The full content of Annals is available to subscribers

Subscribe/Learn More  >
Reviews |

Meta-analysis: Diagnostic Performance of Low-Radiation-Dose Coronary Computed Tomography Angiography

Moritz Wyler von Ballmoos, MD, PhD, MPH; Bernhard Haring, MD; Pascal Juillerat, MD, MSc; and Hatem Alkadhi, MD, MPH
[+] Article, Author, and Disclosure Information

From Harvard School of Public Health, Children's Hospital Boston, Harvard Medical School, Beth Israel Deaconess Hospital, and Massachusetts General Hospital, Boston, Massachusetts, and University Hospital Zurich, Zurich, Switzerland.

Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M10-2039.

Requests for Single Reprints: Moritz Wyler von Ballmoos, MD, PhD, MPH, Department of Cardiac Surgery, Children's Hospital Boston, Harvard Medical School, 300 Longwood Avenue, Boston, MA 02115; e-mail, wylervonballmoos@gmail.com.

Current Author Addresses: Dr. von Ballmoos: Department of Cardiac Surgery, Children's Hospital Boston, Harvard Medical School, 300 Longwood Avenue, Boston, MA 02115.

Dr. Haring: Beth Israel Deaconess Medical Center, Cardiovascular Institute, Harvard Medical School, 3 Blackfan Circle, Boston, MA 02215.

Dr. Juillerat: Crohn's and Colitis Center, Harvard Medical School, 165 Cambridge Street, 9th Floor, Boston, MA 02114.

Dr. Alkadhi: Institute of Diagnostic and Interventional Radiology, University Hospital Zurich, Raemistr. 100, CH-8091 Zurich, Switzerland.

Author Contributions: Conception and design: M.W. von Ballmoos, B. Haring, H. Alkadhi.

Analysis and interpretation of the data: M.W. von Ballmoos, B. Haring, P. Juillerat, H. Alkadhi.

Drafting of the article: M.W. von Ballmoos, B. Haring, H. Alkadhi.

Critical revision of the article for important intellectual content: M.W. von Ballmoos, B. Haring, P. Juillerat, H. Alkadhi.

Final approval of the article: M.W. von Ballmoos, B. Haring, P. Juillerat, H. Alkadhi.

Statistical expertise: M.W. von Ballmoos, H. Alkadhi.

Obtaining of funding: P. Juillerat.

Administrative, technical, or logistic support: M.W. von Ballmoos, B. Haring, H. Alkadhi.

Collection and assembly of data: M.W. von Ballmoos, B. Haring, P. Juillerat, H. Alkadhi.

Ann Intern Med. 2011;154(6):413-420. doi:10.7326/0003-4819-154-6-201103150-00007
Text Size: A A A

This article has been corrected. For original version, click "Original Version (PDF)" in column 2.

Background: A new radiation dose–saving technique for noninvasive coronary artery imaging with computed tomography (CT) is available.

Purpose: To summarize current evidence about the ability of low-dose coronary CT angiography to rule out coronary artery disease (CAD) in symptomatic adults.

Data Sources: Online databases, including MEDLINE, EMBASE, and the Cochrane Library, from inception through 31 October 2010; abstract databases; gray literature; reference lists of identified articles; and experts. No language restrictions were applied.

Study Selection: All investigators screened and selected studies that compared prospective electrocardiography-gated coronary CT angiography with catheter coronary angiography (the reference standard) in symptomatic patients with suspected CAD.

Data Extraction: Two investigators independently extracted patient and study protocol characteristics and rated methodological quality; differences were resolved by consensus or by a third reader. Multivariate random-effects models were used to obtain pooled estimates.

