Two new methods to fit models for network meta-analysis with random inconsistency effects

Efficacy and safety of alcohol reduction pharmacotherapy according to treatment duration in patients with alcohol dependence or alcohol use disorder: A systematic review and network meta-analysis

BACKGROUND AND AIMS

Relapse is common in alcohol dependence (AD) and alcohol use disorder (AUD), so alcohol reduction therapy should be measured over as long a period as possible; however, existing reviews do not consider the duration of treatment and therefore alcohol reduction therapy may not have been appropriately evaluated. This review evaluated the efficacy and safety of alcohol reduction pharmacotherapy in patients with AD or AUD according to the duration of treatment.

METHODS

We conducted a systematic review and network meta-analysis of randomized controlled trials (RCTs) that assessed 15 pharmacological agents. MEDLINE, Embase, PsycINFO, the Cochrane Central Register of Controlled Trials, the ClinicalTrials.gov and the International Clinical Trials Registry Platform were searched for eligible trials through to May 2021. Outcomes were heavy drinking days (HDD), total alcohol consumption (TAC), any adverse event and days without drinking.

RESULTS

Fifty-five RCTs (n = 8891) were included. Nalmefene was superior to placebo for reducing HDD (standard mean difference [SMD] -0.28, 95% confidence interval [CI] -0.37, -0.18) and TAC (SMD -0.25, 95% CI -0.35, -0.16) in the long-term, but not in the short-term. Topiramate was superior to placebo for reducing HDD (SMD -0.35, 95% CI -0.59, -0.12) and days without drinking (SMD 0.46, 95% CI 0.11, 0.82), and baclofen was superior for reducing TAC (SMD -0.70, 95% CI -1.29, -0.11), in the short-term. The frequency of adverse events was higher with nalmefene and topiramate than with placebo.

CONCLUSION

Nalmefene, topiramate and baclofen may be effective as alcohol reduction pharmacotherapy; however, only nalmefene has demonstrated long-term efficacy, and nalmefene and topiramate have a significantly higher frequency of adverse events compared with placebo.

A meta-analysis and real-world cohort study on the sex-related differences in efficacy and safety of immunotherapy for hepatocellular carcinoma

Background & Aims

Sex-related differences in the immune pathogenesis of hepatocellular carcinoma (HCC), particularly related to oestrogen-dependent secretion of pro-tumourigenic cytokines, are well-known. Whether sex influences the efficacy and safety of immunotherapy is not known.

Methods

We performed a restricted maximum likelihood random effects meta-analysis of five phase III trials that evaluated immune checkpoint inhibitors (ICIs) in advanced HCC and reported overall survival (OS) hazard ratios (HRs) stratified by sex to evaluate sex-related differences in OS. In a real-world cohort of 840 patients with HCC from 22 centres included between 2018 and 2023, we directly compared the efficacy and safety of atezolizumab + bevacizumab (A+B) between sexes. Radiological response was reported according to RECIST v1.1. Uni- and multivariable Cox regression analyses were performed for OS and progression-free survival (PFS).

Results

In the meta-analysis, immunotherapy was associated with a significant OS benefit only in male (pooled HR 0.79; 95% CI 0.73-0.86) but not in female (pooled HR 0.85; 95% CI 0.70-1.03) patients with HCC. When directly comparing model estimates, no differences in the treatment effect between sexes were observed. Among 840 patients, 677 (81%) were male (mean age 66 ± 11 years), and 163 (19%) were female (mean age 67 ± 12 years). Type and severity of adverse events were similar between the two groups. OS and PFS were comparable between males and females upon uni- and multivariable analyses (aHR for OS and PFS: 0.79, 95% CI 0.59-1.04; 1.02, 95% CI 0.80-1.30, respectively). Objective response rates (24%/22%) and disease control rates (59%/59%) were also similar between sexes.

Conclusion

Female phase III trial participants experienced smaller OS benefit following ICI therapy for advanced HCC, while outcomes following A+B treatment were comparable between sexes in a large real-world database. Based on the ambiguous sex-related differences in survival observed here, further investigation of sex-specific clinical and biologic determinants of responsiveness and survival following ICIs are warranted.

Impact and implications

While immune checkpoint inhibitors have emerged as standard of care for the treatment of hepatocellular carcinoma, there are conflicting reports on whether the efficacy of cancer immunotherapy differs between females and males. Our study suggests ambiguous sex-related differences in outcomes from immunotherapy in hepatocellular carcinoma. Further investigation of sex-specific clustering in clinicopathologic and immunologic determinants of responsiveness to immune checkpoint inhibitor therapy should be prioritised.

Systematic review registration

PROSPERO CRD42023429625.

Effects of different mind-body exercises on glucose and lipid metabolism in patients with type 2 diabetes: A network meta-analysis

BACKGROUND

This study aims to compare the relative effectiveness of different forms of mind-body exercise in improving glycolipid metabolism in patients with T2DM using a network meta-analysis.

METHODS

The relevant literature was systematically searched in Cochrane Library, Embase, PubMed, Web of Science, China National Knowledge Infrastructure (CNKI), Wanfang database and VIP Chinese Science. A network meta-analysis was conducted using a random-effects model within a frequentist framework.

RESULTS

A total of 52 studies with 6 modalities of mind-body exercise involving 4024 patients were included. Compared with controls, all exercise modalities except Yijinjing significantly improved patients' fasting blood glucose (MDs ranged from -0.72(95% CI: -1.20,-0.25) in dance to -2.06(95% CI: -2.59,-1.54) in yoga), glycated hemoglobin (MDs ranged from -0.62 (95% CI: -0.96) in taijiquan to -0.27,-0.27) in Pilates to -1.18(95% CI: 1.80,-0.57)), total cholesterol (MDs ranged from -0.73(95% CI: -1.29,-0.17) in Pilates to -0.41(95% CI: -0.62,-0.19) in Tai Chi), triglycerides (MDs ranged from -0.56 in Pilates (95% CI: -1.07,-0.04) for Pilates to -0.38 (95% CI: -0.67,-0.10)) for dance; only Tai Chi Chuan (MD:0.12 95% CI:0.04,0.20) and Baduanjin (MD:0.14 95% CI:0.06,0.22) significantly increased high-density lipoprotein cholesterol levels in patients; only yoga(MD: 0.78 95%CI: 1.20,-0.37) and Tai Chi Chuan(MD: 0.32 95%CI: 0.60,-0.04) significantly decreased high-density lipoprotein cholesterol levels in patients.

CONCLUSION

All mind-body exercises help to improve blood glucose and lipid levels in T2DM patients, but the ranking of relative effectiveness needs to be rationalized. Our study suggests that T2DM patients should choose appropriate mind-body exercises according to their conditions and stick to them for a long time under their healthcare professionals' guidance to achieve effective diabetes control.

Robotic colon surgery in obese patients: a systematic review and meta-analysis

BACKGROUND

Colon cancer resection can be technically difficult in the obese (OB) population. Robotic surgery is a promising technique but its benefits remain uncertain in OB patients. The aim of this study is to compare OB versus non-obese (NOB) patients undergoing robotic colon surgery, as well as OB patients undergoing robotic versus open or laparoscopic colonic surgery.

METHODS

A systematic review and meta-analysis was performed. Primary outcome measures included length of stay (LOS), surgical site infection (SSI) rate, complications, anastomotic leak and oncological outcomes.

