STATISTICAL METHOD
Diagnostic Test Accuracy Meta-Analysis
Statistical Consulting for Pooled Test Accuracy
We model sensitivity and specificity jointly across the studies you have identified, using a bivariate or HSROC model, and report pooled estimates with confidence and prediction regions, a summary ROC curve and the QUADAS-2 assessment of study quality.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
METHOD OVERVIEW
What Is Test Accuracy Meta-Analysis?
A diagnostic test accuracy (DTA) meta-analysis combines the sensitivity and specificity that several studies report for one index test, judged against a reference standard. Each study supplies a two-by-two table of true positives, false positives, false negatives and true negatives. Because sensitivity and specificity trade off as the positivity threshold changes, they are modeled together, usually with a bivariate random-effects model on the logit scale or the equivalent HSROC model, giving pooled estimates, a summary ROC curve and the between-study variation. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides this analysis within your review design, with the relevant checks, estimates, uncertainty measures, interpretation and publication-ready reporting.
WHEN TO ENGAGE
When Is Accuracy Meta-Analysis Appropriate?
This analysis fits a set of studies that evaluated one index test against a reference standard, and these are the situations in which researchers most often ask us to plan, run or confirm it.
Several Studies of One Index Test Studies found through a Systematic Review each report a two-by-two table against a reference standard, and summary sensitivity and specificity are needed.
Different Positivity Thresholds Studies applied different cutoffs to the same test, so the threshold effect has to be modeled in the synthesis, not averaged away across studies.
Prespecified Accuracy Question The index test, target condition and reference standard were fixed in a protocol, and the model and reporting must be settled before the pooled results are seen.
Reviewing a Completed DTA Analysis Sensitivity and specificity were pooled and a reviewer asked about their correlation. Independent Statistical Review re-examines the model and the extracted cells.
ANALYTICAL APPROACH
How We Apply Test Accuracy Pooling
Within the family of Evidence Synthesis & Meta-Research, accuracy is pooled only after the studies, thresholds and reference standards have been examined.
Cell Extraction and Quality We verify each two-by-two table against its paper, because swapped cells silently corrupt every estimate, and summarize study quality with QUADAS-2 for bias and applicability.
Joint Model of Accuracy Sensitivity and specificity are modeled together with a bivariate random-effects model on the logit scale or an HSROC model. Pooling them separately ignores their correlation.
Heterogeneity and Thresholds Variation between studies is expected, so we model it through variance terms and a prediction region, and explore threshold and population covariates when studies allow.
Funnel Plots and Small Studies Deeks' funnel plot test is preferred for accuracy studies, but its power is low, so a non-significant result is not read as evidence that small-study effects are absent.
DELIVERABLES
What You Receive
As part of our Meta-Analysis & Evidence Synthesis service, the accuracy synthesis is delivered with the estimates, checks and files needed to report and defend it.
Data & Variable Review Two-by-two cells, thresholds and reference standards are checked against each paper, and missing or inconsistent cells are resolved before any model is fitted.
Method-Specific Checks Model convergence, the correlation of sensitivity and specificity and the influence of single studies are examined, with a QUADAS-2 restricted analysis as a sensitivity check.
Statistical Results Pooled sensitivity and specificity with 95% confidence intervals, confidence and prediction regions, HSROC parameters, and likelihood ratios derived from the fitted model.
Publication-Ready Tables & Figures A summary ROC plot with the pooled point and regions, paired forest plots and a QUADAS-2 summary, formatted to PRISMA-DTA or the journal's requirements.
Interpretation Results are written as accuracy at a stated threshold and setting, with the certainty of evidence in the GRADE framework, and without implying a better patient outcome.
Reproducible Analysis Files R scripts, Stata do-files or SAS programs are provided with the coded two-by-two data and threshold notes, so each summary point and region can be refitted.
ALTERNATIVE METHODS
When Another Method May Be More Appropriate
The method follows the accuracy question and the data available. One study, continuous test values, unusable two-by-two tables or an intervention question each call for another approach.
01
Sensitivity & Specificity (Diagnostic Accuracy)
For one accuracy study, not a synthesis, sensitivity and specificity are estimated directly from a single two-by-two table with confidence intervals.
02
ROC Curve Analysis
When the data are continuous test values from one study, the ROC curve, its AUC and the choice of cutoff are analyzed directly, without a synthesis.
03
Meta-Analysis
For a single effect measure such as a risk ratio from intervention studies, standard pooling applies, and a joint model of accuracy is not needed.
04
SWiM (Synthesis Without Meta-Analysis)
When most studies report only sensitivity or only specificity, or no usable two-by-two table, accuracy is summarized in prespecified groups, without pooling.
CASE EXAMPLE
How We Work in Practice
An illustration of how a pooled accuracy analysis moves from a written plan to a reportable, defensible summary of test performance.
