Hierarchical Random-effects Model Research Articles

A Bayesian Meta-Analysis on the Association Between Beta-Carotene and Bone Mineral Density

The results of epidemiological studies on the association between dietary β-carotene and bone mineral density (BMD) are remaining controversial. This study aimed to investigate the association between dietary β-carotene and BMD using a Bayesian meta-analysis approach. We searched PubMed, Embase, Cochrane Central Register, and Medline databases up to July 2019. Two authors independently screened articles, identified, and assessed the risk of bias. Bayesian hierarchical random effect models were used to synthesize data from individual studies. Five studies that consisted of three cross-sectional studies and two cohort studies involving 12,521 men and women met the inclusion criteria. Overall, dietary β-carotene reduced loss of BMD (risk ratio (RR) 0.89; 95% credible interval (CrI) 0.77–0.99). The probability of that β-carotene dietary reduced loss of BMD was 0.99. Sensitivity analyses indicate that approximately similar findings were observed across different prior information. We suggested that dietary intakes of β-carotene was associated with reduced the loss of bone mineral density for the general population. Further research will be recommended to confirm our findings.

Social transition and health inequality in China: an age-period-cohort analysis

ObjectivesTo examine self-rated health (SRH) trajectories by age, period and cohort in the dynamic environment of China and to focus on sex, educational, family economic status (FES) and regional disparities in these temporal trajectories. Study designRepeated cross-sectional data from the World Value Survey from 1990 to 2012 (n = 7731) were used. MethodsAn age-period-cohort (APC) model with an intrinsic estimator method was applied, using SRH as a continuous variable to analyse SRH trajectories by age, period and cohort in different groups in China. Robust analyses were conducted using SRH as a binary variable and a hierarchical APC cross-classified random effects model. ResultsSRH was adversely associated with age, and a global trend of declining SRH was observed between 1990 and 2012; however, SRH generally increased across successive cohorts in China. SRH was better in males, individuals from the eastern region, and in those with higher levels of education and higher FES. Regional and FES disparities in SRH increased with age and across successive cohorts; however, educational disparities in SRH first decreased and then increased with age and across successive cohorts, and sex disparities in SRH decreased across successive cohorts, in general. ConclusionsA well-performing social environment is related to better population health. The cumulative disadvantage in health among individuals from underprivileged regions/families with less education should be the focus of social transition in China.

Modeling repeated labor curves in consecutive pregnancies: Individualized prediction of labor progression from previous pregnancy data.

The measurement of cervical dilation of a pregnant woman is used to monitor the progression of labor until 10 cm when pushing begins. There is anecdotal evidence that labor tracks across repeated pregnancies; moreover, no statistical methodology has been developed to address this important issue, which can help obstetricians make more informed clinical decisions about an individual woman's progression. Motivated by the NICHD Consecutive Pregnancies Study (CPS), we propose new methodology for analyzing labor curves across consecutive pregnancies. Our focus is both on studying the correlation between repeated labor curves on the same woman and on using the cervical dilation data from prior pregnancies to predict subsequent labor curves. We propose a hierarchical random effects model with a random change point that characterizes repeated labor curves within and between women to address these issues. We employ Bayesian methodology for parameter estimation and prediction. Model diagnostics to examine the appropriateness of the hierarchical random effects structure for characterizing the dependence structure across consecutive pregnancies are also proposed. The methodology was used in analyzing the CPS data and in developing a predictor for labor progression that can be used in clinical practice.