RESULTS

A total of eight studies were included, with five comparing OB and NOB patients undergoing robotic colon surgery included in meta-analysis. A total of 263 OB patients and 400 NOB patients formed the sample for meta-analysis. There was no significant difference between the two groups in operative time, conversion to open, LOS, lymph node yield, anastomotic leak and postoperative ileus. There was a trend towards a significant increase in overall complications and SSI in the OB group (32.3% OB versus 26.8% NOB for complications, 14.2% OB versus 9.9% NOB for SSI). The three included studies comparing surgical techniques were too heterogeneous to undergo meta-analysis.

CONCLUSION

Robotic colon surgery is safe in obese patients, but high-quality prospective evidence is lacking. Future studies should report on oncological safety and the cost-effectiveness of adopting the robotic technique in these challenging patients.

A cumulative Bayesian network meta-analysis on the comparative efficacy of pharmacotherapies for mania over the last 40 years

RATIONALE

Mania (or manic episodes) is a common symptom of bipolar disorder and is frequently accompanied by hyperactivity and delusions; given the cost and resources available, there is a paucity of evidence for direct comparison of different drugs.

OBJECTIVES

We aimed to provide evidence-based recommendations on the efficacy of overall currently used pharmacological treatments for patients with acute bipolar mania.

METHOD

We conducted a systematic review and network meta-analysis (NMA) using a Bayesian network frame. We searched the primary literature databases without language restrictions until Dec 18, 2021, for reports of randomized controlled trials (RCTs) of suspected antimanic drugs used as monotherapy for patients with acute bipolar mania, with the primary outcomes being efficacy (mean difference (MD), standardized mean difference (SMD) in the change of mania score).

RESULTS

Eighty-seven studies were included in which 18,724 manic participants (mean age = 34.6 years, with 50.36% males) were allocated at random to one of 25 active medication drug therapies or placebo, resulting in 87 direct comparisons on 192 data points. Tamoxifen (- 22·00 [- 26·00 to - 18·00]) had the best efficacy over the placebo. Meanwhile, risperidone (- 6·60 [- 8·40 to - 4·90]) was substantially more effective than placebo in treating acute mania. Carbamazepine, haloperidol, ziprasidone, cariprazine, olanzapine, quetiapine, aripiprazole, lithium, paliperidone, asenapine, and divalproex were noticeably more effective than placebo.

CONCLUSIONS

Overall, tamoxifen appears to be the most effective of the currently known pharmaceutical therapy available to treat acute mania or manic episodes; however, this conclusion is restricted by the scale of RCTs conducted, and risperidone was found to be the most effective medication among antipsychotics. Carbamazepine, haloperidol, ziprasidone, cariprazine, olanzapine, quetiapine, aripiprazole, lithium, paliperidone, asenapine, and divalproex were noticeably effective in treating acute mania or manic episodes.

Effect of glucagon-like peptide-1 (GLP-1) analogues on epicardial adipose tissue: A meta-analysis

BACKGROUND AND AIMS

Glucagon-like peptide-1 (GLP-1) analogues reduce body fat and cardiovascular events in patients with type 2 diabetes. Accumulation of epicardial adipose tissue (EAT) is associated with increased cardio-metabolic risks and coronary events in type 2 diabetes.

METHODS

A systematic review and meta-analysis were performed from Glucagon-like peptide-1 analogues therapy on type 2 diabetes patients, reporting data from changes in EAT, after searching the PubMed/MEDLINE, Embase, Science Direct, Scopus, Google Scholar, and Cochrane databases.

RESULTS

It has been found a limited number of studies, a total of 4 studies (n = 160 patients with GLP-1 analogues therapy) were included in the final analysis. Pooled analysis revealed that GLP-1 analogues reduce EAT (MD: 1.83 mm [-2.50; -1.10]; P < 0.01). Compared with the patients before the treatment, the patients after the treatment had a smaller HbA1c (MD -1.10%[-1.80; -0.30]; p = 0.0143) and body mass index was reduced (MD -2.20 kg/m²[-3.70; -0.60]; p = 0.0058), GLP-1 therapy reduced low-density lipoprotein levels (MD-13.53 mg/dL [-21.74; -5.31]; p = 0.001) and reduced triglycerides levels significantly (MD -18.32 -28.20 mg/dL; -8.50); p = 0.0003).

CONCLUSIONS

This meta-analysis suggests that the amount of EAT is significantly reduced in T2D patients with Glucagon-like peptide-1 analogues.

Changes in glycaemic control of oral anti-diabetic medications assessed by continuous glucose monitors among patients with type 2 diabetes: a protocol of network meta-analysis

BACKGROUND

Continuous glucose monitors (CGMs) can measure interstitial fluid glucose levels to provide comprehensive real-time glucose profile among people with type 2 diabetes. These can accurately detect glucose levels, hyperglycaemia and hypoglycaemia events compared with conventional self-monitoring. Increased application of CGMs provides a valuable opportunity to evaluate glucose control on oral anti-diabetic medications. This review will compare the efficacy and safety of oral anti-diabetic medications among patients with type 2 diabetes, evaluated by CGM.

METHODS

The following databases will be searched: Cochrane Library, PubMed, EMBASE, CINAHL, PsycINFO, Scopus and grey literature (ClinicalTrials.gov, PsycEXTRA, ProQuest Dissertations, Google Scholar and Theses Global) for the identification of studies. The review will include and summarise evidence from randomised clinical trials that use CGMs for blood glucose management in adults (aged ≥ 18 years), published in English between January 2000 and May 2021 without any restrictions of countries. Reference list of all selected articles will independently be screened to identify additional studies left out in the initial search. Primary outcomes will be HbA1c (≤ 7.0%), time spent with hypoglycaemia (< 70 mg/dl) or hyperglycaemia (≥ 180 mg/dl). Secondary outcomes will be change in weight, blood pressure and related comorbidities (cardiovascular mortality, heart failure events, myocardial infarction and stroke). Study selection, data extraction and quality assessment will be conducted independently by at least two reviewers. A third reviewer will determine and resolve discrepancies. At least two independent reviewers will cross-check data synthesis. The quality of evidence of the review will be assessed according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) Tool.

DISCUSSION

The review is anticipated to provide up to date evidence for further studies and clinic practices regarding glycaemic control, hypoglycaemia, and hyperglycaemia issues. The results will be published in a peer-reviewed journal.

TRIAL REGISTRATION

PROSPERO CRD42020188399 .

A comprehensive evaluation of the safety of ipilimumab, nivolumab and their combination therapy: A systematic review and network meta-analysis

BACKGROUND

Immune checkpoint inhibitors (ICIs) have changed the landscape of management of advanced cancers. It is imperative to evaluate the safety of nivolumab and ipilimumab based therapies. This study was aimed to assess the comparative safety profiles of ipilimumab, nivolumab and their combinations.

MATERIALS AND METHODS

We searched PubMed, Embase, and the CENTRAL for randomised controlled trials of ipilimumab and nivolumab. The outcome measures were treatment-related adverse events [TRAEs], TRAEs of grade 3-5, treatment discontinuation due to TRAEs [TDTRAEs], TDTRAEs of grade 3-5, serious adverse events [SAEs] and SAEs of grades 3-5. We performed a network meta-analysis using the Bayesian approach in R version 4.0.3.

