Keywords:
Epidemiology, Diagnostic procedure, CT, Cardiac
Authors:
G. J. Pelgrim1, M. D. Dorrius1, X.-Q. Xie2, M. A. den Dekker1, M. Oudkerk1, R. Vliegenthart1; 1Groningen/NL, 2Shanghai/CN
Methods and Materials
Pubmed,
Embase and Web of Knowledge were searched from January 1995 until March 2014.
The search query used in Pubmed was: (“Tomography,
X-Ray Computed”[MeSH] OR “Computed Tomography”[TIAB] OR CT [TIAB]) AND ( myocardial OR cardiac OR “coronary artery”) AND (“perfusion imaging”[MeSH] OR dynamic OR stress perfusion) AND 1995/01:2014/05 [dp].
Bibliographies of the selected articles were searched for potentially relevant articles.
Selection based on title and abstract was performed by one reviewer.
Thereafter,
the inclusion for the meta-analysis based on fulltext articles was performed by two reviewers.
In case of a disagreement,
a third reviewer arbitered.
Studies were included when: 1) it assessed CT perfusion imaging as a diagnostic test to evaluate patients for the presence of CAD; 2) it assessed perfusion using a static or dynamic imaging protocol; 3) perfusion scanning was performed in stable CAD patients; 4) addressed diagnostic accuracy in comparison to other perfusion techniques or morphological imaging.
Studies were rejected when: 1) it concerned a review,
protocol,
letter or case report; 2) scanners < 16 MDCT were used; 3) it were laboratory,
phantom or animal studies.
Referral of patients with either suspected or known CAD was both eligible and therefore not rejected.
Multiple articles written by the same author were only included when it was assured that there was no overlap in the patient data groups.
Articles were included in the meta-analysis when at least 4 studies used the same reference method.
As reference standard SPECT,
PET,
MRI or invasive coronary angiography (ICA) were included.
Also,
test characteristics of CT perfusion per patient,
vessel territory or coronary segment had to be reported in the studies.
Data extracted from the included studies were pooled to calculate measures of diagnostic performance.
These included sensitivity,
specificity,
area under the curve (AUC) and a diagnostic odds ratio (DOR),
with 95% confidence interval (CI).
Test characteristics were calculated using the random effects model.
DORs were used to calculate the summary receiver operating characteristics (SROC) curves.
DOR results are a combination of both sensitivity and specificity in one diagnostic performance parameter.
A score of one indicates the technique has no discriminative ability to determine whether the patient has the disease or not.
Consequently,
a high score means a good discriminative ability of the analyzed CT technique.