Search

Healthcare commissioners and those making reimbursement decisions require data on the “natural history” of diseases to estimate quantities such as life expectancy and quality of life. However, such data are commonly acquired from unrepresentative sources like trial placebo arms or case series. We used Scottish Diabetes Research Network-National Diabetes Dataset to characterise life expectancy for people with type 2 diabetes mellitus (T2DM). Data from >99% of people with diabetes in Scotland are available from primary and secondary care (including from clinical investigations) and can be linked to hospital admission and death records for research purposes. As part of an MRC-funded (MR/T017112/1) systematic review and modelling study (PROSPERO registration CRD42020184174) building on work funded by the Chief Scientific Office for Scotland (CSO HIPS_17_26), we used these linked data to define a treatment-eligible population for three drug classes — SGLT-2 inhibitors, GLP-1 agonists, and DPP-4 inhibitors. Fitting a Gompertz model, we estimated life expectancy for this ‘treatment-eligible’ group. The treatment-eligible population, compared to the entire Scottish T2DM population, was younger (63.8 (12.1) vs. 66.7 (12.7) year-old), with larger proportions of men (60.6% vs. 56.2%), higher HbA1c levels (67.4 (13.1) vs. 60.2 (15.1) mmol/L), larger proportions prescribed metformin (69.2% vs 55.2%) and higher life expectancy at age 65 (in women 21 vs 19 years and in men 20 vs 18 years). Using linked data to define and characterise treatment-eligible populations is feasible and may impact decision models. These findings will be useful to those making resourcing decisions and future economic analyses.

Background Liver cirrhosis is the largest risk factor for developing hepatocellular carcinoma (HCC), and surveillance is therefore recommended among this population. Current guidance recommends surveillance with ultrasound, with or without alpha-fetoprotein (AFP). This review is part of a larger project looking at benefits, harms and costs of surveillance for HCC in people with cirrhosis. It aims to synthesise the evidence on the diagnostic accuracy of imaging or biomarker tests, alone or in combination, to identify HCC in adults with liver cirrhosis in a surveillance programme. Methods We will identify studies through a 2021 Cochrane review with similar eligibility criteria, and a database search of MEDLINE, Embase and the Cochrane Database of Systematic Reviews. We will include diagnostic test accuracy studies with adult cirrhosis patients of any aetiology. Studies must assess at least one of the following index tests: ultrasound (US), magnetic resonance imaging (MRI), computerised tomography (CT), alpha-fetoprotein (AFP), des-gamma-carboxyprothrombin (DCP), lens culinaris agglutinin-reactive fraction of AFP (AFP-L3), a genomic biomarker, or a diagnostic prediction model incorporating at least one of the above-mentioned tests. We will assess studies for risk of bias using QUADAS-2 and QUADAS-C. We will combine data using bivariate random effects meta-analyses. For tests evaluated across varying diagnostic thresholds, we will produce pooled estimates of sensitivity and specificity across the full range of numerical thresholds, where possible. Where sufficient studies compare two or more index tests, we will perform additional analyses to compare the accuracy of different tests. Where feasible, we will stratify all meta-analyses by tumour size and patient characteristics, including cirrhosis aetiology and liver disease severity. Discussion This review will synthesise evidence across the full range of possible surveillance tests, using advanced statistical methods to summarise accuracy across all thresholds and to compare the accuracy of different tests. PROSPERO registration CRD42022357163