A Class of Hierarchical Random Effect Models for Posterior Insurance Ratemaking

Poisson random effect panel data model is widely used in non-life insurance for posterior rate making. Recently, Lee et al. (2020) identify one shortcoming of the shared random effect assumption, which introduces both the marginal count distribution and the serial dependence between counts at different periods. Therefore, it is not possible to accommodate separately these two features. This shortcoming also exists for any other nonlinear (say, gamma, or negative binomial) random effect regression model. In this note we propose a family of hierarchical random effect model, allowing to separate its effect on the marginal and the serial dependence structure of the observable dependent variables. The model we propose has several advantages. First it is very tractable, with closed form likelihood function and posterior rate making formula. Second, it is very general and suitable for a large number of marginal distributions of the random effect and conditional distributions of the response variable, such as the gamma-Poisson, beta-negative binomial pair distributions, as well as some new non conjugate pairs. Finally, it is also applicable to the simultaneous pricing of a large number of risks encountered in fleet insurance.

Influence diagnostics and outlier detection for meta-analysis of diagnostic test accuracy.

Meta-analyses of diagnostic test accuracy (DTA) studies have been gaining prominence in research in clinical epidemiology and health technology development. In these DTA meta-analyses, some studies may have markedly different characteristics from the others and potentially be inappropriate to include. The inclusion of these "outlying" studies might lead to biases, yielding misleading results. In addition, there might be influential studies that have notable impacts on the results. In this article, we propose Bayesian methods for detecting outlying studies and their influence diagnostics in DTA meta-analyses. Synthetic influence measures based on the bivariate hierarchical Bayesian random effects models are developed because the overall influences of individual studies should be simultaneously assessed by the two outcome variables and their correlation information. We propose four synthetic measures for influence analyses: (a) relative distance, (b) standardized residual, (c) Bayesian p-value, and (d) influence statistic on the area under the summary receiver operating characteristic curve. We also show that conventional univariate Bayesian influential measures can be applied to the bivariate random effects models, which can be used as marginal influential measures. Most of these methods can be similarly applied to the frequentist framework. We illustrate the effectiveness of the proposed methods by applying them to a DTA meta-analysis of ultrasound in screening for vesicoureteral reflux among children with urinary tract infections.

Have Hospice Costs Increased After Implementation of the Hospice Quality–Reporting Program?

ContextThe Centers for Medicare & Medicaid Services Hospice Quality–Reporting Program introduced the requirement that hospices nationwide begin collecting and submitting standardized patient-level quality data on July 1, 2014. ObjectivesThis study examined whether this requirement has increased hospice total costs, general costs, and visiting services costs. MethodsWe conducted a cross-sectional study using data from the 2012 and 2014 Medicare hospice cost reports linked to hospice claims. We measured total costs per patient day (PPD), general costs PPD, and visiting services costs PPD for freestanding hospices. We estimated the incremental costs of operating in 2014 vs. 2012 using hierarchical random effects models and adjusting for year, wage index, care volume, case-mix, and hospice and market characteristics, stratified by hospice ownership type. ResultsBoth for-profit and nonprofit hospices reported higher total costs PPD and general services costs PPD in 2014 than 2012. Nonprofit hospices also reported higher general costs PPD in 2014 than 2012. In adjusted models, the total costs PPD in 2014 were $10.55 higher than in 2012 for nonprofit hospices and $6.43 higher for for-profit hospices. The increase in general costs PPD and visiting services costs PPD ranged from $3.15 to $5.87 by ownership and type of costs. Both for-profit and nonprofit hospices showed lower costs PPD for all types associated with more patients and longer length of stay. ConclusionHospice costs increased after the Centers for Medicare & Medicaid Services Hospice Quality–Reporting Program quality data collection/submission requirement. Complementary studies need to understand whether increased costs brought additional benefits.

