John W. Stevens

Background: Thromboprophylaxis can reduce the risk of venous thromboembolism (VTE) during lower-limb immobilisation, but it is unclear whether or not this translates into meaningful health benefit, justifies the risk of bleeding or is cost-effective. Risk assessment models (RAMs) could select higher-risk individuals for thromboprophylaxis.Objectives: To determine the clinical effectiveness and cost-effectiveness of different strategies for providing thromboprophylaxis to people with lower-limb immobilisation caused by injury and to identify priorities for future research.Data sources: Ten electronic databases and research registers (MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, Database of Abstracts of Review of Effects, the Cochrane Central Register of Controlled Trials, Health Technology Assessment database, NHS Economic Evaluation Database, Science Citation Index Expanded, ClinicalTrials.gov and the International Clinical Trials Registry Platform) were searched from inception to May 2017, and this was supplemented by hand-searching reference lists and contacting experts in the field.Review methods: Systematic reviews were undertaken to determine the effectiveness of pharmacological thromboprophylaxis in lower-limb immobilisation and to identify any study of risk factors or RAMs for VTE in lower-limb immobilisation. Study quality was assessed using appropriate tools. A network meta-analysis was undertaken for each outcome in the effectiveness review and the results of risk-prediction studies were presented descriptively. A modified Delphi survey was undertaken to identify risk predictors supported by expert consensus. Decision-analytic modelling was used to estimate the incremental cost per quality-adjusted life-year (QALY) gained of different thromboprophylaxis strategies from the perspectives of the NHS and Personal Social Services.Results: Data from 6857 participants across 13 trials were included in the meta-analysis. Thromboprophylaxis with low-molecular-weight heparin reduced the risk of any VTE [odds ratio (OR) 0.52, 95% credible interval (CrI) 0.37 to 0.71], clinically detected deep-vein thrombosis (DVT) (OR 0.40, 95% CrI 0.12 to 0.99) and pulmonary embolism (PE) (OR 0.17, 95% CrI 0.01 to 0.88). Thromboprophylaxis with fondaparinux (Arixtra^®, Aspen Pharma Trading Ltd, Dublin, Ireland) reduced the risk of any VTE (OR 0.13, 95% CrI 0.05 to 0.30) and clinically detected DVT (OR 0.10, 95% CrI 0.01 to 0.94), but the effect on PE was inconclusive (OR 0.47, 95% CrI 0.01 to 9.54). Estimates of the risk of major bleeding with thromboprophylaxis were inconclusive owing to the small numbers of events. Fifteen studies of risk factors were identified, but only age (ORs 1.05 to 3.48), and injury type were consistently associated with VTE. Six studies of RAMs were identified, but only two reported prognostic accuracy data for VTE, based on small numbers of patients. Expert consensus was achieved for 13 risk predictors in lower-limb immobilisation due to injury. Modelling showed that thromboprophylaxis for all is effective (0.015 QALY gain, 95% CrI 0.004 to 0.029 QALYs) with a cost-effectiveness of £13,524 per QALY, compared with thromboprophylaxis for none. If risk-based strategies are included, it is potentially more cost-effective to limit thromboprophylaxis to patients with a Leiden thrombosis risk in plaster (cast) [L-TRiP(cast)] score of ≥ 9 (£20,000 per QALY threshold) or ≥ 8 (£30,000 per QALY threshold). An optimal threshold on the L-TRiP(cast) receiver operating characteristic curve would have sensitivity of 84–89% and specificity of 46–55%.Limitations: Estimates of RAM prognostic accuracy are based on weak evidence. People at risk of bleeding were excluded from trials and, by implication, from modelling.Conclusions: Thromboprophylaxis for lower-limb immobilisation due to injury is clinically effective and cost-effective compared with no thromboprophylaxis. Risk-based thromboprophylaxis is potentially optimal but the prognostic accuracy of existing RAMs is uncertain.Future work: Research is required to determine whether or not an appropriate RAM can accurately select higher-risk patients for thromboprophylaxis.Study registration: This study is registered as PROSPERO CRD42017058688.Funding: The National Institute for Health Research Health Technology Assessment programme.