Introduction Participants in randomised controlled trials (trials) are generally younger and healthier than many individuals encountered in clinical practice. Consequently, the applicability of trial findings is often uncertain. To address this, results from trials can be calibrated to more representative data sources. In a network meta-analysis, using a novel approach which allows the inclusion of trials whether or not individual-level participant data (IPD) is available, we will calibrate trials for three drug classes (sodium glucose cotransporter 2 (SGLT2) inhibitors, glucagon-like peptide-1 (GLP1) receptor analogues and dipeptidyl peptidase-4 (DPP4) inhibitors) to the Scottish diabetes register.Methods and analysis Medline and EMBASE databases, the US clinical trials registry (clinicaltrials.gov) and the Chinese Clinical Trial Registry (chictr.org.cn) will be searched from 1 January 2002. Two independent reviewers will apply eligibility criteria to identify trials for inclusion. Included trials will be phase 3 or 4 trials of SGLT2 inhibitors, GLP1 receptor analogues or DPP4 inhibitors, with placebo or active comparators, in participants with type 2 diabetes, with at least one of glycaemic control, change in body weight or major adverse cardiovascular event as outcomes. Unregistered trials will be excluded.We have identified a target population from the population-based Scottish diabetes register. The chosen cohort comprises people in Scotland with type 2 diabetes who either (1) require further treatment due to poor glycaemic control where any of the three drug classes may be suitable, or (2) who have adequate glycaemic control but are already on one of the three drug classes of interest or insulin.Ethics and dissemination Ethical approval for IPD use was obtained from the University of Glasgow MVLS College Ethics Committee (Project: 200160070). The Scottish diabetes register has approval from the Scottish A Research Ethics Committee (11/AL/0225) and operates with Public Benefit and Privacy Panel for Health and Social Care approval (1617-0147).PROSPERO registration number CRD42020184174.

Introduction Knee replacements are highly successful for many people, but if a knee replacement fails, revision surgery is generally required. Surgeons and patients may choose from a range of implant components and combinations that make up knee replacement constructs, all with potential implications for how long a knee replacement will last. To inform surgeon and patient decisions, a comprehensive synthesis of data from randomised controlled trials is needed to evaluate the effects of different knee replacement implants on overall construct survival. Due to limited follow-up in trials, joint registry analyses are also needed to assess the long-term survival of constructs. Finally, economic modelling can identify cost-effective knee replacement constructs for different patient groups.Methods and analysis In this protocol, we describe systematic reviews and network meta-analyses to synthesise evidence on the effectiveness of knee replacement constructs used in total and unicompartmental knee replacement and analyses of two national joint registries to assess long-term outcomes. Knee replacement constructs are defined by bearing materials and mobility, constraint, fixation and patella resurfacing. For men and women in different age groups, we will compare the lifetime cost-effectiveness of knee replacement constructs.Ethics and dissemination Systematic reviews are secondary analyses of published data with no ethical approval required. We will design a common joint registry analysis plan and provide registry representatives with information for submission to research or ethics committees. The project has been assessed by the National Health Service (NHS) REC committee and does not require ethical review.Study findings will be disseminated to clinicians, researchers and administrators through open access articles, presentations and websites. Specific UK-based groups will be informed of results including National Institute for Health Research and National Institute for Health and Care Excellence, as well as international orthopaedic associations and charities. Effective dissemination to patients will be guided by our patient–public involvement group and include written lay summaries and infographics.PROSPERO registration number CRD42019134059 and CRD42019138015.