At what age do normal weight Canadian children become overweight adults? Differences according to sex and metric

Background: The prevalence of overweight and obesity doubles between adolescence and young adulthood; however, the exact age, and appropriate metric to use to identify when overweight develops is still debated.Aim: To examine the age of onset of overweight by sex and four metrics: body mass index (BMI), fat mass (%FM), waist circumference (WC) and waist-to-height ratio (WHtR).Subjects and methods: Between 1991 and 2017, serial measures of body composition were taken on 237 (108 males) individuals (aged 8–40 years of age). Hierarchical random effects models were used to develop growth curves. Curves were compared to BMI, %FM and WC overweight age- and sex-specific cut-points.Results: In males, the BMI growth curve crossed the cut-point at 22.0 years, compared to 23.5 and 26.5 years for WHtR and %FM, respectively; WC cut-off was not reached until 36 years. In females, the BMI growth curve crossed the overweight cut-point at 21.5 years, compared to 14.2 years for %FM and 21.9 and 27.5 years for WC and WHtR, respectively.Conclusion: In summary, overweight onset occurs during young adulthood with the exception of WC in males. BMI in males and %FM in females were the metrics identifying overweight the earliest.

Depression, anxiety and delinquency: Results from the Pittsburgh Youth Study

PurposeThe main aim of this research is to investigate to what extent within-individual changes in anxiety and depression are related to within-individual changes in theft and violence. MethodsThe youngest sample of the Pittsburgh Youth Study (PYS), a prospective longitudinal survey of 503 boys followed up from age 7 onwards, was analyzed. Depression and anxiety were measured for boys from ages 11 to 16 as were moderate and serious forms of self-reported theft and violence. A hierarchical linear random effects model was used to investigate anxiety and depression as potential causes or outcomes of these forms of delinquency. ResultsThe results showed that the between-individual correlations were consistently higher than the corresponding within-individual correlations, and provided little evidence to discern the directionality of the potential relationships between depression, anxiety and delinquency. Using a random effects approach, there was limited evidence that prior depression or anxiety was related to later offending, but there was evidence that offending (particularly theft and serious violence) was associated with later increases in anxiety, and to a lesser extent depression. ConclusionsThis study indicates that depression and anxiety were outcomes of offending. If replicated, this would suggest that evidence-based interventions which reduced offending would have a desirable influence in reducing depression and anxiety. However, interventions for depression should still form part of responsive interventions. More research which explores within-individual changes in longitudinal studies with repeated measures is needed.

Probiotics for Preterm Infants: A Strain-Specific Systematic Review and Network Meta-analysis.

Several randomized controlled trials (RCTs) on the use of probiotics to reduce morbidity and mortality in preterm infants have provided inconsistent results. Although meta-analyses that group all of the used strains together are suggesting efficacy, it is not possible to determine the most effective strain that is more relevant to the clinician. We therefore used a network meta-analysis (NMA) approach to identify strains with greatest efficacy. A PubMed search identified placebo-controlled or head-to-head RCTs investigating probiotics in preterm infants. From trials that recorded mortality, necrotizing enterocolitis, late-onset sepsis, or time until full enteral feeding as outcomes, data were extracted and Bayesian hierarchical random-effects models were run to construct a NMA. Fifty-one RCTs involving 11,231 preterm infants were included. Most strains or combinations of strains were only studied in one or a few RCTs. Only 3 of 25 studied probiotic treatment combinations showed significant reduction in mortality rates. Seven treatments reduced necrotizing enterocolitis incidence, 2 reduced late-onset sepsis, and 3 reduced time until full enteral feeding. There was no clear overlap of strains, which were effective on multiple outcome domains. This NMA showed efficacy in reducing mortality and morbidity only in a minority of the studied strains or combinations. This may be due to an inadequate number, or size, of RCTs, or due to a true lack of effect for certain species. Further large and adequately powered RCTs using strains with the greatest apparent efficacy will be needed to more precisely define optimal treatment strategies.

Bayesian Hierarchical Random Effects Models in Forensic Science.