Data Synthesis: 16 studies, comprising 960 patients, were found (7 studies of single-source, 64-slice CT; 4 of dual-source, 64-slice CT; 2 of single-source, 320-slice CT; 1 dual-source, 128-slice CT; 1 of single-source, 128-slice CT; and 1 of single-source, 256-slice CT). On average, 2.4% of the coronary arterial segments were of nondiagnostic image quality, and 1 or more segments were nondiagnostic in 9.5% of the patients. The patient-level sensitivity and specificity of CT angiography were 1.00 (95% CI, 0.98 to 1.00) and 0.89 (CI, 0.85 to 0.92), respectively. The pooled vessel- and segment-level estimates showed lower sensitivity and higher specificity than the patient-level estimates. Statistically significant heterogeneity was found between studies for vessel- and segment-level estimates, which seemed to be associated with body mass index and prevalence of CAD but not with CT scanner characteristics.

Limitations: The small number of studies, half of which were from a single tertiary center, limits generalizability. The potential harms of the imaging tests were not well-evaluated.

Conclusion: Early evidence suggests that low-dose coronary CT angiography matches the sensitivity of catheter-based angiography, has low radiation exposure, and is a potentially valid alternative to catheter angiography for triaging symptomatic patients with a clinical suspicion of CAD.

Primary Funding Source: None.


Grahic Jump Location
Appendix Figure 1.
Summary of evidence search and selection.
Grahic Jump Location
Grahic Jump Location
Figure 1.
Paired forest plots of the sensitivity and specificity of low-dose coronary CT angiography at the segment, patient, and vessel levels for the diagnosis of CAD, compared with catheter angiography.

In this analysis, coronary arterial segments with nondiagnostic image quality are coded as positive for disease. CAD = coronary artery disease; CT = computed tomography.

Grahic Jump Location
Grahic Jump Location
Appendix Figure 2.
Hierarchical SROC curves.

Pooled estimates of the diagnostic performance of low-dose coronary computed tomography angiography obtained from the bivariate random-effects meta-analysis. AUC = area under the curve; SROC = summary receiver-operating characteristic.

Grahic Jump Location
Grahic Jump Location
Figure 2.
Bayesian plots of pre- and posttest probability.

Posttest probability after negative or positive findings on low-dose coronary CT angiography at the patient level for populations with different prevalences (pretest probabilities) of coronary artery disease. CT = computed tomography.

Grahic Jump Location




Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).


Submit a Comment/Letter
Interpreting the Results of Systematic Reviews of Diagnostic Test Accuracy
Posted on March 27, 2011
Ali Salavati MD, MPH
Zentralklinik Bad Berka, Germany
Conflict of Interest: None Declared

To the Editor:

We read with great enthusiasm the comprehensive systematic review on the diagnostic performance of low-radiation-dose coronary computed tomography angiography (1). There are a few concerns about the statistical methods.

Von Ballmoos et al evaluated between-study heterogeneity by calculating the Cochran Q and I2 statistics. These univariate statistics were mainly developed for meta-analysis of interventional studies (2) and calculate heterogeneity for sensitivity and specificity separately when used in systematic reviews of diagnostic test accuracy (SRDTA). The I2 statistic is commonly applied with univariate models like Moses-Littenberg method which both are recommended to be replaced by more robust hierarchical models (3).

The more appropriate methods for investigating heterogeneity in SRDTA are the bivariate or HSROC models which were used to obtain the summary estimates of sensitivity and specificity in the study. However, appropriate methods for calculation of the Cochran Q and I2 statistics for the hierarchical model are not yet developed; the variance parameters of the random-effect in these models give an indication of the degree of heterogeneity in sensitivity and specificity if using the bivariate model, and in accuracy and threshold if the HSROC model is used (3).

The authors investigated the publication bias using the Egger method. Tests for funnel plot asymmetry like Egger, Begg, Peters and Harbord test are primarily designed for use in randomized trials (3, 4) and should not be used with diagnostic studies (3), since the accuracy of these tests deteriorates as the odds ratio moves away from 1 and the expected odds ratio for diagnostic studies is commonly too large (4) (237.3 and 444.3 for vessel and segment levels, respectively in the current study). Even with the development of more accurate methods such as Deeks test, the assessment of publication bias in SRDTA is still controversial (3).

