A statistical method for removing unbalanced trials with multiple covariates in meta-analysis.

In meta-analysis literature, there are several checklists describing the procedures necessary to evaluate studies from a qualitative point of view, whereas preliminary quantitative and statistical investigations on the "combinability" of trials have been neglected. Covariate balance is an important prerequisite to conduct meta-analysis. We propose a method to identify unbalanced trials with respect to a set of covariates, in presence of covariate imbalance, namely when the randomized controlled trials generate a meta-sample that cannot satisfy the requisite of randomization/combinability in meta-analysis. The method is able to identify the unbalanced trials, through four stages aimed at achieving combinability. The studies responsible for the imbalance are identified, and then they can be eliminated. The proposed procedure is simple and relies on the combined Anderson-Darling test applied to the Empirical Cumulative Distribution Functions of both experimental and control meta-arms. To illustrate the method in practice, two datasets from well-known meta-analyses in the literature are used.

Alanine aminotransferase and spleno-portal dynamics affect spleen stiffness measured by point shear-wave elastography in patients with chronic hepatitis C in the absence of significant liver fibrosis.

BACKGROUND: Spleen stiffness (SS) has gained a lot of interest in the context of liver cirrhosis and portal hypertension stratification. However, there is a paucity of data on confounding factors that may alter SS values. METHODS: Between January 2018 and October 2019, we enrolled 120 healthy subjects and 117 patients with hepatitis C virus (HCV) infection who did not have significant liver fibrosis (i.e., F0-1). Abdominal ultrasound evaluation was performed on each individual to measure portal vein diameter, portal flow velocity, spleen bipolar diameter, and splenic area. We also performed liver and spleen elastography. RESULTS: HCV patients had higher SS (p < 0.001), portal vein diameter (p = 0.031), portal flow velocity (p = 0.035), spleen bipolar diameter (p = 0.042) and area (p = 0.025), and ALT levels (p < 0.001). Linear regression models showed that SS increased by 3.220 kPa for each mm of portal vein diameter, by 0.7 kPa for each cm/s of portal flow velocity, by 2.239 kPa for each cm of spleen bipolar diameter, and by 0.233 kPa for each cm2 of spleen area. Patients with HCV infection were stratified according to median ALT levels (i.e. 32 IU/L). SS and spleno-portal axis parameters were significantly higher in patients with an ALT level > 32 IU/L. Besides, the relationship between SS and ALT was described by cubic polynomial regression according to the following equation: 11.735 + 0.404 (ALT)1 - 0.002 (ALT)2 + 4.26 × 10-6 (ALT)3. CONCLUSIONS: Our results bring new light to the role of inflammation as a confounding factor for SS measurement. Therefore, particular attention should be paid to serum transaminase for a correct evaluation of spleen elastography.

The role of elastography in alcoholic liver disease: fibrosis staging and confounding factors, a review of the current literature.

INTRODUCTION: Alcohol-related liver disease (ALD) was estimated to have a prevalence of 2% among the USA population. Since severe fibrosis in compensated patients is the main predictor of long-term survival, it is of utmost importance to early detect patients with severe fibrosis before decompensation occurs. Liver elastography has been used to stage liver fibrosis. However, there is a widespread lack in guidelines for the correct use of liver stiffness (LS) in ALD. EVIDENCE ACQUISITION: A structured search was carried out on MEDLINE/PubMed database. From the original 225 research articles identified, only 12 studies met the inclusion criteria, with 10 studies being eventually included. EVIDENCE SYNTHESIS: According to reported data, patients with aspartate aminotransferase (AST)>100 IU/L and 50 IU/L showed significantly higher values of LS if compared to patients with the same fibrosis stage. Also, excessive alcohol consumption greatly influences elastography, leading to false fibrosis staging. When LS values >5-6 kPa are detected, several aspects should be taken into account. First of all, the patient should be asked about the current alcohol consumption (i.e. active vs. abstinence, determination of abstinence period, and quantification of alcohol intake), and if the patient is an active drinker, liver elastography can be repeated after a complete abstinence period of at least two weeks. and if the patient is an active drinker, liver elastography can be repeated after a complete abstinence period of at least two weeks. Secondly, clinicians should check liver transaminases level, and if AST are above 100 IU/L, they should be aware of a possible overestimation of fibrosis. However, whether transaminases-adapted cut-off values should be used for ad-hoc decisions in patients with no time or option to withdraw from alcohol consumption is still a matter of debate. CONCLUSIONS: We hope that our review article may serve as a reference point in the prospect of futures guidelines.