Background: Most surgical incisions heal by primary intention (i.e. wound edges are apposed with sutures, clips or glue); however, some heal by secondary intention (i.e. the wound is left open and heals by formation of granulation tissue). There is, however, a lack of evidence regarding the epidemiology, management and impact on patients’ quality of life of these surgical wounds healing by secondary intention, resulting in uncertainty regarding effective treatments and difficulty in planning care and research. Objectives: To derive a better understanding of the nature, extent, costs, impact and outcomes of surgical wounds healing by secondary intention, effective treatments, and the value and nature of further research. Design: Cross-sectional survey; inception cohort; cost-effectiveness and value of implementation analyses; qualitative interviews; and pilot, feasibility randomised controlled trial. Setting: Acute and community care settings in Leeds and Hull, Yorkshire, UK. Participants: Adults (or for qualitative interviews, patients or practitioners) with previous experience of a surgical wound healing by secondary intention. Inclusion criteria varied between the individual workstreams. Interventions: The pilot, feasibility randomised controlled trial compared negative-pressure wound therapy – a device applying a controlled vacuum to a wound via a dressing – with usual care (no negative-pressure wound therapy). Results: Survey data estimated that treated surgical wounds healing by secondary intention have a point prevalence of 4.1 per 10,000 population (95% confidence interval 3.5 to 4.7 per 10,000 population). Surgical wounds healing by secondary intention most frequently occurred following colorectal surgery (n = 80, 42.8% cross-sectional survey; n = 136, 39.7% inception cohort) and were often planned before surgery (n = 89, 47.6% cross-sectional survey; n = 236, 60.1% inception cohort). Wound care was frequently delivered in community settings (n = 109, 58.3%) and most patients (n = 184, 98.4%) received active wound treatment. Cohort data identified hydrofibre dressings (n = 259, 65.9%) as the most common treatment, although 29.3% (n = 115) of participants used negative-pressure wound therapy at some time during the study. Surgical wounds healing by secondary intention occurred in 81.4% (n = 320) of participants at a median of 86 days (95% confidence interval 75 to 103 days). Baseline wound area (p ≤ 0.01), surgical wound contamination (determined during surgery; p = 0.04) and wound infection at any time (p ≤ 0.01) (i.e. at baseline or postoperatively) were found to be predictors of prolonged healing. Econometric models, using observational, cohort study data, identified that, with little uncertainty, negative-pressure wound therapy treatment is more costly and less effective than standard dressing treatment for the healing of open surgical wounds. Model A (ordinary least squares with imputation) effectiveness: 73 days longer than those who did not receive negative-pressure wound therapy (95% credible interval 33.8 to 112.8 days longer). Model A cost-effectiveness (associated incremental quality-adjusted life-years): observables –0.012 (standard error 0.005) and unobservables –0.008 (standard error 0.011). Model B (two-stage model, logistic and linear regression) effectiveness: 46 days longer than those who did not receive negative-pressure wound therapy (95% credible interval 19.6 to 72.5 days longer). Model B cost-effectiveness (associated incremental quality-adjusted life-years): observables –0.007 (standard error 0.004) and unobservables –0.027 (standard error 0.017). Patient interviews (n = 20) identified initial reactions to surgical wounds healing by secondary intention of shock and disbelief. Impaired quality of life characterised the long healing process, with particular impact on daily living for patients with families or in paid employment. Patients were willing to try any treatment promising wound healing. Health professionals (n = 12) had variable knowledge of surgical wound healing by secondary intention treatments and, frequently, favoured negative-pressure wound therapy, despite the lack of robust evidence. The pilot feasibility randomised controlled trial screened 248 patients for eligibility and subsequently recruited and randomised 40 participants to receive negative-pressure wound therapy or usual care (no negative-pressure wound therapy). Data indicated that it was feasible to complete a full randomised controlled trial to provide definitive evidence for the clinical effectiveness and cost-effectiveness of negative-pressure wound therapy as a treatment for surgical wounds healing by secondary intention. Key elements and recommendations for a larger randomised controlled trial were identified. Limitations: This research programme was conducted in a single geographical area (i.e. Yorkshire and the Humber, UK) and local guidelines and practices may have affected treatment availability, and so may not represent UK-wide treatment choices. A wide