Statistical modeling of the evaluation of evidence with the use of the likelihood ratio has a long history. It dates from the Dreyfus case at the end of the nineteenth century through the work at Bletchley Park in the Second World War to the present day. The development received a significant boost in 1977 with a seminal work by Dennis Lindley which introduced a Bayesian hierarchical random effects model for the evaluation of evidence with an example of refractive index measurements on fragments of glass. Many models have been developed since then. The methods have now been sufficiently well-developed and have become so widespread that it is timely to try and provide a software package to assist in their implementation. With that in mind, a project (SAILR: Software for the Analysis and Implementation of Likelihood Ratios) was funded by the European Network of Forensic Science Institutes through their Monopoly programme to develop a software package for use by forensic scientists world-wide that would assist in the statistical analysis and implementation of the approach based on likelihood ratios. It is the purpose of this document to provide a short review of a small part of this history. The review also provides a background, or landscape, for the development of some of the models within the SAILR package and references to SAILR as made as appropriate.

Comparison of treatment strategies for femoro-popliteal disease: A network meta-analysis.

We sought to compare treatment strategies in a Bayesian network meta-analysis of randomized controlled trials. Peripheral artery disease (PAD) is a prevalent morbidity that is treated with various strategies. We performed a MEDLINE search for randomized studies comparing at least 2 treatment strategies, including bypass surgery, percutaneous transluminal angioplasty (PTA) balloons, stents, covered stents, drug-eluting stents (DES), and drug-coated balloons (DCB), in patients with native femoro-popliteal disease. Mixed treatment comparison model generation was performed to directly and indirectly compare the strategies in terms of restenosis and target lesion revascularization (TLR) presented as odds ratios (OR, [95% credible intervals]). Twenty-nine studies with 4,820 patients were included in the present study. PTA was the largest group with 1,900 patients, followed by DCB (n = 1,343), bare metal stents (n = 941), covered stents (n = 304), DES (n = 236), and bypass (n = 92). Mean age was 68 ± 9 years, 64% were male, 37% diabetic, and 55% smokers. Mean lesion length was 77 ± 44 mm, and 39% were total occlusions. Bayesian hierarchical random-effects model demonstrated all treatments were significantly better than, or had a trend toward superiority over, PTA, with DCB ranking well in both restenosis (OR = 0.29, [0.17-0.47]) and TLR (OR = 0.31, [0.20-0.46]). Nonetheless, none of the therapies showed superiority in terms of survival or amputations. Treatment of femoro-popliteal disease has significantly evolved in recent years, with higher rates of patency and freedom from TLR. However, the utility of these treatment strategies in terms of reduction of amputations and overall survival remains in question.

The Effect of China's National Essential Medicine Policy on Health Expenses: Evidence From a National Study.

To increase use of medical service across the country, the Chinese government has tried to improve equity in health care access and reduce patients’ medical expenses. For this purpose, the National Essential Medicine Policy (NEMP) was introduced in 2009 to mandate the distribution of medicines to health care facilities at a low cost and without profit. This study aims to evaluate the effect of the essential medicine policy on average per-visit expenses for outpatient and inpatient services. The annual national surveillance system data covering all the grassroots-level primary health care facilities (PHFs) in 2675 counties and 31 provinces in China during 2008 to 2012 were used in this study. The 4-level hierarchical random effects models were utilized to deal with possible dose-response effects of the policy and possible variations of such effects at the provincial, county, and facility levels. Our research findings suggest that the NEMP had positive effects in reducing both outpatient and inpatient expenses at grassroots level, and the policy effects tended to be greater as the exposure time increased. This study provides implications on reforming China’s health system and its medicine cost control policies.

The unrealized potential: cohort effects and age-period-cohort analysis.

This study aims to provide a systematical introduction of age-period-cohort (APC) analysis to South Korean readers who are unfamiliar with this method (we provide an extended version of this study in Korean). As health data in South Korea has substantially accumulated, population-level studies that explore long-term trends of health status and health inequalities and identify macrosocial determinants of the trends are needed. Analyzing long-term trends requires to discern independent effects of age, period, and cohort using APC analysis. Most existing health and aging literature have used cross-sectional or short-term available panel data to identify age or period effects ignoring cohort effects. This under-use of APC analysis may be attributed to the identification (ID) problem caused by the perfect linear dependency across age, period, and cohort. This study explores recently developed three APC models to address the ID problem and adequately estimate the effects of A-P-C: intrinsic estimator-APC models for tabular age by period data; hierarchical cross-classified random effects models for repeated cross-sectional data; and hierarchical APC-growth curve models for accelerated longitudinal panel data. An analytic exemplar for each model was provided. APC analysis may contribute to identifying biological, historical, and socioeconomic determinants in long-term trends of health status and health inequalities as well as examining Korean’s aging trajectories and temporal trends of period and cohort effects. For designing effective health policies that improve Korean population’s health and reduce health inequalities, it is essential to understand independent effects of the three temporal factors by using the innovative APC models.