Background and Objective: The extrapolation of estimated hazard functions can be an important part of cost-effectiveness analyses. Given limited follow-up time in the sample data, it may be expected that the uncertainty in estimates of hazards increases the further into the future they are extrapolated. The objective of this study was to illustrate how the choice of parametric survival model impacts on estimates of uncertainty about extrapolated hazard functions and lifetime mean survival.Methods: We examined seven commonly used parametric survival models and described analytical expressions and approximation methods (delta and multivariate normal) for estimating uncertainty. We illustrate the multivariate normal method using case studies based on four representative hypothetical datasets reflecting hazard functions commonly encountered in clinical practice (constant, increasing, decreasing, or unimodal), along with a hypothetical cost-effectiveness analysis.Results: Depending on the survival model chosen, the uncertainty in extrapolated hazard functions could be constant, increasing or decreasing over time for the case studies. Estimates of uncertainty in mean survival showed a large variation (up to sevenfold) for each case study. The magnitude of uncertainty in estimates of cost effectiveness, as measured using the incremental cost per quality-adjusted life-year gained, varied threefold across plausible models. Differences in estimates of uncertainty were observed even when models provided near-identical point estimates.Conclusions: Survival model choice can have a significant impact on estimates of uncertainty of extrapolated hazard functions, mean survival and cost effectiveness, even when point estimates were similar. We provide good practice recommendations for analysts and decision makers, emphasizing the importance of considering the plausibility of estimates of uncertainty in the extrapolated period as a complementary part of the model selection process.

Background and Aims: There are growing concerns over the respiratory health of people who use illicit opioids due to high rates of opioid inhalation and tobacco smoking in this group. This study aimed to summarize the evidence relating illicit opioid use with poor respiratory health.Methods: A systematic review of the literature on the association between illicit opioid use and respiratory health was undertaken in accordance with Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidance (PROSPERO ID = CRD42017059953). Electronic searches of MEDLINE, Embase, PsycINFO, CINAHL and the Cochrane Library databases were undertaken (English language, published January 1980–November 2018). All study designs excluding case studies were considered. Studies were undertaken in community and hospital settings in the United States (n = 23), United Kingdom (n = 7), Australia (n = 7), the Netherlands (n = 2), Canada (n = 2), Ireland (n = 1), Spain (n = 1) and Iran (n = 1). Measurements of respiratory disease, including asthma and chronic obstructive pulmonary disease (COPD) and related symptoms were extracted. Data on respiratory‐related deaths and hospital admissions were also extracted. Meta‐analysis of prevalence data was undertaken using a random effects meta‐analysis model with parameters estimated using Markov chain Monte Carlo simulation.Results: Meta‐analyses estimated prevalence of asthma in people who inject illicit opioids as 8.5% [95% predictive interval (PrI) = 0.2%, 74.0%] and as 20.2% (95% PrI = 4.2%, 59.2%) in people who inhale illicit opioids. Prevalence of COPD in people who inject illicit opioids was estimated as 2.7% (95% PrI = 0.0%, 50.4%) and as 17.9% (95% PrI = 0.6%, 89.5%) in people who inhale illicit opioids. There was evidence of moderate to extreme heterogeneity across studies.Conclusions: There is evidence of increased burden of respiratory diseases in people who use illicit opioids. Due to the heterogeneity of study design and samples, it is difficult to gain accurate estimates of the prevalence of respiratory disease in this population.

