A random effects meta‐analysis combines the results of several independent studies to summarise the evidence about a particular measure of interest, such as a treatment effect. The approach allows for unexplained between‐study heterogeneity in the true treatment effect by incorporating random study effects about the overall mean. The variance of the mean effect estimate is conventionally calculated by assuming that the between study variance is known; however, it has been demonstrated that this approach may be inappropriate, especially when there are few studies. Alternative methods that aim to account for this uncertainty, such as Hartung–Knapp, Sidik–Jonkman and Kenward–Roger, have been proposed and shown to improve upon the conventional approach in some situations. In this paper, we use a simulation study to examine the performance of several of these methods in terms of the coverage of the 95% confidence and prediction intervals derived from a random effects meta‐analysis estimated using restricted maximum likelihood. We show that, in terms of the confidence intervals, the Hartung–Knapp correction performs well across a wide‐range of scenarios and outperforms other methods when heterogeneity was large and/or study sizes were similar. However, the coverage of the Hartung–Knapp method is slightly too low when the heterogeneity is low (I 2 < 30%) and the study sizes are quite varied. In terms of prediction intervals, the conventional approach is only valid when heterogeneity is large (I 2 > 30%) and study sizes are similar. In other situations, especially when heterogeneity is small and the study sizes are quite varied, the coverage is far too low and could not be consistently improved by either increasing the number of studies, altering the degrees of freedom or using variance inflation methods. Therefore, researchers should be cautious in deriving 95% prediction intervals following a frequentist random‐effects meta‐analysis until a more reliable solution is identified. © 2016 The Authors. Statistics in Medicine Published by John Wiley & Sons Ltd.

No abstract

SummaryBackgroundInfections acquired in hospital are an important cause of morbidity and mortality in very preterm infants. Several small trials have suggested that supplementing the enteral diet of very preterm infants with lactoferrin, an antimicrobial protein processed from cow's milk, prevents infections and associated complications. The aim of this large randomised controlled trial was to collect data to enhance the validity and applicability of the evidence from previous trials to inform practice.MethodsIn this randomised placebo-controlled trial, we recruited very preterm infants born before 32 weeks' gestation in 37 UK hospitals and younger than 72 h at randomisation. Exclusion criteria were presence of a severe congenital anomaly, anticipated enteral fasting for longer than 14 days, or no realistic prospect of survival. Eligible infants were randomly assigned (1:1) to receive either enteral bovine lactoferrin (150 mg/kg per day; maximum 300 mg/day; lactoferrin group) or sucrose (same dose; control group) once daily until 34 weeks' postmenstrual age. Web-based randomisation minimised for recruitment site, gestation (completed weeks), sex, and single versus multifetal pregnancy. Parents, caregivers, and outcome assessors were unaware of group assignment. The primary outcome was microbiologically confirmed or clinically suspected late-onset infection (occurring >72 h after birth), which was assessed in all participants for whom primary outcome data was available by calculating the relative risk ratio with 95% CI between the two groups. The trial is registered with the International Standard Randomised Controlled Trial Number 88261002.FindingsWe recruited 2203 participants between May 7, 2014, and Sept 28, 2017, of whom 1099 were assigned to the lactoferrin group and 1104 to the control group. Four infants had consent withdrawn or unconfirmed, leaving 1098 infants in the lactoferrin group and 1101 in the sucrose group. Primary outcome data for 2182 infants (1093 [99·5%] of 1098 in the lactoferrin group and 1089 [99·0] of 1101 in the control group) were available for inclusion in the modified intention-to-treat analyses. 316 (29%) of 1093 infants in the intervention group acquired a late-onset infection versus 334 (31%) of 1089 in the control group. The risk ratio adjusted for minimisation factors was 0·95 (95% CI 0·86–1·04; p=0·233). During the trial there were 16 serious adverse events for infants in the lactoferrin group and 10 for infants in the control group. Two events in the lactoferrin group (one case of blood in stool and one death after intestinal perforation) were assessed as being possibly related to the trial intervention.InterpretationEnteral supplementation with bovine lactoferrin does not reduce the risk of late-onset infection in very preterm infants. These data do not support its routine use to prevent late-onset infection and associated morbidity or mortality in very preterm infants.FundingUK National Institute for Health Research Health Technology Assessment programme (10/57/49).