range of wound types were included; however, some wound types may be under-represented, meaning that this research may not represent the overall surgical wound healing by secondary intention population. The lack of randomised controlled trial data on the relative effects of negative-pressure wound therapy in surgical wounds healing by secondary intention resulted in much of the economic modelling being based on observational data. Observational data, even with extensive adjustment, do not negate the potential for unresolved confounding to affect the results, which can reduce confidence in conclusions drawn from observational data. Definitive evidence from a randomised controlled trial may be the only way to overcome this lack of confidence. Conclusions: This research has provided new information regarding the nature, extent, costs, impacts and outcomes of surgical wounds healing by secondary intention, treatment effectiveness, and the value and nature of future research, while addressing previous uncertainties regarding the problem of surgical wounds healing by secondary intention. Aspects of our research indicate that negative-pressure wound therapy is more costly and less effective than standard dressing for the healing of open surgical wounds. However, because this conclusion is based solely on observational data, it may be affected by unresolved confounding. Should a future randomised controlled trial be considered necessary, its design should reflect careful consideration of the findings of this programme of research. Future work: This research signals the importance of further research on surgical wound healing by secondary intention. Key research questions raised by this programme of research include (1) which treatments are clinically effective and cost-effective for surgical wound healing by secondary intention for all patients or for particular patient subgroups? (2) Can particular prognostic factors predict time to healing of surgical wound healing by secondary intention? And (3) do psychosocial interventions have the potential to improve quality of life in people with hard-to-heal surgical wound healing by secondary intention? Given that negative-pressure wound therapy has been widely adopted, with relatively little evidence to support its use, the design and outcomes of a randomised controlled trial would need to be carefully considered. We focused in this research on wound healing, and maintain, based on the findings of patient interviews, that this is a key outcome for future research. Impacts of negative-pressure wound therapy on outcomes such as infection and reoperation should also be considered, as should patients’ views of the treatment. The type of patient group recruited and the outcomes of interest will all influence the duration of follow-up of any planned study. The comparator in any future study will also need careful consideration. Trial registration: Current Controlled Trials ISRCTN12761776. Funding: This project was funded by the National Institute for Health Research Programme Grants for Applied Research programme and will be published in full in Programme Grants for Applied Research; Vol. 8, No. 7. See the National Institute for Health Research Journals Library website for further project information.

Background: Despite a growing number of prescriptions for antidepressants (over 70 million in 2018), there is uncertainty about when people with depression might benefit from antidepressant medication and concern that antidepressants are prescribed unnecessarily. Objectives: The main objective of the PANDA (What are the indications for Prescribing ANtiDepressAnts that will lead to a clinical benefit?) research programme was to provide more guidance about when antidepressants are likely to benefit people with depression. We aimed to estimate the minimal clinically important difference for commonly used self-administered scales for depression and anxiety, and to understand more about how patients respond to such assessments. We carried out an observational study of patients with depressive symptoms and a placebo-controlled randomised controlled trial of sertraline versus placebo to estimate the treatment effect in UK primary care. The hypothesis was that the severity and duration of symptoms were related to treatment response. Design: The programme consisted of three phases. The first phase relied on the secondary analysis of existing data extracted from published trials. The second phase was the PANDA cohort study of patients with depressive symptoms who presented to primary care and were followed up 2, 4 and 6 weeks after a baseline assessment. Both quantitative and qualitative methods were used in the analysis. The third phase was a multicentre randomised placebo-controlled double-blind trial of sertraline versus placebo in patients presenting to primary care with depressive symptoms. Setting: UK primary care in Bristol, London, Liverpool and York. Participants: Patients aged 18–74 years who were experiencing depressive symptoms in primary