Search

A P value, or the magnitude or direction of results can influence decisions about whether, when, and how research findings are disseminated. Regardless of whether an entire study or a particular study result is unavailable because investigators considered the results to be unfavorable, bias in a meta‐analysis may occur when available results differ systematically from missing results. In this article, we summarize the empirical evidence for various reporting biases that lead to study results being unavailable for inclusion in systematic reviews, with a focus on health research. These biases include publication bias and selective nonreporting bias. We describe processes that systematic reviewers can use to minimize the risk of bias due to missing results in meta‐analyses of health research, such as comprehensive searches and prospective approaches to meta‐analysis. We also outline methods that have been designed for assessing risk of bias due to missing results in meta‐analyses of health research, including using tools to assess selective nonreporting of results, ascertaining qualitative signals that suggest not all studies were identified, and generating funnel plots to identify small‐study effects, one cause of which is reporting bias. Highlights: Bias in a meta‐analysis may occur when available results differ systematically from missing results.Several different tools, plots, and statistical methods have been designed for assessing risk of bias due to missing results in meta‐analyses. These include comparison of prespecified analysis plans with completed reports to detect selective nonreporting of results, consideration of qualitative signals that suggest not all studies were identified, and the use of funnel plots to identify small‐study effects, for which reporting bias is one of several causes.Information from approaches such as funnel plots and selection models is more difficult to interpret than from less subjective approaches such as detection of incompletely reported results in studies for which prespecified analysis plans were available. [ABSTRACT FROM AUTHOR]

Background: Unhealthy alcohol use among persons living with HIV (PLWH) is linked to significant morbidity, and use of alcohol services may differ by HIV status. Our objective was to compare unhealthy alcohol use screening and treatment by HIV status in primary care. Methods: Cohort study of adult (≥18 years) PLWH and HIV‐uninfected participants frequency matched 20:1 to PLWH by age, sex, and race/ethnicity who were enrolled in a large integrated healthcare system in the United States, with information ascertained from an electronic health record. Outcomes included unhealthy alcohol screening, prevalence, provider‐delivered brief interventions, and addiction specialty care visits. Other predictors included age, sex, race/ethnicity, neighborhood deprivation index, depression, smoking, substance use disorders, Charlson comorbidity index, prior outpatient visits, insurance type, and medical facility. Cox proportional hazards models were used to compute hazard ratios (HR) for the outcomes of time to unhealthy alcohol use screening and time to first addiction specialty visit. Poisson regression with robust standard errors was used to compute prevalence ratios (PR) for other outcomes. Results: 11,235 PLWH and 227,320 HIV‐uninfected participants were included. By 4.5 years after baseline, most participants were screened for unhealthy alcohol use (85% of PLWH and 93% of HIV‐uninfected), but with a lower rate among PLWH (adjusted HR 0.84, 95% CI 0.82 to 0.85). PLWH were less likely, compared with HIV‐uninfected participants, to report unhealthy drinking among those screened (adjusted PR 0.74, 95% CI 0.69 to 0.79), and among those who screened positive, less likely to receive brief interventions (adjusted PR 0.82, 95% CI 0.75 to 0.90), but more likely (adjusted HR 1.7, 95% CI 1.2 to 2.4) to have an addiction specialty visit within 1 year. Conclusions: Unhealthy alcohol use was lower in PLWH, but the treatment approach by HIV status differed. PLWH reporting unhealthy alcohol use received less brief interventions and more addiction specialty care than HIV‐uninfected participants. [ABSTRACT FROM AUTHOR]

Meta‐analyses based on systematic literature reviews are commonly used to obtain a quantitative summary of the available evidence on a given topic. However, the reliability of any meta‐analysis is constrained by that of its constituent studies. One major limitation is the possibility of small‐study effects, when estimates from smaller and larger studies differ systematically. Small‐study effects may result from reporting biases (ie, publication bias), from inadequacies of the included studies that are related to study size, or from reasons unrelated to bias. We propose two estimators based on the median and mode to increase the reliability of findings in a meta‐analysis by mitigating the influence of small‐study effects. By re‐examining data from published meta‐analyses and by conducting a simulation study, we show that these estimators offer robustness to a range of plausible bias mechanisms, without making explicit modelling assumptions. They are also robust to outlying studies without explicitly removing such studies from the analysis. When meta‐analyses are suspected to be at risk of bias because of small‐study effects, we recommend reporting the mean, median and modal pooled estimates. [ABSTRACT FROM AUTHOR]