Background: Breast cancer and its treatment can have an impact on health-related quality of life and survival. Tumour profiling tests aim to identify whether or not women need chemotherapy owing to their risk of relapse.Objectives: To conduct a systematic review of the effectiveness and cost-effectiveness of the tumour profiling tests oncotypeDX^®(Genomic Health, Inc., Redwood City, CA, USA), MammaPrint^®(Agendia, Inc., Amsterdam, the Netherlands), Prosigna^®(NanoString Technologies, Inc., Seattle, WA, USA), EndoPredict^®(Myriad Genetics Ltd, London, UK) and immunohistochemistry 4 (IHC4). To develop a health economic model to assess the cost-effectiveness of these tests compared with clinical tools to guide the use of adjuvant chemotherapy in early-stage breast cancer from the perspective of the NHS and Personal Social Services.Design: A systematic review and health economic analysis were conducted.Review methods: The systematic review was partially an update of a 2013 review. Nine databases were searched in February 2017. The review included studies assessing clinical effectiveness in people with oestrogen receptor-positive, human epidermal growth factor receptor 2-negative, stage I or II cancer with zero to three positive lymph nodes. The economic analysis included a review of existing analyses and the development of a de novo model.Results: A total of 153 studies were identified. Only one completed randomised controlled trial (RCT) using a tumour profiling test in clinical practice was identified: Microarray In Node-negative Disease may Avoid ChemoTherapy (MINDACT) for MammaPrint. Other studies suggest that all the tests can provide information on the risk of relapse; however, results were more varied in lymph node-positive (LN+) patients than in lymph node-negative (LN0) patients. There is limited and varying evidence that oncotypeDX and MammaPrint can predict benefit from chemotherapy. The net change in the percentage of patients with a chemotherapy recommendation or decision pre/post test ranged from an increase of 1% to a decrease of 23% among UK studies and a decrease of 0% to 64% across European studies. The health economic analysis suggests that the incremental cost-effectiveness ratios for the tests versus current practice are broadly favourable for the following scenarios: (1) oncotypeDX, for the LN0 subgroup with a Nottingham Prognostic Index (NPI) of > 3.4 and the one to three positive lymph nodes (LN1–3) subgroup (if a predictive benefit is assumed); (2) IHC4 plus clinical factors (IHC4+C), for all patient subgroups; (3) Prosigna, for the LN0 subgroup with a NPI of > 3.4 and the LN1–3 subgroup; (4) EndoPredict Clinical, for the LN1–3 subgroup only; and (5) MammaPrint, for no subgroups.Limitations: There was only one completed RCT using a tumour profiling test in clinical practice. Except for oncotypeDX in the LN0 group with a NPI score of > 3.4 (clinical intermediate risk), evidence surrounding pre- and post-test chemotherapy probabilities is subject to considerable uncertainty. There is uncertainty regarding whether or not oncotypeDX and MammaPrint are predictive of chemotherapy benefit. The MammaPrint analysis uses a different data source to the other four tests. The Translational substudy of the Arimidex, Tamoxifen, Alone or in Combination (TransATAC) study (used in the economic modelling) has a number of limitations.Conclusions: The review suggests that all the tests can provide prognostic information on the risk of relapse; results were more varied in LN+ patients than in LN0 patients. There is limited and varying evidence that oncotypeDX and MammaPrint are predictive of chemotherapy benefit. Health economic analyses indicate that some tests may have a favourable cost-effectiveness profile for certain patient subgroups; all estimates are subject to uncertainty. More evidence is needed on the prediction of chemotherapy benefit, long-term impacts and changes in UK pre-/post-chemotherapy decisions.Study registration: This study is registered as PROSPERO CRD42017059561.Funding: The National Institute for Health Research Health Technology Assessment programme.

Background: Rheumatoid arthritis (RA) is a chronic, debilitating disease associated with reduced quality of life and substantial costs. It is unclear which tests and assessment tools allow the best assessment of prognosis in people with early RA and whether or not variables predict the response of patients to different drug treatments. Objective: To systematically review evidence on the use of selected tests and assessment tools in patients with early RA (1) in the evaluation of a prognosis (review 1) and (2) as predictive markers of treatment response (review 2). Data sources: Electronic databases (e.g. MEDLINE, EMBASE, The Cochrane Library, Web of Science Conference Proceedings; searched to September 2016), registers, key websites, hand-searching of reference lists of included studies and key systematic reviews and contact with experts. Study selection: Review 1 – primary studies on the development, external validation and impact of clinical prediction models for selected outcomes in adult early RA patients. Review 2 – primary studies on the interaction between selected baseline covariates and treatment (conventional and biological disease-modifying antirheumatic drugs) on salient outcomes in adult early RA patients. Results: Review 1 – 22 model development studies and one combined model development/external validation study reporting 39 clinical prediction models were included. Five external validation studies evaluating eight clinical prediction models for radiographic joint damage were also included. c -statistics from internal validation ranged from 0.63 to 0.87 for radiographic progression (different definitions, six studies) and 0.78 to 0.82 for the Health Assessment Questionnaire (HAQ). Predictive performance in external validations varied considerably. Three models [(1) Active controlled Study of Patients receiving Infliximab for the treatment of Rheumatoid arthritis of Early onset (ASPIRE) C-reactive protein (ASPIRE CRP), (2) ASPIRE erythrocyte sedimentation rate (ASPIRE ESR) and (3) Behandelings Strategie (BeSt)] were externally validated using the same outcome definition in more than one population. Results of the random-effects meta-analysis suggested substantial uncertainty in the expected predictive performance of models in a new sample of patients. Review 2 – 12 studies were identified. Covariates examined included anti-citrullinated protein/peptide anti-body (ACPA) status, smoking status, erosions, rheumatoid factor status, C-reactive protein level, erythrocyte sedimentation rate, swollen joint count (SJC), body mass index and vascularity of synovium on power Doppler ultrasound (PDUS). Outcomes examined included erosions/radiographic progression, disease activity, physical function and Disease Activity Score-28 remission. There was statistical evidence to suggest that ACPA status, SJC and PDUS status at baseline may be treatment effect modifiers, but not necessarily that they are prognostic of response for all treatments. Most of the results were subject to considerable uncertainty and were not statistically significant. Limitations: The meta-analysis in review 1 was limited by the availability of only a small number of external validation studies. Studies rarely investigated the interaction between predictors and treatment. Suggested research priorities: Collaborative research (including the use of individual participant data) is needed to further develop and externally validate the clinical prediction models. The clinical prediction models should be validated with respect to individual treatments. Future assessments of treatment by covariate interactions should follow good statistical practice. Conclusions: Review 1 – uncertainty remains over the optimal prediction model(s) for use in clinical practice. Review 2 – in general, there was insufficient evidence that the effect of treatment depended on baseline characteristics. Study registration: This study is registered as PROSPERO CRD42016042402. Funding: The National Institute for Health Research Health Technology Assessment programme.