care. Eligibility for the PANDA randomised controlled trial included that there was uncertainty about the benefits about treatment with an antidepressant. Interventions: In the PANDA randomised controlled trial, patients were individually randomised to 100 mg daily of sertraline or an identical placebo. The PANDA cohort study was an observational study. Main outcome measures: Depressive symptoms measured using the Patient Health Questionnaire were the primary outcome for the randomised controlled trial. Other outcomes included anxiety symptoms using the Generalised Anxiety Disorder-7; depressive symptoms using the Beck Depression Inventory, version 2; health-related quality of life; self-reported improvement; and cost-effectiveness. Results: The secondary analysis of existing randomised controlled trials [GENetic and clinical Predictors Of treatment response in Depression (GenPod), TREAting Depression with physical activity (TREAD) and Clinical effectiveness and cost-effectiveness of cognitive Behavioural Therapy as an adjunct to pharmacotherapy for treatment-resistant depression in primary care (CoBalT)] found evidence that the minimal clinically important difference increased as the initial severity of depressive symptoms rose. Our estimates of minimal clinically important difference were a 17% and 18% reduction in Beck Depression Inventory scores for GenPod and TREAD, respectively. In CoBalT, a 32% reduction corresponded to the minimal clinically important difference but the participants in this study had depression that had not responded to antidepressants. In the PANDA study cohort, and from our analyses in existing data, we found that the minimal clinically important difference varies considerably with the initial severity of depressive and anxiety symptoms. Expressing the minimal clinically important difference as a percentage reduction reduces this variation at higher scores, but at low scores the percentage reduction increased substantially. The results from the qualitative studies pointed out many limitations of the Patient Health Questionnaire-9 items in assessing change and recovery from depression. In the PANDA randomised controlled trial, there was no evidence that sertraline resulted in a reduction in depressive symptoms within 6 weeks of randomisation, but there was some evidence of a reduction by 12 weeks. However, sertraline led to a reduction in anxiety symptoms, an improvement of mental health-related quality of life and an increased likelihood of reporting improvement. The mean Patient Health Questionnaire-9 items score at 6 weeks was 7.98 (standard deviation 5.63) in the sertraline group and 8.76 (standard deviation 5.86) in the placebo group (5% relative reduction, 95% confidence interval –7% to 15%; p = 0.41). Of the secondary outcomes, there was strong evidence that sertraline reduced anxiety symptoms (Generalised Anxiety Disorder-7 score reduced by 17% (95% confidence interval 9% to 25%; p = 0.00005). Sertraline had a high probability (> 90%) of being cost-effective at 12 weeks. The PANDA randomised controlled trial found no evidence that treatment response or cost-effectiveness was related to severity or duration of depressive symptoms. The minimal clinically important difference estimates suggested that sertraline’s effect on anxiety, but not on depression, was likely to be clinically important. Limitations: The results from the randomised controlled trial and the estimates of minimal clinically important difference were not sufficiently precise to provide specific clinical guidance for individuals. We had low power in testing whether or not initial severity and duration of depressive symptoms are related to treatment response. Conclusions: The results of the trial support the use of sertraline and probably other selective serotonin reuptake inhibitors because of their action in reducing anxiety symptoms and the likelihood of longer-term benefit on depressive symptoms. Sertraline could be prescribed for anxiety symptoms that commonly occur with depression and many patients will experience a clinical benefit. The Patient Health Questionnaire-9 items and similar self-administered scales should not be used on their own to assess clinical outcome, but should be supplemented with further clinical assessment. Future work: We need to examine the longer-term effects of antidepressant treatment. We need more precise estimates of the treatment effects and minimal clinically important difference at different severities to provide more specific guidance for individuals. However, the methods we have developed provide an approach towards providing such detailed guidance. Trial registration: Current Controlled Trials ISRCTN84544741 and EudraCT number 2013-003440-22. Funding: This project was funded by the National Institute for Health Research (NIHR) Programme Grants for Applied Research programme and will be published in full in Programme Grants for Applied Research; Vol. 7, No. 10. See the NIHR Journals Library website for further project information.