Spleen Stiffness Probability Index (SSPI): A simple and accurate method to detect esophageal varices in patients with compensated liver cirrhosis.

INTRODUCTION AND OBJECTIVES: Recent findings pointed out that even low-risk esophageal varices (EVs) are markers of severe prognosis. Accordingly, we analyzed spleen stiffness (SS) as a non-invasive method to predict EVs of any grade in a cohort of patients with compensated liver cirrhosis. METHOD: We measured SS and liver stiffness (LS) using point-Shear-Wave Elastography (pSWE) with Philips Affiniti 70 system in 210 cirrhotic patients who had undergone endoscopic screening for EVs. We compared SS and LS predictive capability for EVs of any grade. RESULTS: SS was higher in cirrhotic patients with EVs if compared to patients without EVs (p<0.001). The cut-off analysis detected 31kPa (100% sensitivity and negative predictive value) as the value to rule-out EVs and 69kPa (100% specificity and positive predictive value) to rule-in EVs. Besides, we developed the Spleen Stiffness Probability Index (SSPI), that can provide a probability of presence/absence of EVs. SSPI was the best model according to all discriminative and calibration metrics (AIC=120, BIC=127, AUROC=0.95, Pseudo-R2=0.74). SS demonstrated higher correlation with spleen bipolar diameter and spleen surface (r=0.52/0.55) if compared to LS (r=0.30/0.25) - and with platelet count as well (r=0.67 vs r=0.4). CONCLUSION: SS showed significantly higher performance than other parameters, proving to be the best non-invasive test in the screening of EVs: by directly applying SS cut-off of 31kPa, our department could have safely avoided endoscopy in 36% of patients. Despite cut-off analyses, it was possible to create a probability model that could further stratify low-risk from high-risk patients (for any grade of EVs).

Evaluation of spleen stiffness in healthy volunteers using point shear wave elastography.

INTRODUCTION AND OBJECTIVES: This study aims to measure the values of spleen stiffness (SS) in healthy subjects, the inter-operator agreement in SS measurement, and to detect statistically significant correlations between SS and age, sex, weight, BMI, portal vein dynamics and splenic dimensions. MATERIALS AND METHODS: The study included 100 healthy volunteers who had no substantial alcohol intake (<30g/daily for man, <20g/daily women), were negative on hepatitis B, hepatitis C, HIV blood serology, and had any history of lymphoproliferative disorders. Abdominal ultrasound, liver and spleen elastography were performed on each patient to search for focal splenic lesions, bile tract or portal vein dilatation, moderate/severe liver steatosis, and to measure liver and spleen stiffness. RESULTS: The mean value was 18.14 (±3.08) kPa. In the group of men (n=49), the mean was 17.73 (±2.91) kPa, whereas in the group of women (n=51) it was 16.72 (±3.32) kPa. Statistical analyses showed no correlation between spleen stiffness and sex, age, weight, and BMI. Regarding their splenoportal axis, statistically significant differences in SS were found in the means of the two subgroups of subjects stratified by their portal flow velocity (p=0.003) and spleen area (p<0.001). Spearman's rank showed a weak association between SS and portal flow velocty (r=0.271) and splenic area (r=-0.237). ICC showed excellent (0.96) inter-operator agreement and Bland-Altman plot demonstrated no systematic over/under-estimation of spleen stiffness values. CONCLUSIONS: Our results may serve as a reference point in the evaluation of SS especially in patients affected by advanced liver disease.

Incidence of Hepatocellular Carcinoma in Patients With HCV-Associated Cirrhosis Treated With Direct-Acting Antiviral Agents.

