Beta-binomial analysis of variance model for network meta-analysis of diagnostic test accuracy data

An overview of systematic reviews on imaging tests for diagnosis of pulmonary embolism applying different network meta-analytic methods

PURPOSE

This study aimed to apply different methods of diagnostic test accuracy network meta-analysis (DTA-NMA) for studies reporting results of five imaging tests for the diagnosis of suspected pulmonary embolism (PE): pulmonary angiography (PA), computed tomography angiography (CTPA), magnetic resonance angiography (MRA), planar ventilation/perfusion (V/Q) scintigraphy and single-photon emission computed tomography ventilation/perfusion (SPECT V/Q).

METHODS

We searched four databases (MEDLINE [via PubMed], Cochrane CENTRAL, Scopus, and Epistemonikos) from inception until June 2, 2022 to identify systematic reviews (SRs) describing diagnostic accuracy of PA, CTPA, MRA, V/Q scan and SPECT V/Q for suspected PE. Study-level data were extracted and pooled using a hierarchical summary receiver operating characteristic (HSROC) meta-regression approach and two DTA-NMA models to compare accuracy estimates of different imaging tests. Risk of bias was assessed using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies-2) tool and certainty of evidence using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework.

RESULTS

We identified 13 SRs, synthesizing data from 33 primary studies and for four imaging tests (PA, CTPA, MRA and V/Q scan). The HSROC meta-regression model using PA as the reference standard showed that MRA had the best overall diagnostic performance with sensitivity of 0.93 (95% confidence interval [CI]: 0.76, 1.00) and specificity of 0.94 (95% CI: 0.84, 0.99). However, DTA-NMA models indicated that V/Q scan had the highest sensitivity, while CTPA was most specific.

CONCLUSION

Selecting a different DTA-NMA method to assess multiple diagnostic tests can affect estimates of diagnostic accuracy. There is no established method, but the choice depends on the data and familiarity with Bayesian statistics.

Transdiagnostic cognitive biases in psychiatric disorders: A systematic review and network meta-analysis

Psychiatric disorders are characterized by cognitive deficits, which have been proposed as a transdiagnostic feature of psychopathology ("C" factor). Similarly, cognitive biases (e.g., in attention, memory, and interpretation) represent common tendencies in information processing that are often associated with psychiatric symptoms. However, the question remains whether cognitive biases are also transdiagnostic or are specific to certain psychiatric disorders/symptoms. The current systematic review sought to address whether the proposed "C" factor of transdiagnostic cognitive dysfunction in psychopathology can be extended to cognitive biases. Overall, 31 studies comprising 4401 participants (2536 patients, 1865 non-clinical controls) met inclusion criteria, assessing 19 cognitive biases across 20 diagnostic categories, with most studies focusing on interpretation (k = 22) and attention (k = 11) biases and only 2 assessing memory biases. Traditional meta-analyses found a moderate effect size (g = 0.32) for more severe cognitive biases in all patients relative to non-clinical controls, as well as small but significant associations between interpretation biases and transdiagnostic symptom categories (general psychopathology: r = 0.20, emotion dysfunction: r = 0.17, psychotic symptoms: r = 0.25). Network meta-analyses revealed significant patient versus non-clinical control differences on attention and interpretation biases across diagnoses, as well as significant differences between diagnoses, with highest severity in panic disorder for attention biases and obsessive-compulsive disorder for interpretation biases. The current findings extend the big "C" interpretation of transdiagnostic cognitive dysfunction in psychiatric disorders to cognitive biases and transdiagnostic symptom dimensions. Results also suggest that while the presence of cognitive biases is transdiagnostic, bias severity differs across diagnoses, as in traditional neurocognitive deficits.

Complement levels during the first trimester predict disease flare and adverse pregnancy outcomes in systemic lupus erythematosus: A network meta-analysis on 532 pregnancies

BACKGROUND

Complement levels have been proposed as candidate biomarkers of disease activity and obstetric risk in systemic lupus erythematosus (SLE) pregnancies, but their reliability has been questioned due to the physiologic fluctuations of complement during gestation. Thus, this network meta-analysis aimed at assessing the clinical significance of complement fluctuations in lupus pregnant women.