Background: Obsessive–compulsive disorder (OCD) is a relatively common and disabling condition. Objectives: To determine the clinical effectiveness, acceptability and cost-effectiveness of pharmacological and psychological interventions for the treatment of OCD in children, adolescents and adults. Data sources: We searched the Cochrane Collaboration Depression, Anxiety and Neurosis Trials Registers, which includes trials from routine searches of all the major databases. Searches were conducted from inception to 31 December 2014. Review methods: We undertook a systematic review and network meta-analysis (NMA) of the clinical effectiveness and acceptability of available treatments. Outcomes for effectiveness included mean differences in the total scores of the Yale–Brown Obsessive–Compulsive Scale or its children’s version and total dropouts for acceptability. For the cost-effectiveness analysis, we developed a probabilistic model informed by the results of the NMA. All analyses were performed using OpenBUGS version 3.2.3 (members of OpenBUGS Project Management Group; see www.openbugs.net). Results: We included 86 randomised controlled trials (RCTs) in our systematic review. In the NMA we included 71 RCTs (54 in adults and 17 in children and adolescents) for effectiveness and 71 for acceptability (53 in adults and 18 in children and adolescents), comprising 7643 and 7942 randomised patients available for analysis, respectively. In general, the studies were of medium quality. The results of the NMA showed that in adults all selective serotonin reuptake inhibitors (SSRIs) and clomipramine had greater effects than drug placebo. There were no differences between SSRIs, and a trend for clomipramine to be more effective did not reach statistical significance. All active psychological therapies had greater effects than drug placebo. Behavioural therapy (BT) and cognitive therapy (CT) had greater effects than psychological placebo, but cognitive–behavioural therapy (CBT) did not. BT and CT, but not CBT, had greater effects than medications, but there are considerable uncertainty and methodological limitations that should be taken into account. In children and adolescents, CBT and BT had greater effects than drug placebo, but differences compared with psychological placebo did not reach statistical significance. SSRIs as a class showed a trend for superiority over drug placebo, but the difference did not reach statistical significance. However, the superiority of some individual drugs (fluoxetine, sertraline) was marginally statistically significant. Regarding acceptability, all interventions except clomipramine had good tolerability. In adults, CT and BT had the highest probability of being most cost-effective at conventional National Institute for Health and Care Excellence thresholds. In children and adolescents, CBT or CBT combined with a SSRI were more likely to be cost-effective. The results are uncertain and sensitive to assumptions about treatment effect and the exclusion of trials at high risk of bias. Limitations: The majority of psychological trials included patients who were taking medications. There were few studies in children and adolescents. Conclusions: In adults, psychological interventions, clomipramine, SSRIs or combinations of these are all effective, whereas in children and adolescents, psychological interventions, either as monotherapy or combined with specific SSRIs, were more likely to be effective. Future RCTs should improve their design, in particular for psychotherapy or combined interventions. Study registration: The study is registered as PROSPERO CRD42012002441. Funding details: The National Institute for Health Research Health Technology Assessment programme.

Background: Joint hypermobility syndrome (JHS) is a heritable disorder associated with laxity and pain in multiple joints. Physiotherapy is the mainstay of treatment, but there is little research investigating its clinical effectiveness. Objectives: To develop a comprehensive physiotherapy intervention for adults with JHS; to pilot the intervention; and to conduct a pilot randomised controlled trial (RCT) to determine the feasibility of conducting a future definitive RCT. Design: Patients’ and health professionals’ perspectives on physiotherapy for JHS were explored in focus groups (stage 1). A working group of patient research partners, clinicians and researchers used this information to develop the physiotherapy intervention. This was piloted and refined on the basis of patients’ and physiotherapists’ feedback (stage 2). A parallel two-arm pilot RCT compared ‘advice’ with ‘advice and physiotherapy’ (stage 3). Random allocation was via an automated randomisation service, devised specifically for the study. Owing to the nature of the interventions, it was not possible to blind clinicians or patients to treatment allocation. Setting: Stage 1 – focus groups were conducted in four UK locations. Stages 2 and 3 – piloting of the intervention and the pilot RCT were conducted in two UK secondary care NHS trusts. Participants: Stage 1 – patient focus group participants (n = 25, three men) were aged > 18 years, had a JHS diagnosis and had received physiotherapy within the