RESULTS

We identified 42 RCTs for final analysis. The treatment ranking for TRAEs revealed that nivolumab 240 mg/week and nivolumab 3 mg/kg/week were safer (0.84 and 0.81 in SUCRA); for TRAEs of grade 3-5, nivolumab 3 mg/kg/week and nivolumab 240 mg/week were safer (0.83 and 0.75 in SUCRA); for TDTRAEs nivolumab 3 mg/kg/week and ipilimumab in combination with other drugs were safer (0.87 and 0.64 in SUCRA) and for TDTRAEs of grade 3-5, nivolumab 3 mg/kg/week was safer (0.85 in SUCRA). Nivolumab 3 mg/kg/week and nivolumab 240 mg/week were safer (0.79 and 0.76 in SUCRA) for SAEs and nivolumab 3 mg/kg/week was safer for SAEs of grade 3-5 (0.78 in SUCRA).

CONCLUSION

Nivolumab 3 mg/kg biweekly, nivolumab 240 mg weekly and nivolumab 3 mg/kg plus ipilimumab 1 mg/kg triweekly could be preferred due to the relatively low risk of TRAEs, TDAEs and SAEs.

Maintenance Therapy for Preventing Endometrioma Recurrence after Endometriosis Resection Surgery-A Systematic Review and Network Meta-analysis

OBJECTIVE

To evaluate the efficacy of different hormone therapies in preventing postoperative endometrioma recurrence.

DATA SOURCES

The MEDLINE, COCHRANE and Embase electronic databases were searched from inception to 30 April 2021.

METHODS OF STUDY SELECTION

Randomized, controlled trials (RCTs) or cohort studies including reproductive age women with endometriosis undergoing ovarian cystectomy or excision of endometriotic lesions compared the effects of postoperative adjuvant therapy [gonadotropin-releasing hormone agonist (GnRHa)] and postoperative maintenance hormone interventions for more than one year [i.e., oral contraceptive pills (OCPs), dienogest (DNG), levonorgestrel-releasing intrauterine system (LNGIUS)] on endometrioma recurrence .

TABULATION, INTEGRATION AND RESULTS

Data collection and analysis of the data were independently performed by two reviewers. A total of 11 studies were included, of which 2 were RCTs, and 9 were cohort studies. There were 2394 patients with 6 interventions (cases: 1665, 69.6%) and expectant management (cases: 729, 30.4%). Relative treatment effects were estimated using network meta-analysis (NMA) and ranked in descending order. The clinical effectiveness of these drugs (versus expectant management) was as follows: GnRHa plus DNG [odds ratio (OR), 0.04; 95% confidence interval (CI), 0.01-0.27], surface under the cumulative ranking (SUCRA)= 94.0; DNG (OR, 0.11; 95% CI, 0.04-0.32), SUCRA= 69.7; GnRHa plus OCP (OR, 0.12; 95% CI, 0.02-0.64), SUCRA= 63.4; GnRHa plus LNGIUS (OR, 0.13; 95% CI, 0.03-0.66), SUCRA= 59.4; and OCP (OR, 0.21; 95% CI, 0.13-0.36), SUCRA= 43.6. The effectiveness of GnRHa (OR, 0.47; 95% CI, 0.12-1.89), SUCRA= 17.3 was not significantly different from that of controls.

CONCLUSION

In NMA, combined postoperative adjuvant therapy and longer maintenance hormone treatment are better than a single agent in preventing postoperative endometrioma recurrence. GnRHa plus DNG maintenance treatment might be the most effective intervention. Large-scale RCTs of these agents are still required.

REGISTRATION OF SYSTEMATIC REVIEWS

Date of PROSPERO registration: 17 April 2021.

PROSPERO REGISTRATION NUMBER

CRD42021247291.

Some types of exercise are more effective than others in people with chronic low back pain: a network meta-analysis

QUESTION

What are the effects of specific types of exercise treatments on pain intensity and functional limitation outcomes for adults with chronic low back pain?

DESIGN

Systematic review with network meta-analysis of randomised controlled trials.

PARTICIPANTS

Adults with non-specific low back pain for ≥ 12 weeks.

INTERVENTION

Exercise treatments prescribed or planned by a health professional that involved conducting specific activities, postures and/or movements with a goal to improve low back pain outcomes.

OUTCOME MEASURES

Pain intensity (eg, visual analogue scale or numerical rating scale) and back-related functional limitations (eg, Roland Morris Disability Questionnaire or Oswestry Disability Index), each standardised to range from 0 to 100.

RESULTS

This review included 217 randomised controlled trials with 20,969 participants and 507 treatment groups. Most exercise types were more effective than minimal treatment for pain and functional limitation outcomes. Network meta-analysis results were compatible with moderate to clinically important treatment effects for Pilates, McKenzie therapy, and functional restoration (pain only) and flexibility exercises (function only) compared with minimal treatment, other effective treatments and other exercise types. The estimated mean differences for these exercise types compared with minimal treatment ranged from -15 to -19 for pain and from -10 to -12 for functional limitation.

CONCLUSION

This review found evidence that Pilates, McKenzie therapy and functional restoration were more effective than other types of exercise treatment for reducing pain intensity and functional limitations. Nevertheless, people with chronic low back pain should be encouraged to perform the exercise that they enjoy to promote adherence.

REGISTRATION

DOI:10.1002/14651858.CD009790.

Placental location site and adverse antepartum pregnancy complications: a meta-analysis and review of the literature

PURPOSE

The purpose to the study was to determine the relationship, if any, between the placental location site and antepartum complications of pregnancy.

METHODS

A University research librarian conducted a comprehensive literature search using the search engines PubMed and Web of Science. The search terms were "placental location" AND "pregnancy complications" OR "perinatal complications. There were no limits put on the years of the search.

RESULTS

The search identified 110 articles. After reviewing all the abstracts, relevant full articles, and references of full articles, there were 22 articles identified specific to antepartum complications. Central + fundal locations compared to all lateral were associated with a lower risk of hypertension during pregnancy RR = 0.47, 95% CI: 0.31-0.71]. Central location compared to all lateral was also associated with lower risk of hypertension during pregnancy [RR = 0.39, 95% CI: 0.26-0.59]. Placenta locations in the lower uterine segment were associated with greater risk of antepartum hemorrhage (APH) [RR = 2.99, 95% CI: 1.16-7.75] compared to above the lower uterine segment. No differences were observed in placental locations and gestational diabetes (GDM), preterm prelabor rupture of membranes (PPROM), preterm delivery (PTD) or on a placental abruption.

CONCLUSION

Central and fundal location sites and central location alone decreased the risk of hypertension during pregnancy. Low uterine segment location sites increased the risk for APH. There were no effects of placenta location sites on the development of GDM, PPROM, PTD or abruption.

Concepts and emerging issues of network meta-analysis

Most diseases have more than two interventions or treatment methods, and the application of network meta-analysis (NMA) studies to compare and evaluate the superiority of each intervention or treatment method is increasing. Understanding the concepts and processes of systematic reviews and meta-analyses is essential to understanding NMA. As with systematic reviews and meta-analyses, NMA involves specifying the topic, searching for and selecting all related studies, and extracting data from the selected studies. To evaluate the effects of each treatment, NMA compares and analyzes three or more interventions or treatment methods using both direct and indirect evidence. There is a possibility of several biases when performing NMA. Therefore, key assumptions like similarity, transitivity, and consistency should be satisfied when performing NMA. Among these key assumptions, consistency can be evaluated and quantified by statistical tests. This review aims to introduce the concepts of NMA, analysis methods, and interpretation and presentation of the results of NMA. It also briefly introduces the emerging issues in NMA, including methods for evaluation of consistency.