Although the sensitivity analysis was performed on an interesting hypothesis, subgroup analysis is not a particularly convincing method since removing a small number of studies is unlikely to have a major effect on the summary estimates. It would be more appropriate to incorporate a covariate in the bivariate model to test for possible differences between subgroups (3).

The promising diagnostic performance of low-dose coronary CT angiography ,however, could end the controversy surrounding the radiation dose and application of myocardial perfusion imaging or coronary CT angiography (5), we suggest interpreting the results of systematic reviews of diagnostic test accuracy more cautiously, considering the lack of a uniform methodology for SRDTA and while many relevant methods are yet to be developed. This issue is of increased importance when applying tests that may be considered more appropriate for systematic review of interventional studies.


1.von Ballmoos MW, Haring B, Juillerat P, Alkadhi H. Meta-analysis: Diagnostic Performance of Low-Radiation-Dose Coronary Computed Tomography Angiography. Ann Intern Med. 2011;154(6):413-20.

2.Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557-60.

3.Macaskill P, Gatsonis C, Deeks JJ, Harbord RM, Takwoingi Y. Chapter 10: Analysing and Presenting Results. In: Deeks JJ, Bossuyt PM, Gatsonis C, eds. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy Version 1.0. : The Cochrane Collaboration, 2010. Available from: http://srdta.cochrane.org/.

4.Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol. 2005;58(9):882- 93.

5.Salavati A, Radmanesh F, Esfahani SA. Diagnostic accuracy and clinical utility of noninvasive testing for coronary artery disease. Ann Intern Med. 2011;154(4):290: author reply

Conflict of Interest:

None declared

In Response
Posted on March 30, 2011
Hatem Alkadhi
University Hospital Zurich
Conflict of Interest: None Declared

We thank Salavati and Radmanesh for their interest in our study. In their comment they touch on several important points of statistical analysis in research synthesis of diagnostic test studies. Frequently, data obtained from diagnostic test studies with a binary outcome tend to be overdispersed. This requires appropriate analytic methodology for calculation of unbiased estimates and valid test statistics. We fitted non -linear mixed models to calculate parameter estimates by maximizing an approximation to the likelihood integrated over the random effects, and used the second derivative matrix of the likelihood function to compute approximate standard errors as described previously (1).

Eventually, these models returns estimates for the within as well as the between study variance allowing for calculation of the I2 statistic as presented in our paper. To our knowledge there is currently no simulation or other validation study available for the use of the I2 statistics in meta-analyses of diagnostic accuracy studies specifically. Yet, the theoretical framework underlying this statistic of heterogenity is not specific to a certain data type or modeling technique per se, and should remain its validity as long as variance estimates used to calculate it are unbiased (2).

Further, we agree with Salavati and Radmanesh that currently accepted methods to evaluate publication bias are insufficient to address the specific needs of meta-analysis evaluating diagnostic test studies and thus remain controversial. Finally, it is a known fact that stratification of analysis by restriction rather than modeling may decrease power and hence result in higher P-values and larger confidence intervals around point estimates. As described in the methods, we included study location as a random effect in our models, but in addition also used restriction as part of our sensitivity analysis for study location.


1. Harbord RM, Deeks JJ, Egger M, Whiting P, Sterne JA. A unification of models for meta-analysis of diagnostic accuracy studies. Biostatistics. 2007;8(2):239-51.

2. Higgins, J. P. T., and S. G. Thompson. 2002. Quantifying heterogeneity in a meta- analysis. Stat Med 21(11): 1539?1558.

Conflict of Interest:

None declared

Submit a Comment/Letter

Summary for Patients

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.


Buy Now for $32.00

to gain full access to the content and tools.

Want to Subscribe?

Learn more about subscription options

Related Articles
Topic Collections
PubMed Articles
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.