BACKGROUND & AIMS: Studies have produced conflicting results of the incidence of hepatocellular carcinoma (HCC) in patients with hepatitis C virus-associated cirrhosis treated with direct-acting antivirals (DAAs). Data from clinics are needed to accurately assess the occurrence rate of HCC in patients with cirrhosis in the real world. METHODS: We collected data from a large prospective study of 2,249 consecutive patients (mean age = 65.4 years, 56.9% male) with hepatitis C virus-associated cirrhosis (90.5% with Child-Pugh class A and 9.5% with Child-Pugh class B) treated with DAAs from March 2015 through July 2016 at 22 academic and community liver centers in Sicily, Italy. HCC occurrence was evaluated by Kaplan-Meier curves. Cox regression analysis was used to identify variables associated with HCC development. RESULTS: A sustained virologic response (SVR) was achieved by 2,140 patients (total = 95.2%; 95.9% with Child Pugh class A and 88.3% with Child Pugh class B; P < .001). Seventy-eight patients (3.5%) developed HCC during a mean follow-up of 14 months (range = 6-24 months). At 1 year after exposure to DAAs, HCC developed in 2.1% of patients with Child-Pugh class A with an SVR and 6.6% of patients with no SVR and in 7.8% of patients with Child-Pugh class B with an SVR and 12.4% of patients with no SVR (P < .001 by log-rank test). Albumin level below 3.5 g/dL (hazard ratio = 1.77, 95% confidence interval = 1.12-2.82, P = .015), platelet count below 120 × 109/L (hazard ratio = 3.89, 95% confidence interval = 2.11-7.15, P < .001), and absence of an SVR (hazard ratio = 3.40, 95% confidence interval = 1.89-6.12, P < .001) were independently associated increased risk for HCC. The mean interval from exposure to DAAs to an HCC diagnosis was 9.8 months (range = 2-22 months) and did not differ significantly between patients with (n = 64, 9.2 months) and without (n = 14, 12.0 months) an SVR (P = .11). A larger proportion of patients with an SVR had a single HCC lesion (78% vs 50% without an SVR; P = .009) or an HCC lesion smaller than 3 cm (58% vs 28% without an SVR; P = .07). CONCLUSIONS: In an analysis of data from a large prospective study of patients with hepatitis C virus-associated compensated or decompensated cirrhosis, we found that the SVR to DAA treatment decreased the incidence of HCC over a mean follow-up of 14 months.

Nurse in patients' health status assessment: Data from a pilot study assessing agreement among nurse and gastroenterologist in computing IBD-clinical scores.

INTRODUCTION: Crohn's Disease (CD) and Ulcerative Colitis (UC) are chronic, systemic Inflammatory Bowel Diseases (IBDs) that need a multidisciplinary approach involving not only different medical specialists but also qualified nurses. AIM: We evaluated the concordance between IBD-nurse and physician in computing Clinical Activity Scores in IBD-patients treated with biologics. METHODS: We enrolled all consecutive IBD-patients treated with biologics in two referral centers for IBD-care. For each patient, a gastroenterologist and a nurse blindly filled-out a form to assess the Harvey-Bradshaw Index (HBI) in CD or the partial MAYO score in UC. All data were recorded to assess the beyond chance agreement (concordance) using the k statistic. RESULTS: 87 patients were enrolled. The agreement in all patients by k value was substantial (66%), ranging from moderate to substantial (95% CI from 51% to 80%). The main reason of disagreement was about the scoring of remission versus mild activity, and that of mild versus moderate activity, both in CD and UC. For the HBI, the best agreement was for well-being on the previous day (k 62%) and the least one for abdominal mass (k 35%). CONCLUSIONS: Our study shows an acceptable strength of agreement among nurse and gastroenterologist in evaluating the disease activity of IBD-patients through the calculation of clinical scores.

Evidence of bias in randomized clinical trials of hepatitis C interferon therapies.