Leveraging public health's participation in a Health Information Exchange to improve communicable disease reporting.

Infectious diseases can appear and spread rapidly. Timely information about disease patterns and trends allows public health agencies to quickly investigate and efficiently contain those diseases. But disease case reporting to public health has traditionally been paper-based, resulting in somewhat slow, burdensome processes. Fortunately, the expanding use of electronic health records and health information exchanges has created opportunities for more rapid, complete, and easily managed case reporting and investigation. To assess how this new service might impact the efficiency and quality of a public health agency's case investigations, we compared the timeliness of usual case investigation to that of case investigations based on case report forms that were partially pre-populated with electronic data. Between September 2013-March 2014, chlamydia disease report forms for certain clinics in Indianapolis were electronically pre-populated with clinical, lab and patient data available through the Indiana Health Information Exchange, then provided to the patient’s doctor. Doctors could then sign the form and deliver it to public health for investigation and population-level disease tracking. Methods: We utilized a novel matched case analysis of timeliness changes in receipt and processing of communicable disease report forms. Each Chlamydia cases reported with the pre-populated form were matched to cases reported in usual ways. We assessed the time from receipt of the case at the public health agency: 1) inclusion of the case into the public health surveillance system and 2) to close to case. A hierarchical random effects model was used to compare mean difference in each outcome between the target cases and the matched cases, with random intercepts for case. Twenty-one Chlamydia cases were reported to the public health agency using the pre-populated form. Sixteen of these pre-populated form cases were matched to at least one other case, with a mean of 23 matches per case. The mean Reporting Lag for the pre-populated form cases was 2.5 days, which was 2.7 days shorter than the mean Reporting Lag for the matched controls (p = <0.001). The mean time to close a pre-populated form case was 4.7 days, which was 0.2 days shorter than time to close for the matched controls (p = 0.792). Use of pre-populated forms significantly decreased the time it took for the local public health agency to begin documenting and closing chlamydia case investigations. Thoughtful use of electronic health data for case reporting may decrease the per-case workload of public health agencies, and improve the timeliness of information about the pattern and spread of disease.

A hierarchical estimation of school quality capitalisation in house prices in Orange County, California

Bidding for proximity to a good school can lead to a pattern of spatial distribution in which households with similar socio-economic status and willingness-to-pay for school quality cluster together. In this paper, we adopt a three-level hierarchical framework using residential house prices in Orange County, California, in 2001 and 2011, to estimate how much homebuyers pay for school quality. Our data show that, during this period, the Academic Performance Index (API) scores of elementary schools in Orange County increased by 16.4% yet converged while the house prices rose by 50.3%. The variation in house prices attributed to school district boundaries was at the same level in both years, but the variation in the API scores shrank. Using a hierarchical random effects model, our estimation results show that, on average, a 10% increase in the API raised the house prices by 1.9% in 2001 and by 3.4% in 2011. Ten years apart, a one standard deviation increase in school quality in the sample increased house prices by a surprisingly similar percentage: 2.7% in 2001, and 2.6% in 2011, respectively. Our findings also reveal that, in both years, there was a significant spatial heterogeneity of school premiums in house prices across school districts. This research provides a spatial understanding of the education capitalisation effects and sheds light on the effectiveness of urban education policy.