Evidence from randomised controlled trials (RCTs) is used to support regulatory approval and reimbursement decisions. I discuss how these decisions are typically made and argue that the amount of sample data and regulatory authorities’ concerns over multiplicity are irrelevant when making reimbursement decisions. Decision analytic models (DAMs) are usually necessary to meet the requirements of an economic evaluation. DAMs involve inputs relating to health benefits and resource use that represent unknown true population parameters. Evidence about parameters may come from a variety of sources, including RCTs, and uncertainty about parameters is represented by their joint posterior distribution. Any impact of multiplicity is mitigated through the prior distribution. I illustrate my perspective with three examples: the estimation of a treatment effect on a rare event; the number of RCTs available in a meta-analysis; and the estimation of population mean overall survival. I conclude by recommending that reimbursement decisions should be followed by an assessment of the value of sample information and the DAM revised structurally as necessary and to include any new sample data that may be generated.

A network meta‐analysis allows a simultaneous comparison between treatments evaluated in randomised controlled trials that share at least one treatment with at least one other study. Estimates of treatment effects may be required for treatments across disconnected networks of evidence, which requires a different statistical approach and modelling assumptions to account for imbalances in prognostic variables and treatment effect modifiers between studies. In this paper, we review and discuss methods for comparing treatments evaluated in studies that form disconnected networks of evidence. Several methods have been proposed but assessing which are appropriate often depends on the clinical context as well as the availability of data. Most methods account for sampling variation but do not always account for others sources of uncertainty. We suggest that further research is required to assess the properties of methods and the use of approaches that allow the incorporation of external information to reflect parameter and structural uncertainty.

Background: Pairwise and network meta-analyses using fixed effect and random effects models are commonly applied to synthesise evidence from randomised controlled trials. The models differ in their assumptions and the interpretation of the results. The model choice depends on the objective of the analysis and knowledge of the included studies. Fixed effect models are often used because there are too few studies with which to estimate the between-study standard deviation from the data alone. Objectives: The aim is to propose a framework for eliciting an informative prior distribution for the between-study standard deviation in a Bayesian random effects meta-analysis model to genuinely represent heterogeneity when data are sparse. Methods: We developed an elicitation method using external information such as empirical evidence and experts' beliefs on the 'range' of treatment effects in order to infer the prior distribution for the between-study standard deviation. We also developed the method to be implemented in R. Results: The three-stage elicitation approach allows uncertainty to be represented by a genuine prior distribution to avoid making misleading inferences. It is flexible to what judgments an expert can provide, and is applicable to all types of outcome measure for which a treatment effect can be constructed on an additive scale. Conclusions: The choice between using a fixed effect or random effects meta-analysis model depends on the inferences required and not on the number of available studies. Our elicitation framework captures external evidence about heterogeneity and overcomes the often implausible assumption that studies are estimating the same treatment effect, thereby improving the quality of inferences in decision making.