Effect of Bempedoic Acid on atherogenic lipids and inflammation: A meta-analysis

BACKGROUND

Bempedoic acid is a novel non-statin drug that was developed to treat hyperlipidemia in combination with other lipid-lowering drugs in those patients who need additional lipid lowering.

OBJECTIVES

(1) To investigate the lipid efficacy of bempedoic acid; (2) to analyze the anti-inflammatory effects of bempedoic acid estimated through high sensitivity C-reactive protein (hsCRP).

METHODS

We performed a meta-analysis including randomized trials of bempedoic acid therapy, reporting low-density lipoprotein cholesterol (LDL-C), non-high-density lipoprotein cholesterol (non-HDL-C), apolipoprotein B and hsCRP with a minimum of 4 weeks of follow-up. The primary endpoint was defined as the percentage change in lipids and hsCRP levels measured from baseline to follow-up, comparing groups of subjects on bempedoic acid versus placebo.

RESULTS

Seven eligible trials of bempedoic acid (3892 patients) were included. The bempedoic acid therapy was associated with a significant reduction in LDL-C levels [-20.3% (CI 95% -23.5 to -17.1)]; I²=43%]. Similarly, a significant percentage reduction in the apolipoprotein B levels [-14.3% (CI 95% -16.4 to -12.1)]; p<0.05; I²=46%], non-HDL-C levels [-15.5% (CI 95% -18.1 to -13.0)]; p<0.05; I²=53%] and hsCRP [-23.4% (CI 95% -32.6 to -14.2)]; p<0.05; I²=69%] was demonstrated with the bempedoic acid use. The sensitivity analysis showed that the results were robust.

CONCLUSION

Our data suggests that the use of bempedoic acid significantly reduces the levels of all atherogenic lipid markers, including LDL-C, non-HDL-C and apolipoprotein B. Furthermore, considering hsCRP levels, the drug produces an anti-inflammatory effect.

Comparative efficacy of pharmacological treatments on measures of self-rated functional outcomes using the Sheehan Disability Scale in patients with major depressive disorder: a systematic review and network meta-analysis

OBJECTIVE

More than 50% patients with major depressive disorder (MDD) have severe functional impairment. The restoration of patient functioning is a critical therapeutic goal among patients with MDD. We conducted a systematic review and network meta-analysis to evaluate the efficacy of pharmacological treatments on self-rated functional outcomes using the Sheehan Disability Scale in adults with MDD in randomized clinical trials.

METHODS

PubMed, EMBASE, PsycINFO, Cochrane Library, and ClinicalTrials.gov were searched from inception to December 10, 2019. Summary statistics are reported as weighted mean differences with 95% confidence intervals. Interventions were ranked using the surface under the cumulative ranking probabilities.

RESULTS

We included 42 randomized controlled trials (RCTs) (n = 18 998) evaluating the efficacy of 13 different pharmacological treatments on functional outcomes, as measured by the Sheehan Disability Scale (SDS). Duloxetine was the most effective pharmacological agent on functional outcomes, followed by (ranked by efficacy): paroxetine, levomilnacipran, venlafaxine, quetiapine, desvenlafaxine, agomelatine, escitalopram, amitriptyline, bupropion, sertraline, vortioxetine, and fluoxetine. Serotonin and norepinephrine reuptake inhibitors were more effective than other drug classes. Additionally, the comparison-adjusted funnel plot suggested the publication bias between small and large studies was relatively low.

CONCLUSIONS

Our results indicate that there may be differences across antidepressant agents and classes with respect to self-reported functional outcomes. Validation and replication of these findings in large-scale RCTs are warranted. Our research results will be clinically useful for guiding psychiatrists in treating patients with MDD and functional impairment. PROSPERO registration number CRD42018116663.

Utility of intra-operative flexible sigmoidoscopy to assess colorectal anastomosis: a systematic review and meta-analysis

BACKGROUND

Anastomotic leak (AL) after colorectal resection leads to increased oncological and non-oncological, morbidity and mortality. Intra-operative assessment of a colorectal anastomosis with intra-operative flexible sigmoidoscopy (IOFS) has become increasingly prevalent and is an alternative to conventional air leak test. It is thought that intra-operative identification of an AL or anastomotic bleeding (AB) allows for immediate reparative intervention at the time of anastomosis formation itself. We aim to assess the available evidence for the use of IOFS to prevent complications following colorectal resection.

METHODS

Following Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines, a systematic review of the literature between January 1980 and June 2020 was performed. Comparative studies assessing IOFS versus conventional air leak test were compared, and outcomes were pooled.

RESULTS

A total of 4512 articles were assessed, of which eight were found to meet the inclusion criteria. A total of 1792 patients were compared; 884 in the IOFS arm and 908 in the control arm. IOFS was associated with an increase in the rate of positive leak test (odds ratio (OR) 5.21, P > 0.001), a decrease in AL (OR 0.45, P = 0.006) and a decrease in post-operative AB requiring intervention (OR 0.40, P = 0.037).

CONCLUSION

In a non-randomized meta-analysis, IOFS increases the likelihood of identifying an anastomotic defect or bleeding intra-operatively. This allows for immediate intervention that decreases the rate of AL and AB. This adds impetus for performing routine IOFS after a left-sided colorectal resection with anastomosis and highlights the need for randomized controlled trial to confirm the finding.

Role of non-statin lipid-lowering therapy in coronary atherosclerosis regression: a meta-analysis and meta-regression

BACKGROUND

Several studies have investigated the association between non-statin lipid-lowering therapy and regression of atherosclerosis. However, these studies were mostly small and their results were not always robust. The objectives were: (1) to define if a dual lipid-lowering therapy (statin + non-statin drugs) is associated with coronary atherosclerosis regression, estimated by intravascular ultrasound (IVUS); (2) to assess the association between dual lipid-lowering-induced changes in low density lipoprotein cholesterol (LDL-C) and non-high-density-lipoprotein cholesterol (non-HDL-C) levels and atherosclerosis regression.

METHODS

A meta-analysis including trials of non-statin lipid-lowering therapy, reporting LDL-C, non-HDL-C and total atheroma volume (TAV) with a minimum of 6 months of follow-up was performed. The primary endpoint was defined as the change in TAV measured from baseline to follow-up, comparing groups of subjects on statins alone versus combination of statin and non-statin drugs. The random-effects model and meta-regression were performed.

RESULTS

Eight eligible trials of non-statin lipid-lowering drugs (1759 patients) were included. Overall, the dual lipid-lowering therapy was associated with a significant reduction in TAV [- 4.0 mm³ (CI 95% -5.4 to - 2.6)]; I² = 0%]. The findings were similar in the stratified analysis according to the lipid-lowering drug class (ezetimibe or PCSK9 inhibitors). In the meta-regression, a 10% decrease in LDL-C or non-HDL-C levels, was associated, respectively, with 1.0 mm³ and 1.1 mm³ regressions in TAV.

CONCLUSION

These data suggests the addition of ezetimibe or PCSK9 inhibitors to statin therapy results in a significant regression of TAV. Reduction of coronary atherosclerosis observed with non-statin lipid-lowering therapy is associated to the degree of LDL-C and non-HDL-C lowering. Therefore, it seems reasonable to achieve lipid goals according to cardiovascular risk and regardless of the lipid-lowering strategy used (statin monotherapy or dual treatment).