Safety Profile of Biologic Drugs in the Treatment of Inflammatory Bowel Diseases: A Systematic Review and Network Meta-analysis of Randomized Controlled Trials.

Biologic drugs are used in innovative therapies for the management of inflammatory bowel diseases (IBDs). The aim of this study was to compare the safety profile of biologic drugs in patients with IBD. A systematic literature search was performed using PubMed, Embase, and CENTRAL databases, up to 22 August 2016. We included randomized, placebo-controlled, or head-to-head clinical trials that compared the safety of different biologics in patients with IBDs. Two reviewers independently conducted the search and selection of studies and rated each trial's risk of bias. The network meta-analysis (NMA) was conducted for a mid-term (20-30weeks) and long-term (≥52weeks) follow-up with a Bayesian hierarchical random effects model using the ADDIS® software. The PROSPERO registration number was CRD42015029884. Sixteen randomized controlled trials were included in the systematic review with NMA. In the case of the mid-term follow-up, it was possible to conduct the NMA for assessing the relative safety profile of certolizumab pegol and infliximab, and in the case of the long-term follow-up, of infliximab, adalimumab, golimumab, and vedolizumab. There were no significant differences in the rate of adverse events in patients treated with all analyzed biologic drugs for IBD. The analysis of probability for being the safest treatment showed that infliximab was the best option in most analyzed endpoints both in mid-term and in long-term follow-ups. We showed no significant differences in the relative safety profile of the analyzed biologic drugs. Further studies are needed to confirm our findings, including head-to-head comparisons between these drugs.

Accuracy of Onsite Tests to Detect Asymptomatic Bacteriuria in Pregnancy: A Systematic Review and Meta-analysis.

To estimate the accuracy of onsite tests to detect asymptomatic bacteriuria among pregnant women. We searched MEDLINE, EMBASE, Web of Science, Scopus, and Latin-American Literature from inception until June 2015 without language restrictions. The ClinicalTrials.gov register database was screened to identify any recently completed studies. Two independent reviewers selected studies that recruited asymptomatic pregnant women to evaluate the accuracy of onsite tests in detecting the presence of bacteria in the urine using urine culture as a reference standard. Women's characteristics, study design, urine sample collection, and handling were extracted along with the test accuracy data. Where possible, we pooled the data using a bivariate, hierarchical random-effects model. Of 1,360 screened references, 27 articles (13,641 women) with test accuracy data on nine tests met the inclusion criteria. The most commonly evaluated test was urine dipstick. The pooled sensitivity and specificity of nitrites detected by dipstick to detect asymptomatic bacteriuria were 0.55 (95% confidence interval [CI] 0.42-0.67) and 0.99 (95% CI 0.98-0.99), respectively. The Griess test to detect nitrites had a sensitivity of 0.65 (95% CI 0.50-0.78) and specificity of 0.99 (95% CI 0.98-1.00). Dipslide with Gram staining had a pooled sensitivity of 0.86 (95% CI 0.80-0.91) and specificity of 0.97 (95% CI 0.93-0.99). The specificity of onsite tests is high; however, the sensitivity is not with the result that they will fail to detect a substantial number of cases of asymptomatic bacteriuria. PROSPERO International prospective register of systematic reviews, http://www.crd.york.ac.uk/PROSPERO/, CRD42015027905.

Comparative safety profile of hyaluronic acid products for knee osteoarthritis: a systematic review and network meta-analysis