METHODS

Corresponding authors of 19 studies meeting inclusion criteria were invited to contribute with additional data including C3 and C4 levels [before pregnancy, at conception, in every trimester (T) and 3 months after delivery]; data were pooled together in a network meta-analysis.

RESULTS

A total of 532 lupus women from four studies were included in the analysis. In SLE women, C3 and C4 increased progressively during gestation: levels remained stable during T1 and peaked in T2 to decrease in T3. Patients with previous lupus nephritis (LN) and those who experienced flares during pregnancy had significantly lower mean levels of C3 and C4 at all timepoints. The lowest levels of complement were observed, particularly during T1, in patients with LN and gestational flare. Both reduction and the lack of increase of C3 and C4 levels at T1 versus conception were associated with gestational flares, particularly in LN patients. Pregnancies with flare had a statistically significant higher rate of maternal and fetal complications(60% versus 50.3%; p = 0.03).

CONCLUSIONS

Low complement levels, particularly in T1, were associated with a higher frequency of gestational flare. Either reduction or smaller increase of C3 and/or C4 levels, even within normal range, might predict flares especially in early gestation.

Diagnostic accuracy of screening algorithms to identify persons with active pulmonary tuberculosis at prison entry: protocol of a systematic review and network meta-analysis

Prison inmates are a high-risk group for tuberculosis (TB) infection and disease due to the increasing number of vulnerable fringe groups, risk factors (e.g., alcohol and drug addictions), contagious diseases (HIV, hepatitis), and their high-risk behavior. Compared to the general population, TB incidence and prevalence rates are significantly higher among prison inmates. Early identification of potentially infectious pulmonary TB (PTB) and targeted care of sick inmates are essential to effectively control TB within the prison system. The WHO recommends combining active and passive case-finding in prisons. No study has been published comparing the broad spectrum of screening tools using a diagnostic accuracy network meta-analysis (NMA). We aim to identify the most accurate TB case-finding algorithm at prison entry that is feasible in resource-limited prisons of high-burden TB countries and ensures continuous comprehensive TB detection services in such settings. Evidence generated by this NMA can provide important decision support in selecting the most (cost-) effective algorithms for screening methods for resource-limited settings in the short, medium, and long terms.

VALCOR: a protocol for the validation of SARS-corona virus-2 assays

BACKGROUND

Testing for SARS-CoV-2, together with vaccination, is one of the most vital strategies in curbing the current COVID-19 pandemic. The pandemic has led to an unprecedented need for diagnostic testing and the rapid emergence of an abundance of commercial assays on the market. Due to the nature of the pandemic and in the interest of health protection, many of these assays received provisional authorisation for emergency use without thorough validation. To limit false negative and false positive results, it is key to define common criteria that SARS-CoV-2 assays need to fulfil. VALCOR or "VALidation of SARS-CORona Virus-2 assays" is a protocol designed to set up a framework for test validation of SARS-CoV-2 virus assays.

OBJECTIVES

VALCOR is a study protocol for the validation of assays used for confirmation of the presence of SARS-CoV-2 in patients with COVID-19 disease or the screening of carriers of SARS-CoV-2 virus by the identification of viral RNA in oropharyngeal and/or nasopharyngeal specimens or other specimens from the human respiratory tract.

METHODS

The VALCOR panel of samples will contain clinical human specimens and standardised artificial specimens. The collection of clinical specimens will include nasopharyngeal or oropharyngeal specimens or other specimens from the respiratory tract obtained from COVID-19 patients and healthy carriers of SARS-CoV-2 as well as specimens from subjects not carrying SARS-CoV-2. Artificial specimens include calibrated amounts of viral RNA of SARS-CoV-2 sequences provided by established competent agencies that produce reference materials for the assessment of the limit of detection of each assay. The panel of samples are sent from a central reference laboratory (having access to biobanks of clinical specimens tested already for SARS-CoV-2 with a reference comparator assay) to participating laboratories for testing with a SARS-CoV-2 index assay that requires evaluation.