A simulation study to compare different estimation approaches for network meta-analysis and corresponding methods to evaluate the consistency assumption

Background

Network meta-analysis (NMA) is becoming increasingly popular in systematic reviews and health technology assessments. However, there is still ambiguity concerning the properties of the estimation approaches as well as for the methods to evaluate the consistency assumption.

Methods

We conducted a simulation study for networks with up to 5 interventions. We investigated the properties of different methods and give recommendations for practical application. We evaluated the performance of 3 different models for complex networks as well as corresponding global methods to evaluate the consistency assumption. The models are the frequentist graph-theoretical approach netmeta, the Bayesian mixed treatment comparisons (MTC) consistency model, and the MTC consistency model with stepwise removal of studies contributing to inconsistency identified in a leverage plot.

Results

We found that with a high degree of inconsistency none of the evaluated effect estimators produced reliable results, whereas with moderate or no inconsistency the estimator from the MTC consistency model and the netmeta estimator showed acceptable properties. We also saw a dependency on the amount of heterogeneity. Concerning the evaluated methods to evaluate the consistency assumption, none was shown to be suitable.

Conclusions

Based on our results we recommend a pragmatic approach for practical application in NMA. The estimator from the netmeta approach or the estimator from the Bayesian MTC consistency model should be preferred. Since none of the methods to evaluate the consistency assumption showed satisfactory results, users should have a strong focus on the similarity as well as the homogeneity assumption.

Comparison of Cardiovascular Events Among Users of Different Classes of Antihypertension Medications: A Systematic Review and Network Meta-analysis

Importance

Antihypertension medications have been associated with prevention of cardiovascular events, although less is known about the comparative effectiveness of different medication classes.

Objective

To compare contemporary aggregated first-in-trial cardiovascular events among patients with hypertension and no substantial comorbidities.

Data Sources

The PubMed, Embase, and Cochrane Library databases were systematically searched for articles published between January 1, 1990, and October 24, 2017.

Study Selection

Randomized clinical trials that tested commonly used antihypertension medications (angiotensin-converting enzyme inhibitors, dihydropyridine calcium channel blockers, nondihydropyridine calcium channel blockers, β-blockers, angiotensin receptor blockers, and diuretics) and that reported selected cardiovascular outcomes for at least 6 months of follow-up.

Data Extraction and Synthesis

The analysis was conducted from October 2017 to December 2019. Two reviewers extracted the number of cardiovascular events at the end of treatment for all study groups. For each outcome, a frequentist network meta-analysis was used to compare risk reductions between medication classes (random-effects models weighted by the inverse variance). The dose-response association between a 10-mm Hg reduction of systolic blood pressure and a 5-mm Hg reduction of diastolic blood pressure and the risk of first-in-trial cardiovascular events was estimated.

Main Outcomes and Measures

First-in-trial cardiovascular events, including cardiovascular death, myocardial infarction, stroke, and revascularization.

Results

In this systematic review and network meta-analysis, data were pooled from 46 eligible clinical trials (248 887 total participants with a mean [SD] age of 65.6 [5.8] years; 52.8% men). In the network meta-analysis, compared with placebo, angiotensin-converting enzyme inhibitors, dihydropyridine calcium channel blockers, and thiazide diuretics were reported to be similarly effective in reducing overall cardiovascular events (25%), cardiovascular death (20%), and stroke (35%); angiotensin-converting enzyme inhibitors were reported to be the most effective in reducing the risk of myocardial infarction (28%); and diuretics were reported to be the most effective in reducing revascularization (33%). In the metaregression analyses, each 10-mm Hg reduction in systolic blood pressure and 5-mm Hg reduction in diastolic blood pressure was significantly associated with a lower risk of cardiovascular death, stroke, and overall cardiovascular events.

Conclusions and Relevance

In this network meta-analysis of clinical trials of patients with hypertension and no substantial comorbidities, different classes of antihypertension medications were associated with similar benefits in reducing cardiovascular events. Future studies should compare the effectiveness of combinations of antihypertension medications in reducing cardiovascular events.

Identifying inconsistency in network meta-analysis: Is the net heat plot a reliable method?

One of the biggest challenges for network meta‐analysis is inconsistency, which occurs when the direct and indirect evidence conflict. Inconsistency causes problems for the estimation and interpretation of treatment effects and treatment contrasts. Krahn and colleagues proposed the net heat approach as a graphical tool for identifying and locating inconsistency within a network of randomized controlled trials. For networks with a treatment loop, the net heat plot displays statistics calculated by temporarily removing each design one at a time, in turn, and assessing the contribution of each remaining design to the inconsistency. The net heat plot takes the form of a matrix which is displayed graphically with coloring indicating the degree of inconsistency in the network. Applied to a network of individual participant data assessing overall survival in 7531 patients with lung cancer, we were surprised to find no evidence of important inconsistency from the net heat approach; this contradicted other approaches for assessing inconsistency such as the Bucher approach, Cochran's Q statistic, node‐splitting, and the inconsistency parameter approach, which all suggested evidence of inconsistency within the network at the 5% level. Further theoretical work shows that the calculations underlying the net heat plot constitute an arbitrary weighting of the direct and indirect evidence which may be misleading. We illustrate this further using a simulation study and a network meta‐analysis of 10 treatments for diabetes. We conclude that the net heat plot does not reliably signal inconsistency or identify designs that cause inconsistency.

Autologous Transplantation for Newly Diagnosed Multiple Myeloma in the Era of Novel Agent Induction: A Systematic Review and Meta-analysis

Importance

The role of high-dose therapy with melphalan followed by autologous stem cell transplant (HDT/ASCT) in patients with multiple myeloma continues to be debated in the context of novel agent induction.

Objective

To perform a systematic review, conventional meta-analysis, and network meta-analysis of all phase 3 randomized clinical trials (RCTs) evaluating the role of HDT/ASCT.

Data Sources

We performed a systematic literature search of Cochrane Central, MEDLINE, and Scopus from January 2000 through April 2017 and relevant annual meeting abstracts from January 2014 to December 2016. The following search terms were used: "myeloma" combined with "autologous," "transplant," "myeloablative," or "stem cell."

Study Selection

Phase 3 RCTs comparing HDT/ASCT with standard-dose therapy (SDT) using novel agents were assessed. Studies comparing single HDT/ASCT with bortezomib, lenalidomide, and dexamethasone consolidation and tandem transplantation were included for network meta-analysis.

Data Extraction And Synthesis

For the random effects meta-analysis, we used hazard ratios (HRs) and corresponding 95% CIs.

Main Outcomes and Measures

The primary outcome was progression-free survival (PFS). Overall survival (OS), complete response, and treatment-related mortality were secondary outcomes.

