After cutting off the clinical crown at CEJ by diamond fissure burs (MANI, Japan) using a high-speed hand piece with water spray cooling system, an ISO size K-file #10 (MANI, Japan) was inserted into the root canal and instrumented to apical foramen to ensure patency. The working length was defined as 0.5 mm short of the foramen. Then 5 millimeters of the coronal third of each canal was enlarged by no #3, no #2, and no #1 Gates-Glidden drills (MANI, Japan). The remaining of the root canal was hand instrumented with K-file ISO size #25 to #50 (MANI, Japan) using a step-back technique. Each sample was buried vertically from its coronal side in self-curing acrylic resin (Aria Dent, Iran). The apical 3 mm of each root was then cut perpendicular to the long axis using a #0.8 fissure diamond bur in a high-speed hand piece with water spray (
Fifty samples were randomly selected for creating artificial cracks. Consequently, a cylindrical steel wedge (1.2 millimeter in diameter) was attached to a force application universal testing machine (Instron, Germany) that was used to vertically apply the force to the apical end of the root (
The machine had a sensitive pen to measure the applied forces and register it on graduated paper. Immediately after a crack formed on the root-end surface, a serrated line was registered on the paper. The amount of applied forces was between 50 and 60 N.
In order to confirm artificial cracks, the same stereomicroscope was used at 50x magnification (cracked samples), but samples with visible cracks by unaided vision were excluded from the study and replaced with new ones. Subsequently, all parts of each sample were covered by a rubber dam except for 2 mm of the apical third (
Four dental students who were blind to the procedures of this study, evaluated all 100 specimens (cracked=50 and non-cracked=50) using each of the following diagnostic methods. Between four blind observers, the observer with the least correct diagnosis was excluded. Then, the diagnoses of the three other observers were recorded. In the final analysis, minimum two observers with a similar opinion was considered.
Prior to the main study the inter-observer reliability was measured in a pilot study.
Sensitivity and specificity were calculated by the following equations:
To select the suitable method for detecting cracks, Youden index was used.
To compare agreement between studied methods and the gold standard, kappa statistics and odds ratio of McNemar were calculated. To compare sensitivities and specificities between different methods, Cochran’s Q test was performed and adjusted p-values for pair wise comparisons were reported. In this study, p-values of less than 0.05 were considered statistically significant.