INTRODUCTION: Bias may occur in randomized clinical trials in favor of the new experimental treatment because of unblinded assessment of subjective endpoints or wish bias. Using results from published trials, we analyzed and compared the treatment effect of hepatitis C antiviral interferon therapies experimental or control. METHODS: Meta-regression of trials enrolling naïve hepatitis C virus patients that underwent four therapies including interferon alone or plus ribavirin during past years. The outcome measure was the sustained response evaluated by transaminases and/or hepatitis C virus-RNA serum load. Data on the outcome across therapies were collected according to the assigned arm (experimental or control) and to other trial and patient-level characteristics. RESULTS: The overall difference in efficacy between the same treatment labeled experimental or control had a mean of +11.9% (p < 0.0001). The unadjusted difference favored the experimental therapies of group IFN-1 (+6%) and group IFN-3 (+10%), while there was no difference for group IFN-2 because of success rates from large multinational trials. In a meta-regression model with trial-specific random effects including several trial and patient-level variables, treatment and arm type remained significant (p < 0.0001 and p = 0.0009 respectively) in addition to drug-schedule-related variables. CONCLUSION: Our study indicates the same treatment is more effective when labeled "experimental" compared to when labeled "control" in a setting of trials using an objective endpoint and even after adjusting for patient and study-level characteristics. We discuss several factors related to design and conduct of hepatitis C trials as potential explanations of the bias toward the experimental treatment.

Assessing covariate imbalance in meta-analysis studies.

The main goal of meta-analysis is to combine data across studies or data sets to obtain summary estimates. In this paper, the novelty is to propose a statistical tool to assess a possible covariate imbalance in baseline variables to investigate similarity of trials. We conducted the detection of the covariate imbalance, first, through some graphical comparison of the empirical cumulative distribution functions or ECDFs, which are built by putting together arms or trials according to some risk factor, and second, through some non-parametric tests such as the Kolmogorov-Smirnov and the Anderson-Darling tests. To overcome the huge presence of ties, we conducted the statistical tests on perturbed versions of the original data sets. The applications concern two real meta-analyses of RCTs: the first one, on interferon-alpha treatment of chronic hepatitis C, with 107 studies involved, and the second one, on cholesterol-lowering treatment with statins, with 14 studies involved. The applications allow for analysis of both when risk factors reflecting demographic or clinical differences in experimental and control arms are balanced or not and when there are structural differences between the levels of some study variables, in order to proceed eventually with the pooling of the studies. We developed our suggestion, which is a quantitative way to assess combinability in meta-analysis, only with respect to RCTs, but it could be applied to a minor extent to non-RCTs.

An open-safety study of dual antiviral therapy in real-world patients with chronic hepatitis C.

PURPOSE: Treatment of patients with chronic hepatitis C with alpha-interferon and ribavirin usually produces adverse events within the first 3 months. We aimed to assess safety and predictors of discontinuation or dose modification of these drugs. METHODS: Observational study of 312 patients with predominantly genotype 1 chronic hepatitis C treated openly along 5 years in a clinical practice setting. RESULTS: Eighty-four percent of patients experienced at least one adverse event (853 events in total, 3.3 per patient on average). Incidence rate was higher during the first 90 days and decreased thereafter (<5%). Discontinuation rates at 30 and 90 days and at end of treatment were 2, 4 and 8%, respectively. Seventy percent of discontinuation cases were due to adverse events rather than to laboratory abnormalities. Serious adverse events were rare (<1%). Dose modifications were made in 158 patients (51%) on 237 occasions. After adjusting for covariates, older age was a predictor of early discontinuation, whereas HCV genotypes 1-4 and daily ribavirin dose of 1000 mg or more were predictors of dose modification. CONCLUSIONS: The majority of real-world patients with chronic hepatitis C tolerate acceptably dual therapy and very few discontinue it. Subjective decisions on dose reduction of either compound appears to have a major impact on adherence of patients. There is a need to better define, collect and analyse clinical features which may predict adverse events and safety-related decisions during therapy of chronic hepatitis C.

Music for colonoscopy: A single-blind randomized controlled trial.