Results

A total of 4 RCTs (2421 patients) for conventional meta-analysis and 5 RCTs (3171 patients) for network meta-analysis were selected. The combined odds for complete response were 1.27 (95% CI, 0.97-1.65; P = .07) with HDT/ASCT when compared with SDT. The combined HR for PFS was 0.55 (95% CI, 0.41-0.74; P < .001) and 0.76 for OS (95% CI, 0.42-1.36; P = .20) in favor of HDT. Meta-regression showed that longer follow-up was associated with superior PFS (HR/mo, 0.98; 95% CI, 0.96-0.99; P = .03) and OS (HR/mo, 0.90; 95% CI, 0.84-0.96; P = .002). For PFS, tandem HDT/ASCT had the most favorable HR (0.49; 95% CI, 0.37-0.65) followed by single HDT/ASCT with bortezomib, lenalidomide, and dexamethasone (HR, 0.53; 95% CI, 0.37-0.76) and single HDT/ASCT alone (HR, 0.68; 95% CI, 0.53-0.87) compared with SDT. For OS, none of the HDT/ASCT-based approaches had a significant effect on survival. Treatment-related mortality with HDT/ASCT was minimal (<1%).

Conclusions and Relevance

The results of the conventional meta-analysis and network meta-analysis of all the phase 3 RCTs showed that HDT/ASCT was associated with superior PFS with minimal toxic effects compared with SDT. Both tandem HDT/ASCT and single HDT/ASCT with bortezomib, lenalidomide, and dexamethasone were superior to single HDT/ASCT alone and SDT for PFS, but OS was similar across the 4 approaches. Longer follow-up may better delineate any OS benefit; however, is likely to be affected by effective postrelapse therapy.

Oncological safety of laparoscopic versus open colorectal cancer surgery in obesity: a systematic review and meta-analysis

BACKGROUND

Colorectal cancer resection in the obese (OB) patients can be technically challenging. With the increasing adoption of laparoscopic surgery, the benefits remain uncertain. Hence, the aim of this study is to assess the short- and long-term outcomes of laparoscopic compared to open colorectal cancer resection in the OB patients.

METHODS

A systematic review and meta-analysis was performed according to the PRISMA guidelines. The outcome measures were 5-year disease-free survival, overall survival, circumferential resection margin and local and distant recurrence.

RESULTS

A total of 20 studies were included, with a total number of 6779 participants, of whom 1785 (26.3%) were OB and 4994 (73.7%) were non-obese (NOB) participants. The OB patients had higher R1 resection (OB 6.9% versus NOB 3.1%; P = 0.011) and lower mean number of lymph nodes harvested, with standard mean difference of -0.29; P = 0.023, favouring the NOB patients. However, there was no statistical difference for local (OB 2.8% versus NOB 3.4%) or distant recurrence (OB 12.9% versus NOB 15.2%) rate between the two cohorts. There was no difference in 5-year disease-free survival (OB 81% versus NOB 77.4%; odds ratio 1.25, P = 0.215) and overall survival (OB 89.4% versus NOB 87.9%; odds ratio 1.16, P = 0.572). Lastly, the OB group had higher mean total blood loss, total operative time and length of hospital stay when compared to NOB patients.

CONCLUSION

From a pooled non-randomized study, laparoscopic colorectal cancer resection is safe in OB patients with equivalent long-term outcomes compared to NOB patients. However, there is a higher morbidity rate with an increased demand on hospital resources for the OB cohort.

Prognostic Impact of Tumor-Infiltrating Lymphocytes in Primary and Metastatic Colorectal Cancer: A Systematic Review and Meta-analysis

BACKGROUND

There is increasing literature emerging on the significance of tumor-infiltrating lymphocytes in colorectal cancer. However, there have been inconsistent findings, secondary to small patient numbers and varied methods for identifying these lymphocytes.

OBJECTIVE

The aim of this study was to determine the prognostic and predictive power of tumor-infiltrating lymphocytes in colon, rectal (in neoadjuvant setting), and metastatic colorectal cancer.

DATA SOURCES

A comprehensive search of PubMed and Embase was undertaken from January 2006 to December 2016.

STUDY SELECTION

The inclusion criteria included a description of the tumor-infiltrating lymphocyte subset(s) assessed with reporting of associated short- and long-term outcomes.

MAIN OUTCOME MEASURES

The main outcome measures, were disease-free and overall survival.

RESULTS

A total of 25 studies were included, 15 for primary colorectal cancer (4719 patients), 7 for locally advanced rectal cancer (727 patients), and 3 studies for metastatic colorectal cancer (418 patients). High CD3, CD8, FoxP3, and CD45RO densities were associated with improved overall survival for primary colorectal cancer, with pooled estimated HRs of 0.88, 0.81, 0.70, and 0.63 (all p < 0.001) respectively. Furthermore, in locally advanced rectal cancer, the levels of CD8 cells were a significant predictor of good tumor regression grade after chemoradiotherapy.

LIMITATIONS

The retrospective nature of included studies and the significant interstudy heterogeneity were limitations.

CONCLUSIONS

There is increasing evidence that tumor-infiltrating lymphocytes play an important role in predicting prognosis in colorectal cancer and tumor regression after neoadjuvant chemoradiotherapy in locally advanced rectal cancer. Clinical researchers are now in a unique position to build on this work to identify robust predictive markers to stratify patients not only to currently available therapies but also to immunotherapy, which has demonstrated success in improving patient outcomes.

Two new approaches for the visualisation of models for network meta-analysis

Background

Meta-analysis is a useful tool for combining evidence from multiple studies to estimate a pooled treatment effect. An extension of meta-analysis, network meta-analysis, is becoming more commonly used as a way to simultaneously compare multiple treatments in a single analysis. Despite the variety of approaches available for presenting fitted models, ascertaining an intuitive understanding of these models is often difficult. This is especially challenging in large networks with many different treatments. Here we propose two visualisation methods, so that network meta-analysis models can be more easily interpreted.

Methods

Our methods can be used irrespective of the statistical model or the estimation method used and are grounded in network analysis. We define three types of distance measures between the treatments that contribute to the network. These three distance measures are based on 1) the estimated treatment effects, 2) their standard errors and 3) the corresponding p-values. Then, by using a suitable threshold, we categorise some treatment pairs as being “close” (short distances). Treatments that are close are regarded as “connected” in the network analysis theory. Finally, we group the treatments into communities using standard methods for network analysis. We are then able to identify which parts of the network are estimated to have similar (or different) treatment efficacy and which parts of the network are better identified. We also propose a second method using parametric bootstrapping, where a heat map is used in the visualisation. We use the software R and provide the code used.

Results

We illustrate our new methods using a challenging dataset containing 22 treatments, and a previously fitted model for this data. Two communities of treatments that appear to have similar efficacy are identified. Furthermore using our methods we can identify parts of the network that are better (and less well) identified.

Conclusions

Our new visualisation approaches may be used by network meta-analysts to gain an intuitive understanding of the implications of their fitted models. Our visualisation methods may be used informally, to identify the most salient features of the fitted models that can then be reported, or more formally by presenting the new visualisation devices within published reports.

Electronic supplementary material

The online version of this article (10.1186/s12874-019-0689-9) contains supplementary material, which is available to authorized users.

Safety of laparoscopic appendicectomy during pregnancy: a systematic review and meta-analysis

BACKGROUND

Acute appendicitis is the most common non-obstetric surgical presentation during pregnancy. There were concerns that laparoscopic appendicectomy increases the risk of foetal loss compared to an open approach. Therefore, with recent advances in perioperative care, it is likely the risk has changed. Here, we performed an updated meta-analysis assessing the safety of laparoscopic appendicectomy in pregnant women.

METHODS

A meta-analysis was performed according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A search was undertaken between 2000 and 2017 on Ovid Medline and Embase. The primary outcome measures were foetal loss and preterm delivery, whereas secondary outcome measures were operative time and hospital length of stay. A random-effect model was performed to pool odds ratio (OR) and standardized mean difference (SMD).