DISCUSSION

VALCOR provides a harmonised and standard framework to benchmark the testing performance of SARS-CoV-2 assays that are rapidly evolving. As the pandemic incited an urgent need for testing capacity, there is a gap in the comprehensive validation of SARS-CoV-2 assays. This study will generate comprehensive validation data for assays used for the diagnosis of SARS-CoV-2 and may serve as a basis for other validation protocols.

Diagnostic test accuracy network meta-analysis methods: A scoping review and empirical assessment

OBJECTIVE

To a) identify methodological and application papers reporting a model developed specifically for diagnostic test accuracy network meta-analysis (DTA-NMA) or a hierarchical meta-regression method for comparing at least three index tests; b) review and summarize the characteristics of the methods and the application papers; and c) compare DTA-NMA and hierarchical meta-regression methods empirically.

STUDY DESIGN AND SETTINGS

We performed a scoping review and searched major databases until 3^rd March 2021. We assessed the characteristics of the identified methods, conducted a descriptive analysis of characteristics of the application articles, and applied the DTA-NMA and meta-regression methods to the available data.

RESULTS

We included 49 articles, of which 9 were methodological (describing 11 DTA-NMA methods) and 40 were application papers (data available for 32 DTA-NMAs). Our results showed a steep increase in recent years in DTA-NMA publications. DTA-NMA models may lead to different results. Although sensitivity estimates were comparable between meta-regression and DTA-NMA models, specificity estimates were higher in meta-regression.

CONCLUSIONS

The choice of a DTA-NMA model will depend on the available data, including the use of different thresholds for test positivity, different study designs, and software familiarity. Selection between the methods may impact on the NMA results, especially for specificity.

Challenges in Comparative Meta-Analysis of the Accuracy of Multiple Diagnostic Tests

The rapid increase in diagnostic and screening techniques has urged the need to choose among multiple diagnostic tests. For the majority of diseases, there is more than a single test available, and studies usually compare a subset of these tests. In such cases, a separate meta-analysis of each test cannot provide a reliable answer on the relative accuracy of the multiple available tests. Extensions of standard (hierarchical) meta-analysis to network meta-analysis (NMA) models for the comparison of at least three diagnostic tests have been the subject of methodological research in recent years. NMA can be used to jointly analyze the totality of evidence in order to provide estimates of relative accuracy (sensitivity and specificity ), to compare tests that have not been compared head-to-head, and to obtain a ranking of all competing tests in order to further facilitate the decision-making process.In this chapter, we illustrate current methodology for meta-analysis of multiple test comparisons, introduce NMA methods of diagnostic tests as an extension to the standard meta-analysis of diagnostic test accuracy (DTA) studies, and present existing approaches to rank tests according to their accuracy, specificity , and sensitivity . We also describe the basic concepts, underlying assumptions, and challenges in NMA of multiple diagnostic tests.

Meta-Analysis Methods of Diagnostic Test Accuracy Studies

Meta-analytic techniques are used to combine the results of different studies that have evaluated the accuracy of diagnostic tests. In this article, we present univariate and multivariate meta-analysis methods for a single test and we provide an extensive description of methods for meta-analysis and comparison of multiple diagnostic tests. We close with a practical example of a meta-analysis that aimed to determine whether Rheumatoid Factor identifies patients with Rheumatoid Arthritis.

Statistical methods for detecting outlying and influential studies in meta-analysis of diagnostic test accuracy studies

Bivariate random-effects models are currently widely used to synthesize pairs of test sensitivity and specificity across studies. Inferences drawn based on these models may be distorted in the presence of outlying or influential studies. Currently, subjective methods such as inspection of forest plots are used to identify outlying studies in meta-analysis of diagnostic test accuracy studies. We proposed objective methods based on solid statistical reasoning for identifying outlying and/or influential studies. The proposed methods have been validated using simulation study and illustrated on two published meta-analysis data. Our methods outperform and neglect the subjectivity of the currently used ad hoc methods. The proposed methods can be used as a sensitivity analysis tool concurrently with the current bivariate random-effects models or as a preliminary analysis tool for robust models that accommodate outlying and/or influential studies in meta-analysis of diagnostic test accuracy studies.