BACKGROUND: Several methods have been reported to minimize patient discomfort during colonoscopy, none are currently recommended by clinical practice guidelines. We performed a single-blind randomized controlled trial to assess the efficacy of music for colonoscopy. METHODS: 109 patients were randomized to music-delivering or mute headphones before and during colonoscopy. Physicians were blinded to the trial. Sedation was given on demand. Primary outcome was pain measured on linear analogue scale from 0 to 10. Secondary endpoints were the difficulty of the procedure, need of sedation, overall patient satisfaction and willingness to repeat the procedure. RESULTS: Mean pain score was 5.9±2.2 in the control group vs. 3.8±1.9 in the music group (p<0.00001); correspondingly overall satisfaction and willingness to repeat the procedure were significantly improved by music and the difficulty perceived by physicians was significantly reduced. Total administered midazolam was 36mg in the control group vs. 13 in the music group (p<0.007), pethidine was 860mg vs. 465mg (p=0.07) and patients requiring sedation were 22 vs. 9, respectively (p=0.003). A multivariable analysis to adjust treatment effect for potential confounding factors confirmed the significant beneficial effect of music. CONCLUSIONS: Music significantly reduces discomfort and should be routinely offered to patients undergoing colonoscopy.

Patient's preferences for health scenarios associated with hepatitis C and its treatment.

BACKGROUND AND AIM: Antiviral therapy with peg-interferon and ribavirin induces sustained virus eradication in 40%-80% of patients with chronic hepatitis C virus (HCV). We investigated patient views on their involvement in therapeutic decision making and on the desirability of disease and treatment-related outcomes. METHODS: The control preferences and visual analog scales were administered in a pencil and paper format to a series of 45 patients in order to assess their decisional role, preferences for scenarios of HCV disease and antiviral treatment, and estimates of success required to recommend treatment. RESULTS: The preferred decisional role of patients was passive in 26 (58%), collaborative in 12 (27%) and active in 7 (15%). Median preference scores ranged from 0.30 to 0.90 for scenarios of disease, from 0.05 to 0.80 for side effects and from 25% to 100% for estimates of benefit to recommend treatment. CONCLUSIONS: Our patients prefer to defer to the doctor the final decision in starting therapy in a context of shared decision making. Reported preferences for HCV scenarios are in the range discussed in the literature. The wide variability in the values attributed to side effects by patients with chronic hepatitis C as well as in expected probabilities of successful treatment suggests a need for decision analysis tailored to the individual patient.

A decade of trials of interferon-alpha for chronic hepatitis C. A meta-regression analysis.

The most relevant randomized controlled trials of interferon-alpha (IFN) for naive patients with chronic hepatitis C (CHC) published in a decade, just before appearance of pegylated IFN trials in 2000, were included in this paper. Its purpose is to review the relationship between sustained biochemical response in active versus control group versus usual clinical variables as IFN regimens, cirrhosis, genotype and versus less frequently addressed variables as funding, methodological quality or location of principal author. Meta-analysis estimates of global treatment effect varied according to trial design: group 1=IFN versus placebo/no treatment, 32 RCTs, 2499 pts, OR 9.5 (6.3-14.2); group 2a=comparison of IFN schedules, 43 RCTs, 7454 pts, OR 1.6 (1.4-1.9); group 2b=IFN+other drugs versus standard IFN, 30 RCTs, 4737 pts, OR 2.0 (1.6-2.6). Fixed effects (arm-level) meta-regression on the complete data set (171 arms, 10,580 pts) revealed that sustained response was most likely in experimental arms of IFN+ribavirin or other drugs (OR 2.4), arms using yearly schedule (OR 2.0), trial principal author from Asia (OR 1.7), trial sample size >200 (OR 1.4) and arms enrolling less than 50% of cirrhotics (OR 1.3). Moreover, focus was on some significant interactions too, as the effect of trial's quality interacting to the recorded funding (more benefit if no-profit, less if for-profit) and the effect of trial funding interacting to the location of first author (more benefit if from Asia). Three main effects (experimental arm, cirrhosis, funding) and one interaction (funding*location of principal author) explained 31% of between study variability in a random-effect meta-regression. In a subgroup analysis on a data set including available information on HCV genotype (93 arms, around 7000 pts), meta-regression revealed that genotype 1 or 4 less than 50% per arm and specialistic journal were significant predictors of either biochemical (transaminases) or virological (HCV-RNA) sustained response, in a model including the same main effects identified in the complete data set analysis. Finally, although mostly captured by different IFN regimens along time, heterogeneity of effect in a large set of (not-pegylated) IFN trials was also explained by HCV genotype and variables of quality and reporting, such as trial's principal author from Asia.