RESULTS

Seventeen observational studies were included, with 1886 patients in the laparoscopic and 4261 patients in the open group. Comparing laparoscopic versus open appendicectomy, there were 54 (5.96%) and 136 (3.73%) foetal losses, respectively. However, preterm delivery was much higher in the open approach (8.99%) compared to laparoscopic approach (2.84%). Pooled OR for foetal loss was 1.84 (95% confidence interval (CI) 1.31-2.58, P < 0.001), whereas OR for preterm delivery was 0.39 (95% CI 0.27-0.55, P < 0.001). There was no significant difference between both approaches for operative time (SMD -0.07; 95% CI -0.43 to 0.30, P = 0.71) or hospital length of stay (SMD -0.34; 95% CI -0.83 to 0.16, P = 0.18).

CONCLUSION

In a pooled analysis of level III evidence, laparoscopic appendicectomy posed a higher risk of foetal loss but lower risk of preterm delivery. Caution and informed consent are crucial when offering a laparoscopic approach.

Hypothesis testing in Bayesian network meta-analysis

Background

Network meta-analysis is an extension of the classical pairwise meta-analysis and allows to compare multiple interventions based on both head-to-head comparisons within trials and indirect comparisons across trials. Bayesian or frequentist models are applied to obtain effect estimates with credible or confidence intervals. Furthermore, p-values or similar measures may be helpful for the comparison of the included arms but related methods are not yet addressed in the literature. In this article, we discuss how hypothesis testing can be done in a Bayesian network meta-analysis.

Methods

An index is presented and discussed in a Bayesian modeling framework. Simulation studies were performed to evaluate the characteristics of this index. The approach is illustrated by a real data example.

Results

The simulation studies revealed that the type I error rate is controlled. The approach can be applied in a superiority as well as in a non-inferiority setting.

Conclusions

Test decisions can be based on the proposed index. The index may be a valuable complement to the commonly reported results of network meta-analyses. The method is easy to apply and of no (noticeable) additional computational cost.

Assessing the consistency assumptions underlying network meta-regression using aggregate data

When numerous treatments exist for a disease (Treatments 1, 2, 3, etc), network meta‐regression (NMR) examines whether each relative treatment effect (eg, mean difference for 2 vs 1, 3 vs 1, and 3 vs 2) differs according to a covariate (eg, disease severity). Two consistency assumptions underlie NMR: consistency of the treatment effects at the covariate value 0 and consistency of the regression coefficients for the treatment by covariate interaction. The NMR results may be unreliable when the assumptions do not hold. Furthermore, interactions may exist but are not found because inconsistency of the coefficients is masking them, for example, when the treatment effect increases as the covariate increases using direct evidence but the effect decreases with the increasing covariate using indirect evidence.

We outline existing NMR models that incorporate different types of treatment by covariate interaction. We then introduce models that can be used to assess the consistency assumptions underlying NMR for aggregate data. We extend existing node‐splitting models, the unrelated mean effects inconsistency model, and the design by treatment inconsistency model to incorporate covariate interactions. We propose models for assessing both consistency assumptions simultaneously and models for assessing each of the assumptions in turn to gain a more thorough understanding of consistency.

We apply the methods in a Bayesian framework to trial‐level data comparing antimalarial treatments using the covariate average age and to four fabricated data sets to demonstrate key scenarios.

We discuss the pros and cons of the methods and important considerations when applying models to aggregated data.

Prognostic value of tumour regression grade in locally advanced rectal cancer: a systematic review and meta-analysis

AIM

The current standard of care for locally advanced rectal cancer involves neoadjuvant chemoradiotherapy (CRT) followed by total mesorectal excision. There is a spectrum of response to neoadjuvant therapy; however, the prognostic value of tumour regression grade (TRG) in predicting disease-free survival (DFS) or overall survival (OS) is inconsistent in the literature.

METHOD

This study was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A systematic search was undertaken using Ovid MEDLINE, Embase and Google Scholar. Inclusion criteria were Stage II and III locally advanced rectal cancer treated with long-course CRT followed by radical surgery. The aim of the meta-analysis was to assess the prognostic implication of each TRG for rectal cancer following neoadjuvant CRT. Long-term prognosis was assessed. The main outcome measures were DFS and OS. A random effects model was performed to pool the hazard ratio (HR) from all included studies.

RESULTS

There were 4875 patients from 17 studies, with 775 (15.9%) attaining a pathological complete response (pCR) and 719 (29.9%) with no response. A significant association with OS was identified from a pooled-estimated HR for pCR (HR = 0.47, P = 0.002) and nonresponding tumours (HR = 2.97; P < 0.001). Previously known tumour characteristics, such as ypN, lymphovascular invasion and perineural invasion, were also significantly associated with DFS and OS, with estimated pooled HRs of 2.2, 1.4 and 2.3, respectively.

CONCLUSION

In conclusion, the degree of TRG was of prognostic value in predicting long-term outcomes. The current challenge is the development of a high-validity tests to predict pCR.

A framework for identifying treatment-covariate interactions in individual participant data network meta-analysis

Background: Stratified medicine seeks to identify patients most likely to respond to treatment. Individual participant data (IPD) network meta‐analysis (NMA) models have greater power than individual trials to identify treatment‐covariate interactions (TCIs). Treatment‐covariate interactions contain “within” and “across” trial interactions, where the across‐trial interaction is more susceptible to confounding and ecological bias.

Methods: We considered a network of IPD from 37 trials (5922 patients) for cervical cancer (2394 events), where previous research identified disease stage as a potential interaction covariate. We compare 2 models for NMA with TCIs: (1) 2 effects separating within‐ and across‐trial interactions and (2) a single effect combining within‐ and across‐trial interactions. We argue for a visual assessment of consistency of within‐ and across‐trial interactions and consider more detailed aspects of interaction modelling, eg, common vs trial‐specific effects of the covariate. This leads us to propose a practical framework for IPD NMA with TCIs.

Results: Following our framework, we found no evidence in the cervical cancer network for a treatment‐stage interaction on the basis of the within‐trial interaction. The NMA provided additional power for an across‐trial interaction over and above the pairwise evidence. Following our proposed framework, we found that the within‐ and across‐trial interactions should not be combined.

Conclusion: Across‐trial interactions are susceptible to confounding and ecological bias. It is important to separate the sources of evidence to check their consistency and identify which sources of evidence are driving the conclusion. Our framework provides practical guidance for researchers, reducing the risk of unduly optimistic interpretation of TCIs.

Estimating the variance for heterogeneity in arm-based network meta-analysis

Network meta-analysis can be implemented by using arm-based or contrast-based models. Here we focus on arm-based models and fit them using generalized linear mixed model procedures. Full maximum likelihood (ML) estimation leads to biased trial-by-treatment interaction variance estimates for heterogeneity. Thus, our objective is to investigate alternative approaches to variance estimation that reduce bias compared with full ML. Specifically, we use penalized quasi-likelihood/pseudo-likelihood and hierarchical (h) likelihood approaches. In addition, we consider a novel model modification that yields estimators akin to the residual maximum likelihood estimator for linear mixed models. The proposed methods are compared by simulation, and 2 real datasets are used for illustration. Simulations show that penalized quasi-likelihood/pseudo-likelihood and h-likelihood reduce bias and yield satisfactory coverage rates. Sum-to-zero restriction and baseline contrasts for random trial-by-treatment interaction effects, as well as a residual ML-like adjustment, also reduce bias compared with an unconstrained model when ML is used, but coverage rates are not quite as good. Penalized quasi-likelihood/pseudo-likelihood and h-likelihood are therefore recommended.