Intra-articular (IA) hyaluronic acid (HA) is considered a safer alternative to oral Non-Steroidal Antiinflammatory Drugs (NSAIDs) and opioids for knee osteoarthritis (OA). A recent review raised potential safety concerns about HA, warranting further review of safety outcomes. We examined the risks of HA compared with IA placebo and investigated whether the risks vary among individual HA preparations. We searched all relevant databases from inception to October 2015 and sought unpublished data. We included all knee OA trials which compared any of the 18 HA products and reported on adverse events (AEs) and withdrawals. We calculated odds ratios for safety data reported at the longest follow-up. Network meta-analysis was performed using a Bayesian hierarchical random effects model for mixed multiple treatment comparisons. We identified 74 studies involving 13,032 participants aged between 45 and 75 years. The proportion of women ranged from 28% to 100%. The overall incidence of local reactions reported across all products was 8.5%. Commonly reported AEs were transient local reactions, such as pain, swelling and arthralgia, which subsided rapidly. None of the HA products were statistically significantly different from IA placebo or from each other with regard to incidence of AEs. Three treatment-related serious adverse events (SAEs) were reported among 9214 participants. Given the very low incidence of any particular AEs, we conclude that HA products are relatively well tolerated. These products have a similar safety profile compared to each other. This information along with the comparative effectiveness profile and relative cost would be helpful for clinicians in delivering individualized patient care.

Safety profile of biologic drugs in the therapy of Crohn disease: A systematic review and network meta-analysis

Crohn disease (CD) is an inflammatory bowel disease which occurs especially in developed countries of Western Europe and North America. The aim of the study was to compare the safety profile of biologic drugs in patients with CD. A systematic literature search was performed using PubMed, Embase, and CENTRAL databases, until April 27, 2016. We included randomized controlled trials (RCTs) that compared the safety of biologic drugs (infliximab, adalimumab, vedolizumab, certolizumab pegol, and ustekinumab) with one another or with placebo in patients with CD. The network meta-analysis (NMA) was conducted for an induction phase (6-10 weeks) and maintenance phase (52-56 weeks) with a Bayesian hierarchical random effects model in the ADDIS® software. The PROSPERO registration number was CRD42016032606. Ten RCTs were included in the systematic review with NMA. In the case of the induction phase, the NMA could be conducted for the assessment of the relative safety profile of adalimumab, vedolizumab, certolizumab pegol, and ustekinumab, and in the case of the maintenance phase-of infliximab, adalimumab, and vedolizumab. There were no significant differences in the rate of adverse events in patients treated with biologics. Statistical analysis revealed that vedolizumab had the greatest probability of being the safest treatment in most endpoints in the induction phase and adalimumab-in the maintenance phase. No significant differences between the biologics in the relative safety profile analysis were observed. Further studies are needed to confirm our findings, including head-to-head comparisons between the analyzed biologics.

Safety Profile of Biologic Drugs in the Therapy of Ulcerative Colitis: A Systematic Review and Network Meta-Analysis.

We compared the safety profile of biologic drugs in patients with moderately to severely active ulcerative colitis (UC). A systematic literature search was performed using Medline (PubMed), Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) databases through February 9, 2016. We included randomized controlled trials (RCTs) that compared the safety of biologic drugs (infliximab, adalimumab, golimumab, and vedolizumab) with one another or with placebo in patients with UC. Two reviewers independently conducted the search and selection of studies and rated the risk of bias in each trial. The network meta-analysis (NMA) was conducted for an induction phase (6-8 weeks) and maintenance phase (52-54 weeks) with a Bayesian hierarchical random effects model in Aggregate Data Drug Information System (ADDIS) software. The PROSPERO registration number was CRD42016032607. Seven RCTs were included in the systematic review with NMA. In the case of the induction phase, the NMA could be conducted for the assessment of the relative safety profile of adalimumab, golimumab, and vedolizumab, and in the case of the maintenance phase of infliximab, adalimumab, golimumab, and vedolizumab. The methodological quality of the included RCTs was evaluated as low risk of bias, but high risk of bias in the case of attrition bias (incomplete outcome data) according to the Cochrane criteria. No significant differences were found in the rate of adverse events in patients treated with the reviewed biologics. Vedolizumab was most likely to have the most favorable safety profile in the induction phase as was infliximab for the maintenance phase. The assessment of the relative safety profile revealed no significant differences between the biologic drugs. Further studies are needed to confirm our findings including head-to-head comparisons between the analyzed biologics.