Identification as a study of diagnostic accuracy using at least one measure of accuracy (such as sensitivity, specificity, predictive values, or AUC) 1, 5

Summary Background Risk factors for maternal infection are clearly recognised, including caesarean section and operative vaginal birth. Antibiotic prophylaxis at caesarean section is widely recommended because there is clear systematic review evidence that it reduces incidence of maternal infection. Current WHO guidelines do not recommend routine antibiotic prophylaxis for women undergoing operative vaginal birth because of insufficient evidence of effectiveness. We aimed to investigate whether antibiotic prophylaxis prevented maternal infection after operative vaginal birth. Methods In a blinded, randomised controlled trial done at 27 UK obstetric units, women (aged ≥16 years) were allocated to receive a single dose of intravenous amoxicillin and clavulanic acid or placebo (saline) following operative vaginal birth at 36 weeks gestation or later. The primary outcome was confirmed or suspected maternal infection within 6 weeks of delivery defined by a new prescription of antibiotics for specific indications, confirmed systemic infection on culture, or endometritis. We did an intention-to-treat analysis. This trial is registered with ISRCTN, number 11166984, and is closed to accrual. Findings Between March 13, 2016, and June 13, 2018, 3427 women were randomly assigned to treatment: 1719 to amoxicillin and clavulanic acid, and 1708 to placebo. Seven women withdrew, leaving 1715 in the amoxicillin and clavulanic acid group and 1705 in the placebo groups. Primary outcome data were missing for 195 (6%) women. Significantly fewer women allocated to amoxicillin and clavulanic acid had a confirmed or suspected infection (180 [11%] of 1619) than women allocated to placebo (306 [19%] of 1606; risk ratio 0·58, 95% CI 0·49–0·69; p<0·0001). One woman in the placebo group reported a skin rash and two women in the amoxicillin and clavulanic acid reported other allergic reactions, one of which was reported as a serious adverse event. Two other serious adverse events were reported, neither was considered causally related to the treatment. Interpretation This trial shows benefit of a single dose of prophylactic antibiotic after operative vaginal birth and guidance from WHO and other national organisations should be changed to reflect this. Funding NIHR Health Technology Assessment programme.

Background Interferon gamma release assays (IGRAs) are blood tests recommended for the diagnosis of tuberculosis (TB) infection. There is currently uncertainty about the role and clinical utility of IGRAs in the diagnostic workup of suspected active TB in routine NHS clinical practice. Objectives To compare the diagnostic accuracy and cost-effectiveness of T-SPOT.TB ® (Oxford Immunotec, Abingdon, UK) and QuantiFERON® TB GOLD In-Tube (Cellestis, Carnegie, VIC, Australia) for diagnosis of suspected active TB and to estimate the diagnostic accuracy of second-generation IGRAs. Design Prospective within-patient comparative diagnostic accuracy study. Setting Secondary care. Participants Adults (aged ≥ 16 years) presenting as inpatients or outpatients at 12 NHS hospital trusts in London, Slough, Oxford, Leicester and Birmingham with suspected active TB. Interventions The index tests [T-SPOT.TB and QuantiFERON GOLD In-Tube (QFT-GIT)] and new enzyme-linked immunospot assays utilising novel Mycobacterium tuberculosis antigens (Rv3615c, Rv2654, Rv3879c and Rv3873) were verified against a composite reference standard applied by a panel of clinical experts blinded to IGRA results. Main outcome measures Sensitivity, specificity, predictive values and likelihood ratios were calculated to determine diagnostic accuracy. A decision tree model was developed to calculate the incremental costs and incremental health utilities [quality-adjusted life-years (QALYs)] of changing from current practice to using an IGRA as an initial rule-out test. Results A total of 363 patients had active TB (culture-confirmed and highly probable TB cases), 439 had no active TB and 43 had an indeterminate final diagnosis. Comparing T-SPOT.TB and QFT-GIT, the sensitivities [95% confidence interval (CI)] were 82.3% (95% CI 77.7% to 85.9%) and 67.3% (95% CI 62.1% to 72.2%), respectively, whereas specificities were 82.6% (95% CI 78.6% to 86.1%) and 80.4% (95% CI 76.1% to 84.1%), respectively. T-SPOT.TB was more sensitive than QFT-GIT (relative sensitivity 1.22, 95% CI 1.14 to 1.31; p < 0.001), but the specificities were similar (relative specificity 1.02, 95% CI 0.97 to 1.08; p = 0.3). For both IGRAs the sensitivity was lower and the specificity was higher for human immunodeficiency virus (HIV)-positive than for HIV-negative patients. The most promising novel antigen was Rv3615c. The added value of Rv3615c to T-SPOT.TB was a 9% (95% CI 5% to 12%) relative increase in sensitivity at the expense of specificity, which had a relative decrease of 7% (95% CI 4% to 10%). The use of current IGRA tests for ruling out active TB is unlikely to be considered cost-effective if a QALY was valued at £20,000 or £30,000. For T-SPOT.TB, the probability of being cost-effective for a willingness to pay of £20,000/QALY was 26% and 21%, when patients with indeterminate test results were excluded or included, respectively. In comparison, the QFT-GIT probabilities were 8% and 6%. Although the use of IGRAs is cost saving, the health detriment is large owing to delay in diagnosing active TB, leading to prolonged illness. There was substantial between-patient variation in the tests used in the diagnostic pathway. Limitations The recruitment target for the HIV co-infected population was not achieved. Conclusions Although T-SPOT.TB was more sensitive than QFT-GIT for the diagnosis of active TB, the tests are insufficiently sensitive for ruling out active TB in routine clinical practice in the UK. Novel assays offer some promise. Future work The novel assays require evaluation in distinct clinical settings and in immunosuppressed patient groups. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and the NIHR Health Protection Research Unit in Respiratory Infections, Imperial College London, London, UK.