A design-by-treatment interaction model for network meta-analysis and meta-regression with integrated nested Laplace approximations

Network meta-analysis (NMA) is gaining popularity for comparing multiple treatments in a single analysis. Generalized linear mixed models provide a unifying framework for NMA, allow us to analyze datasets with dichotomous, continuous or count endpoints, and take into account multiarm trials, potential heterogeneity between trials and network inconsistency. To perform inference within such NMA models, the use of Bayesian methods is often advocated. The standard inference tool is Markov chain Monte Carlo (MCMC), which is computationally expensive and requires convergence diagnostics. A deterministic approach to do fully Bayesian inference for latent Gaussian models can be achieved by integrated nested Laplace approximations (INLA), which is a fast and accurate alternative to MCMC. We show how NMA models fit in the class of latent Gaussian models and how NMA models are implemented using INLA and demonstrate that the estimates obtained by INLA are in close agreement with the ones obtained by MCMC. Specifically, we emphasize the design-by-treatment interaction model with random inconsistency parameters (also known as the Jackson model). Also, we have proposed a network meta-regression model, which is constructed by incorporating trial-level covariates to the Jackson model to explain possible sources of heterogeneity and/or inconsistency in the network. A publicly available R package, nmaINLA, is developed to automate the INLA implementation of NMA models, which are considered in this paper. Three applications illustrate the use of INLA for a NMA.

Paule-Mandel estimators for network meta-analysis with random inconsistency effects

Network meta-analysis is used to simultaneously compare multiple treatments in a single analysis. However, network meta-analyses may exhibit inconsistency, where direct and different forms of indirect evidence are not in agreement with each other, even after allowing for between-study heterogeneity. Models for network meta-analysis with random inconsistency effects have the dual aim of allowing for inconsistencies and estimating average treatment effects across the whole network. To date, two classical estimation methods for fitting this type of model have been developed: a method of moments that extends DerSimonian and Laird's univariate method and maximum likelihood estimation. However, the Paule and Mandel estimator is another recommended classical estimation method for univariate meta-analysis. In this paper, we extend the Paule and Mandel method so that it can be used to fit models for network meta-analysis with random inconsistency effects. We apply all three estimation methods to a variety of examples that have been used previously and we also examine a challenging new dataset that is highly heterogenous. We perform a simulation study based on this new example. We find that the proposed Paule and Mandel method performs satisfactorily and generally better than the previously proposed method of moments because it provides more accurate inferences. Furthermore, the Paule and Mandel method possesses some advantages over likelihood-based methods because it is both semiparametric and requires no convergence diagnostics. Although restricted maximum likelihood estimation remains the gold standard, the proposed methodology is a fully viable alternative to this and other estimation methods.

Multivariate and network meta-analysis of multiple outcomes and multiple treatments: rationale, concepts, and examples

Organisations such as the National Institute for Health and Care Excellence require the synthesis of evidence from existing studies to inform their decisions—for example, about the best available treatments with respect to multiple efficacy and safety outcomes. However, relevant studies may not provide direct evidence about all the treatments or outcomes of interest. Multivariate and network meta-analysis methods provide a framework to address this, using correlated or indirect evidence from such studies alongside any direct evidence. In this article, the authors describe the key concepts and assumptions of these methods, outline how correlated and indirect evidence arises, and illustrate the contribution of such evidence in real clinical examples involving multiple outcomes and multiple treatments

A matrix-based method of moments for fitting multivariate network meta-analysis models with multiple outcomes and random inconsistency effects

Random-effects meta-analyses are very commonly used in medical statistics. Recent methodological developments include multivariate (multiple outcomes) and network (multiple treatments) meta-analysis. Here, we provide a new model and corresponding estimation procedure for multivariate network meta-analysis, so that multiple outcomes and treatments can be included in a single analysis. Our new multivariate model is a direct extension of a univariate model for network meta-analysis that has recently been proposed. We allow two types of unknown variance parameters in our model, which represent between-study heterogeneity and inconsistency. Inconsistency arises when different forms of direct and indirect evidence are not in agreement, even having taken between-study heterogeneity into account. However, the consistency assumption is often assumed in practice and so we also explain how to fit a reduced model which makes this assumption. Our estimation method extends several other commonly used methods for meta-analysis, including the method proposed by DerSimonian and Laird (). We investigate the use of our proposed methods in the context of both a simulation study and a real example.

Network meta-analysis: a technique to gather evidence from direct and indirect comparisons

Systematic reviews and pairwise meta-analyses of randomized controlled trials, at the intersection of clinical medicine, epidemiology and statistics, are positioned at the top of evidence-based practice hierarchy. These are important tools to base drugs approval, clinical protocols and guidelines formulation and for decision-making. However, this traditional technique only partially yield information that clinicians, patients and policy-makers need to make informed decisions, since it usually compares only two interventions at the time. In the market, regardless the clinical condition under evaluation, usually many interventions are available and few of them have been studied in head-to-head studies. This scenario precludes conclusions to be drawn from comparisons of all interventions profile (e.g. efficacy and safety). The recent development and introduction of a new technique – usually referred as network meta-analysis, indirect meta-analysis, multiple or mixed treatment comparisons – has allowed the estimation of metrics for all possible comparisons in the same model, simultaneously gathering direct and indirect evidence. Over the last years this statistical tool has matured as technique with models available for all types of raw data, producing different pooled effect measures, using both Frequentist and Bayesian frameworks, with different software packages. However, the conduction, report and interpretation of network meta-analysis still poses multiple challenges that should be carefully considered, especially because this technique inherits all assumptions from pairwise meta-analysis but with increased complexity. Thus, we aim to provide a basic explanation of network meta-analysis conduction, highlighting its risks and benefits for evidence-based practice, including information on statistical methods evolution, assumptions and steps for performing the analysis.

Deriving percentage study weights in multi-parameter meta-analysis models: with application to meta-regression, network meta-analysis and one-stage individual participant data models

Many meta-analysis models contain multiple parameters, for example due to multiple outcomes, multiple treatments or multiple regression coefficients. In particular, meta-regression models may contain multiple study-level covariates, and one-stage individual participant data meta-analysis models may contain multiple patient-level covariates and interactions. Here, we propose how to derive percentage study weights for such situations, in order to reveal the (otherwise hidden) contribution of each study toward the parameter estimates of interest. We assume that studies are independent, and utilise a decomposition of Fisher's information matrix to decompose the total variance matrix of parameter estimates into study-specific contributions, from which percentage weights are derived. This approach generalises how percentage weights are calculated in a traditional, single parameter meta-analysis model. Application is made to one- and two-stage individual participant data meta-analyses, meta-regression and network (multivariate) meta-analysis of multiple treatments. These reveal percentage study weights toward clinically important estimates, such as summary treatment effects and treatment-covariate interactions, and are especially useful when some studies are potential outliers or at high risk of bias. We also derive percentage study weights toward methodologically interesting measures, such as the magnitude of ecological bias (difference between within-study and across-study associations) and the amount of inconsistency (difference between direct and indirect evidence in a network meta-analysis).