preceding 12 months. The health professional focus group participants (n = 16, three men; 14 physiotherapists, two podiatrists) had experience of managing JHS. Stage 2 – patient participants (n = 8) were aged > 18 years, had a JHS diagnosis and no other musculoskeletal conditions causing pain. Stage 3 – patient participants for the pilot RCT (n = 29) were as for stage 2 but the lower age limit was 16 years. Intervention: For the pilot RCT (stage 3) the advice intervention was a one-off session, supplemented by advice booklets. All participants could ask questions specific to their circumstances and receive tailored advice. Participants were randomly allocated to ‘advice’ (no further advice or physiotherapy) or ‘advice and physiotherapy’ (an additional six 30-minute sessions over 4 months). The physiotherapy intervention was supported by a patient handbook and was delivered on a one-to-one patient–therapist basis. It aimed to increase patients’ physical activity through developing knowledge, understanding and skills to better manage their condition. Main outcome measures: Data from patient and health professional focus groups formed the main outcome from stage 1. Patient and physiotherapist interview data also formed a major component of stages 2 and 3. The primary outcome in stage 3 related to the feasibility of a future definitive RCT [number of referrals, recruitment and retention rates, and an estimate of the value of information (VOI) of a future RCT]. Secondary outcomes included clinical measures (physical function, pain, global status, self-reported joint count, quality of life, exercise self-efficacy and adverse events) and resource use (to estimate cost-effectiveness). Outcomes were recorded at baseline, 4 months and 7 months. Results: Stage 1 – JHS is complex and unpredictable. Physiotherapists should take a long-term holistic approach rather than treating acutely painful joints in isolation. Stage 2 – a user-informed physiotherapy intervention was developed and evaluated positively. Stage 3 – recruitment to the pilot RCT was challenging, primarily because of a perceived lack of equipoise between advice and physiotherapy. The qualitative evaluation provided very clear guidance to inform a future RCT, including enhancement of the advice intervention. Some patients reported that the advice intervention was useful and the physiotherapy intervention was again evaluated very positively. The rate of return of questionnaires was low in the advice group but reasonable in the physiotherapy group. The physiotherapy intervention showed evidence of promise in terms of primary and secondary clinical outcomes. The advice arm experienced more adverse events. The VOI analysis indicated the potential for high value from a future RCT. Such a trial should form the basis of future research efforts. Conclusion: A future definitive RCT of physiotherapy for JHS seems feasible, although the advice intervention should be made more robust to address perceived equipoise and subsequent attrition. Trial registration: Current Controlled Trials ISRCTN29874209. Funding: This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 20, No. 47. See the NIHR Journals Library website for further project information.

Treatment by covariate interactions can be explored in reviews using interaction analyses (e.g., subgroup analysis). Such analyses can provide information on how the covariate modifies the treatment effect and is an important methodological approach for personalising medicine. Guidance exists regarding how to apply such analyses but little is known about whether authors follow the guidance.Using published recommendations, we developed criteria to assess how well interaction analyses were designed, applied, interpreted, and reported. The Cochrane Database of Systematic Reviews was searched (8th August 2013). We applied the criteria to the most recently published review, with an accessible protocol, for each Cochrane Review Group. We excluded review updates, diagnostic test accuracy reviews, withdrawn reviews, and overviews of reviews. Data were summarised regarding reviews, covariates, and analyses.Each of the 52 included reviews planned or did interaction analyses; 51 reviews (98%) planned analyses and 33 reviews (63%) applied analyses. The type of analysis planned and the type subsequently applied (e.g., sensitivity or subgroup analysis) was discrepant in 24 reviews (46%). No review reported how or why each covariate had been chosen; 22 reviews (42%) did state each covariate a priori in the protocol but no review identified each post-hoc covariate as such. Eleven reviews (21%) mentioned five covariates or less. One review reported planning to use a method to detect interactions (i.e., interaction test) for each covariate; another review reported applying the method for each covariate. Regarding interpretation, only one review reported whether an interaction was detected for each covariate and no review discussed the importance, or plausibility, of the results, or the possibility of confounding for each covariate.Interaction analyses in Cochrane Reviews can be substantially improved. The proposed criteria can be used to help guide the reporting and conduct of analyses.