Background Infections acquired in hospital are an important cause of morbidity and mortality in very preterm infants. Several small trials have suggested that supplementing the enteral diet of very preterm infants with lactoferrin, an antimicrobial protein processed from cow’s milk, prevents infections and associated complications. Objective To determine whether or not enteral supplementation with bovine lactoferrin (The Tatua Cooperative Dairy Company Ltd, Morrinsville, New Zealand) reduces the risk of late-onset infection (acquired > 72 hours after birth) and other morbidity and mortality in very preterm infants. Design Randomised, placebo-controlled, parallel-group trial. Randomisation was via a web-based portal and used an algorithm that minimised for recruitment site, weeks of gestation, sex and single versus multiple births. Setting UK neonatal units between May 2014 and September 2017. Participants Infants born at < 32 weeks’ gestation and aged < 72 hours at trial enrolment. Interventions Eligible infants were allocated individually (1 : 1 ratio) to receive enteral bovine lactoferrin (150 mg/kg/day; maximum 300 mg/day) or sucrose (British Sugar, Peterborough, UK) placebo (same dose) once daily from trial entry until a postmenstrual age of 34 weeks. Parents, caregivers and outcome assessors were unaware of group assignment. Outcomes Primary outcome – microbiologically confirmed or clinically suspected late-onset infection. Secondary outcomes – microbiologically confirmed infection; all-cause mortality; severe necrotising enterocolitis (NEC); retinopathy of prematurity (ROP); bronchopulmonary dysplasia (BPD); a composite of infection, NEC, ROP, BPD and mortality; days of receipt of antimicrobials until 34 weeks’ postmenstrual age; length of stay in hospital; and length of stay in intensive care, high-dependency and special-care settings. Results Of 2203 enrolled infants, primary outcome data were available for 2182 infants (99%). In the intervention group, 316 out of 1093 (28.9%) infants acquired a late-onset infection versus 334 out of 1089 (30.7%) infants in the control group [adjusted risk ratio (RR) 0.95, 95% confidence interval (CI) 0.86 to 1.04]. There were no significant differences in any secondary outcomes: microbiologically confirmed infection (RR 1.05, 99% CI 0.87 to 1.26), mortality (RR 1.05, 99% CI 0.66 to 1.68), NEC (RR 1.13, 99% CI 0.68 to 1.89), ROP (RR 0.89, 99% CI 0.62 to 1.28), BPD (RR 1.01, 99% CI 0.90 to 1.13), or a composite of infection, NEC, ROP, BPD and mortality (RR 1.01, 99% CI 0.94 to 1.08). There were no differences in the number of days of receipt of antimicrobials, length of stay in hospital, or length of stay in intensive care, high-dependency or special-care settings. There were 16 reports of serious adverse events for infants in the lactoferrin group and 10 for infants in the sucrose group. Conclusions Enteral supplementation with bovine lactoferrin does not reduce the incidence of infection, mortality or other morbidity in very preterm infants. Future work Increase the precision of the estimates of effect on rarer secondary outcomes by combining the data in a meta-analysis with data from other trials. A mechanistic study is being conducted in a subgroup of trial participants to explore whether or not lactoferrin supplementation affects the intestinal microbiome and metabolite profile of very preterm infants. Trial registration Current Controlled Trials ISRCTN88261002. 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. 22, No. 74. See the NIHR Journals Library website for further project information. This trial was also sponsored by the University of Oxford, Oxford, UK. The funder provided advice and support and monitored study progress but did not have a role in study design or data collection, analysis and interpretation.

Objective: The objective of this study was to examine methodological and reporting characteristics of systematic reviews and metaanalyses which compare diagnostic test accuracy (DTA) of multiple index tests, identify good practice, and develop guidance for better reporting.Study Design and Setting: Methodological survey of 127 comparative or multiple tests reviews published in 74 different general medical and specialist journals. We summarized methods and reporting characteristics that are likely to differ between reviews of a single test and comparative reviews. We then developed guidance to enhance reporting of test comparisons in DTA reviews.Results: Of 127 reviews, 16 (13%) reviews restricted study selection and test comparisons to comparative accuracy studies while the remaining 111 (87%) reviews included any study type. Fifty-three reviews (42%) statistically compared test accuracy with only 18 (34%) of these using recommended methods. Reporting of several itemsdin particular the role of the index tests, test comparison strategy, and limitations of indirect comparisons (i.e., comparisons involving any study type)dwas deficient in many reviews. Five reviews with exemplary methods and reporting were identified.Conclusion: Reporting quality of reviews which evaluate and compare multiple tests is poor. The guidance developed, complemented with the exemplars, can assist review authors in producing better quality comparative reviews.