year-olds (adjusted Hazard Ratio aHR: 0.10 (95%CI: 0.06-0.19)), and CYP in unvaccinated households were less likely to receive their first vaccine compared to CYP in partially vaccinated households (aHR: 0.19, 95%CI 0.13-0.29).
Network meta-analysis synthesises data from a number of clinical trials in order to assess the comparative efficacy of multiple healthcare interventions in similar patient populations. In situations ...where clinical trial data are heterogeneously reported i.e. data are missing for one or more outcomes of interest, synthesising such data can lead to disconnected networks of evidence, increased uncertainty, and potentially biased estimates which can have severe implications for decision-making. To overcome this issue, strength can be borrowed between outcomes of interest in multivariate network meta-analyses. Furthermore, in situations where there are relatively few trials informing each treatment comparison, there is a potential issue with the sparsity of data in the treatment networks, which can lead to substantial parameter uncertainty. A multivariate network meta-analysis approach can be further extended to borrow strength between interventions of the same class using hierarchical models.
We extend the trivariate network meta-analysis model to incorporate the exchangeability between treatment effects belonging to the same class of intervention to increase precision in treatment effect estimates. We further incorporate a missing data framework to estimate uncertainty in trials that did not report measures of variability in order to maximise the use of all available information for healthcare decision-making. The methods are applied to a motivating dataset in overactive bladder syndrome. The outcomes of interest were mean change from baseline in incontinence, voiding and urgency episodes. All models were fitted using Bayesian Markov Chain Monte Carlo (MCMC) methods in WinBUGS.
All models (univariate, multivariate, and multivariate models incorporating class effects) produced similar point estimates for all treatment effects. Incorporating class effects in multivariate models often increased precision in treatment effect estimates.
Multivariate network meta-analysis incorporating class effects allowed for the comparison of all interventions across all outcome measures to ameliorate the potential impact of outcome reporting bias, and further borrowed strength between interventions belonging to the same class of treatment to increase the precision in treatment effect estimates for healthcare policy and decision-making.
There is still limited understanding of how chronic conditions co-occur in patients with multimorbidity and what are the consequences for patients and the health care system. Most reported clusters ...of conditions have not considered the demographic characteristics of these patients during the clustering process. The study used data for all registered patients that were resident in Fife or Tayside, Scotland and aged 25 years or more on 1st January 2000 and who were followed up until 31st December 2018. We used linked demographic information, and secondary care electronic health records from 1st January 2000. Individuals with at least two of the 31 Elixhauser Comorbidity Index conditions were identified as having multimorbidity. Market basket analysis was used to cluster the conditions for the whole population and then repeatedly stratified by age, sex and deprivation. 318,235 individuals were included in the analysis, with 67,728 (21·3%) having multimorbidity. We identified five distinct clusters of conditions in the population with multimorbidity: alcohol misuse, cancer, obesity, renal failure, and heart failure. Clusters of long-term conditions differed by age, sex and socioeconomic deprivation, with some clusters not present for specific strata and others including additional conditions. These findings highlight the importance of considering demographic factors during both clustering analysis and intervention planning for individuals with multiple long-term conditions. By taking these factors into account, the healthcare system may be better equipped to develop tailored interventions that address the needs of complex patients.
Abstract Background Network meta-analysis (NMA) is commonly used in evidence synthesis; however, in situations in which there are a large number of treatment options, which may be subdivided into ...classes, and relatively few trials, NMAs produce considerable uncertainty in the estimated treatment effects, and consequently, identification of the most beneficial intervention remains inconclusive. Objective To develop and demonstrate the use of evidence synthesis methods to evaluate extensive treatment networks with a limited number of trials, making use of classes. Methods Using Bayesian Markov chain Monte Carlo methods, we build on the existing work of a random effects NMA to develop a three-level hierarchical NMA model that accounts for the exchangeability between treatments within the same class as well as for the residual between-study heterogeneity. We demonstrate the application of these methods to a continuous and binary outcome, using a motivating example of overactive bladder. We illustrate methods for incorporating ordering constraints in increasing doses, model selection, and assessing inconsistency between the direct and indirect evidence. Results The methods were applied to a data set obtained from a systematic literature review of trials for overactive bladder, evaluating the mean reduction in incontinence episodes from baseline and the number of patients reporting one or more adverse events. The data set involved 72 trials comparing 34 interventions that were categorized into nine classes of interventions, including placebo. Conclusions Bayesian three-level hierarchical NMAs have the potential to increase the precision in the effect estimates while maintaining the interpretability of the individual interventions for decision making.
We aimed to develop a network meta-analytic model for the evaluation of treatment effectiveness within predictive biomarker subgroups, by combining evidence from individual participant data (IPD) ...from digital sources (in the absence of randomized controlled trials) and aggregate data (AD).
A Bayesian framework was developed for modeling time-to-event data to evaluate predictive biomarkers. IPD were sourced from electronic health records, using a target trial emulation approach, or digitized Kaplan-Meier curves. The model is illustrated using two examples: breast cancer with a hormone receptor biomarker, and metastatic colorectal cancer with the Kirsten Rat Sarcoma (KRAS) biomarker.
The model allowed for the estimation of treatment effects in two subgroups of patients defined by their biomarker status. Effectiveness of taxanes did not differ in hormone receptor positive and negative breast cancer patients. Epidermal growth factor receptor inhibitors were more effective than chemotherapy in KRAS wild type colorectal cancer patients but not in patients with KRAS mutant status. Use of IPD reduced uncertainty of the subgroup-specific treatment effect estimates by up to 49%.
Utilization of IPD allowed for more detailed evaluation of predictive biomarkers and cancer therapies and improved precision of the estimates compared to use of AD alone.