ILLUSTRATIVE EXAMPLE · DIAGNOSTIC DEVICE EVIDENCE REVIEW · UNITED STATES
An evidence group at a diagnostic device company needed pooled accuracy of a high-sensitivity troponin assay in suspected myocardial infarction, for a clinical evidence report.
The protocol, written before extraction, named a bivariate random-effects model on the logit scale and a prespecified analysis at the most common cutoff. Duplicate extraction showed that two studies had false positive and false negative counts transposed, which would have distorted both their sensitivity and their specificity. We corrected the cells and queried one author. Because studies used different cutoffs, the main model pooled only those at the prespecified cutoff, and an HSROC analysis of all studies served as a sensitivity analysis. Between-study variation was reported through a prediction region, with QUADAS-2 judgments beside the estimates.
Outcome: The group received pooled sensitivity and specificity with their 95% confidence intervals and a prediction region, the summary ROC curve, a QUADAS-2 summary and the analysis scripts.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
SOFTWARE
Statistical Software for Accuracy Synthesis
Bivariate and HSROC models are fitted in R, Stata and SAS, with Python used for custom or Bayesian implementations, and scripts provided where applicable.
RESEARCH DISCIPLINES
Test Accuracy Evidence Across Research Fields
Accuracy synthesis is used wherever a test classifies people against a reference standard, each field with its own conditions, thresholds and reporting conventions.
A Defined Process for Your Analysis
We review your dataset, research question and design before any work begins, so the scope, deliverables, timeline and fixed project fee are clear from the outset.
01
Share Your Research Materials
Send your dataset, research question, study design and any existing analysis plan or output. We will let you know if anything else is needed to assess the project.
02
Receive Your Assessment
Your statistically trained Research Coordinator reviews the project and confirms the required analysis, deliverables, timeline and fixed project price before work begins.
03
Analysis & Delivery
Once you confirm the scope, an in-house statistician completes the agreed analysis and delivers the outputs, interpretation and reporting files.
RESEARCHER VOICE
What Researchers Say
Feedback from a clinical researcher who needed to judge whether studies were comparable enough to pool and how heterogeneity should be handled.
“Our dataset looked straightforward until we got into Diagnostic Test Accuracy Meta-Analysis and realized whether the studies were comparable enough to pool and how heterogeneity should be handled. The team sorted out the analytical setup and gave us a clear explanation of what could and could not be concluded. The final output was concise, clearly interpreted, and ready to use in our research reporting.”
RESEARCH RESPONSIBILITY
Statistical Support Without Taking Over Your Research
We provide the statistical work for your accuracy synthesis, and the review, its meaning and its authorship stay entirely with you.
Agreed Statistical Support We provide the agreed methodological, analytical and statistical reporting support for your accuracy synthesis, within the scope confirmed before work begins.
No Intellectual Contribution OnlineSPSS claims no intellectual contribution to your research. Our work is statistical support, and authorship remains yours.
Your Research Decisions The target condition, eligibility criteria and clinical conclusions remain with you, and our statistical interpretation stays within what the pooled estimates and the studies support.
Confidential Handling Research data and project details are handled confidentially and are not published. An NDA is available where required, and can be arranged before sensitive materials are shared.
FREQUENTLY ASKED QUESTIONS
Questions About Test Accuracy Meta-Analysis
Short answers to what researchers ask before pooling accuracy studies, from joint models and thresholds to confidentiality and the assessment process.
Why not pool sensitivity and specificity separately?
Sensitivity and specificity are correlated across studies through the positivity threshold. A stricter cutoff raises specificity and lowers sensitivity, so two separate univariate pools ignore that link and can give misleading summary points and intervals. A bivariate random-effects model keeps the correlation, and we report it so readers can see the trade-off in the data.
How do I handle studies that use different thresholds?
Different thresholds create genuine heterogeneity, and we model it instead of averaging it away. Options include restricting the main analysis to a prespecified cutoff, using an HSROC model that estimates a threshold parameter, or analyzing cutoff groups separately. The choice is stated in the protocol, and the other options are shown as sensitivity analyses.
Should I use the bivariate model or the HSROC model?
Without covariates the two are equivalent parameterizations of one model, so the fitted curve is the same. The bivariate form is convenient for a summary point with its confidence and prediction regions, and the HSROC form for the curve and a threshold parameter. With covariates their interpretation differs, so the choice follows the review question.
Can you review an analysis that has already been completed?
Yes, we can review a completed accuracy meta-analysis before submission or a response to reviewers. We check the extracted two-by-two cells against the source papers, the model, the handling of thresholds and the correlation, the QUADAS-2 judgments and the reporting, and confirm what stands or needs to change.
How do you protect my dataset and research confidentiality?
Your extraction sheets, output files and any protocol or analysis plan are handled confidentially, and project information is not published. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Need Support With
Accuracy Meta-Analysis?
Share your review question, the extracted two-by-two data and your protocol. We assess whether the studies can be pooled and confirm the approach before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014