When estimating the effect of treatment on HIV using data from observational studies, standard methods may produce biased estimates due to the presence of time-dependent confounders. Such confounding can be present when a covariate, affected by past exposure, is both a predictor of the future exposure and the outcome. One example is the CD4 cell count, being a marker for disease progression for HIV patients, but also a marker for treatment initiation and influenced by treatment. Fitting a marginal structural model (MSM) using inverse probability weights is one way to give appropriate adjustment for this type of confounding. In this paper we study a simple and intuitive approach to estimate similar treatment effects, using observational data to mimic several randomized controlled trials. Each 'trial' is constructed based on individuals starting treatment in a certain time interval. An overall effect estimate for all such trials is found using composite likelihood inference. The method offers an alternative to the use of inverse probability of treatment weights, which is unstable in certain situations. The estimated parameter is not identical to the one of an MSM, it is conditioned on covariate values at the start of each mimicked trial. This allows the study of questions that are not that easily addressed fitting an MSM. The analysis can be performed as a stratified weighted Cox analysis on the joint data set of all the constructed trials, where each trial is one stratum. The model is applied to data from the Swiss HIV cohort study. Copyright © 2010 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Collections of meta-analyses assembled in meta-epidemiological studies are used to study associations of trial characteristics with intervention effect estimates. However, methods and findings are not consistent across studies. To combine data from 10 meta-epidemiological studies into a single database, and derive a harmonized dataset without overlap between meta-analyses. The database design allowed trials to be contained in different meta-analyses, multiple meta-analyses in systematic reviews, overlapping meta-analyses between systematic reviews, and multiple references to the same trial or review. Unique identifiers were assigned to each reference and used to identify duplicate trials. Sets of meta-analyses with overlapping trials were identified and duplicates removed. Overlapping trials were used to examine agreement between assessments of trial characteristics. The combined database contained 427 reviews, 454 meta-analyses and 4874 trial results. Of these, 258 meta-analyses were unique, while for 196 at least one trial overlapped with another meta-analysis. Median kappa statistics for reliability of assessments were 0.60 for sequence generation, 0.58 for allocation concealment and 0.87 for blinding. Based on inspection of sets of overlapping meta-analyses, 91 meta-analyses containing 1344 trial results were removed. Additionally, 24 duplicated trial results were removed from 16 meta-analyses, to derive a final database containing 363 meta-analyses and 3477 unique trial results. The final database will be used to examine the combined evidence on sources of bias in randomized controlled trials. The strategy used to remove overlap between meta-analyses may be of use for future empirical research. Copyright © 2010 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Observational epidemiological studies suffer from many potential biases, from confounding and from reverse causation, and this limits their ability to robustly identify causal associations. Several high-profile situations exist in which randomized controlled trials of precisely the same intervention that has been examined in observational studies have produced markedly different findings. In other observational sciences, the use of instrumental variable (IV) approaches has been one approach to strengthening causal inferences in non-experimental situations. The use of germline genetic variants that proxy for environmentally modifiable exposures as instruments for these exposures is one form of IV analysis that can be implemented within observational epidemiological studies. The method has been referred to as 'Mendelian randomization', and can be considered as analogous to randomized controlled trials. This paper outlines Mendelian randomization, draws parallels with IV methods, provides examples of implementation of the approach and discusses limitations of the approach and some methods for dealing with these. Copyright © 2007 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Publication bias and related bias in meta-analysis is often examined by visually checking for asymmetry in funnel plots of treatment effect against its standard error. Formal statistical tests of funnel plot asymmetry have been proposed, but when applied to binary outcome data these can give false-positive rates that are higher than the nominal level in some situations (large treatment effects, or few events per trial, or all trials of similar sizes). We develop a modified linear regression test for funnel plot asymmetry based on the efficient score and its variance, Fisher's information. The performance of this test is compared to the other proposed tests in simulation analyses based on the characteristics of published controlled trials. When there is little or no between-trial heterogeneity, this modified test has a false-positive rate close to the nominal level while maintaining similar power to the original linear regression test ('Egger' test). When the degree of between-trial heterogeneity is large, none of the tests that have been proposed has uniformly good properties. Copyright © 2005 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

The process of developing and validating a prognostic model for survival time data has been much discussed in the literature. Assessment of the performance of candidate prognostic models on data other than that used to fit the models is essential for choosing a model that will generalize well to independent data. However, there remain difficulties in current methods of measuring the accuracy of predictions of prognostic models for censored survival time data. In this paper, flexible parametric models based on the Weibull, loglogistic and lognormal distributions with spline smoothing of the baseline log cumulative hazard function are used to fit a set of candidate prognostic models across k data sets. The model that generalizes best to new data is chosen using a cross-validation scheme which fits the model on k-1 data sets and tests the predictive accuracy on the omitted data set. The procedure is repeated, omitting each data set in turn. The quality of the predictions is measured using three different methods: two commonly proposed validation methods, Harrell's concordance statistic and the Brier statistic, and a novel method using deviance differences. The results show that the deviance statistic is able to discriminate between quite similar models and can be used to choose a prognostic model that generalizes well to new data. The methods are illustrated by using a model developed to predict progression to a new AIDS event or death in HIV-1 positive patients starting antiretroviral therapy. [ABSTRACT FROM AUTHOR]

The article cites a study on a systematic review design which aims to determine the accuracy of magnetic resonance imaging (MRI) criteria for the early diagnosis of multiple sclerosis in patient suspected with the disease. The review found out that 29 studies produced higher estimated diagnostic odds ratio than did cohort studies, and 15 studies produced higher estimates of specificity and lower estimates of sensitivity. The results are believed to have implications for the use of MRI.

Objectives: To examine the construct validity of the RSI in a contemporary cohort., Design: A cross-sectional investigation of 1,717 employed individuals who responded to the second stage of a study of occupational stress., Methods: Scores on the RSI are compared to smoking and drinking habits, social class, and two measures of health (number of days sick leave, and number of visits to a GP during the previous year). The RSI was compared to three questionnaires measuring concepts related to stress: the 12-item General Health Questionnaire (GHQ-12), the Hospital Anxiety and Depression Scale (HADS), and the Karasek Job Strain Questionnaire., Results: Higher levels of stress, as measured by the RSI, were associated with smoking a greater number of cigarettes, and, if the respondent drank alcohol, greater consumption of alcoholic drinks. High levels of stress were also associated with having taken more days sick leave, having made more frequent visits to a GP, and, somewhat unexpectedly, with being a teetotaller, and with holding a non-manual occupation. Of the questionnaire measures, the strongest association was between the RSI and HADS anxiety subscale, consistent with an overlap between the concepts of stress and anxiety., Conclusions: This study supports the construct validity of the RSI in a sample of employed individuals.