40 subjects (17 females and 23 males) participated in this study. The descriptive statistics for age, CBCT and USD measurements are reported in Table 1.
Table 1. Descriptive statistics of all gingival thickness measurements (in mm).
|
Mean (Std.Dev.)
|
Min
|
Max
|
Age
|
24.48 (6.68)
|
18.00
|
45.00
|
Mandibular left central incisor
|
|
|
|
CBCT: Examiner #1
|
0.93 (0.24)
|
0.55
|
1.51
|
CBCT: Examiner #2
|
1.01 (0.24)
|
0.61
|
1.59
|
USD
|
0.80 (0.26)
|
0.50
|
1.50
|
Mandibular right central incisor
|
|
|
|
CBCT: Examiner #1
|
0.95 (0.27)
|
0.50
|
1.50
|
CBCT: Examiner #2
|
0.99 (0.25)
|
0.42
|
1.43
|
USD
|
0.80 (0.22)
|
0.50
|
1.50
|
Table Legends. CBCT: Cone Beam Computed Tomography, USD: Ultrasound Device
Repeatability assessment
The results of the paired t-tests for bias between the 1st and the 2nd CBCT measurements made by the first examiner (DK) are reported in Table 2. Repeatability of USD measurements were performed in a previous cross-sectional study with the same methodology and objective [22]. Normality assumption was not violated for any of the differences between the repeated measurements of USD in two time points (Mean diff. 0.00, 95% CI -0.05, 0.05, p=1.00).
Table 2. Results of the paired t-tests between the repeated CBCT measurements by examiner #1.
Tooth
|
Bias (Std.Er.)
|
95% CI
|
p-value
|
Mandibular left central incisor
|
0.06 (0.02)
|
(0.01, 0.11)
|
0.014
|
Mandibular right central incisor
|
0.05 (0.03)
|
(-0.01, 0.11)
|
0.104
|
Bias= 2nd - 1st measurements (in mm)
|
Table Legends. CI: Confidence Interval
Statistical analysis indicated that the repeated CBCT measurements were not identical for the mandibular left central incisor (Bias= 0.06 mm, 95% CI= 0.01, 0.11, p-value= 0.014), whereas the respective for the mandibular right central incisor could be considered identical (Bias= 0.05 mm, 95% CI= -0.01, 0.11, p-value= 0.104). Nevertheless, clinically, a difference of 0.06 mm in repeated measurements can be regarded mostly as unimportant. The corresponding Bland – Altman plots are displayed in figures 1a & 1b. Neither a magnitude nor a dispersion related trend could be identified graphically. Absence of the former trend was also implied by the Spearman’s rank correlation coefficient (Mandibular left central incisor: Spearman’s rho= 0.10, p-value= 0.554; mandibular right central incisor: Spearman’s rho= -0.14, p-value= 0.377). The Shapiro – Wilk test results showed that the normality hypothesis was valid for both left and right mandibular incisors (p-value= 0.595 and 0.614 respectively).
Reproducibility assessment
The results of the paired t-tests for bias between the two examiners (DK, LK) are reported in Table 3. Again, statistical analysis indicated that the repeated measurements were not identical for the mandibular left central incisor (Bias= 0.06 mm, 95% CI= 0.01, 0.11, p-value= 0.014), whereas the respective for the mandibular right central incisor could be considered identical (Bias= 0.05 mm, 95% CI= -0.01, 0.11, p-value= 0.246). Nevertheless, clinically, a difference of 0.06 mm in different doctors’ measurements can be regarded mostly as unimportant. The corresponding Bland – Altman plots are displayed in figures 1c & 1d. Neither a magnitude nor a dispersion related trend could be identified (Mandibular left central incisor: Spearman’s rho= -0.03, p-value= 0.836; mandibular right central incisor: Spearman’s rho= -0.12, p-value= 0.377). The normality hypothesis was valid for both left and right mandibular central incisors (Shapiro – Wilk test p-value= 542 and 0.475 respectively).
Table 3. Results of the paired t-tests between the CBCT measurements made by the two examiners.
Tooth
|
Bias (Std.Er.)
|
95% CI
|
p-value
|
Mandibular left central incisor
|
0.06 (0.02)
|
(0.01, 0.11)
|
0.014
|
Mandibular right central incisor
|
0.05 (0.03)
|
(-0.01, 0.11)
|
0.246
|
Bias= examiner #2 – examiner #1 (in mm)
|
Table Legends. CI: Confidence Interval
Method agreement (comparability)
The results of the paired t-tests between the two GT measuring techniques as well as the estimated corresponding 95% LOA and the respective 95% CIs are reported in Table 4. The respective Bland – Altman plots are displayed in figure 2a-d. There was no evidence of a magnitude related trend for either the differences or for their dispersion after graphical evaluation.
Finally, all normality assumptions could not be rejected after either graphical evaluation or using Shapiro-Wilk tests (Mandibular left central incisor, CBCT measurements from examiner #1: p-value= 0.163; CBCT measurements from examiner #2: p-value= 0.561; Mandibular right central incisor, CBCT measurements from examiner #1: p-value= 0.157; CBCT measurements from examiner #2: p-value= 0.097).
Table 4. Results of the paired t-tests (Bias, Std.Er., 95% CI, p-value) between the CBCT and USD measurements along with the corresponding 95% LOA and the respective 95% CI for the LOA.
|
Bias
(Std.Er.)
|
95% CI
for bias
|
p-value
|
95% LOA
|
95% CI for the LOA
|
|
|
|
|
|
Lower LOA
|
Upper LOA
|
Mandibular left
central incisor
|
|
|
|
|
|
|
CBCT measurements
from examiner #1
|
0.13 (0.03)
|
(0.07, 0.19)
|
<0.001
|
(-0.26, 0.51)
|
(-0.35, -0.17)
|
(0.42, 0.60)
|
CBCT measurements
from examiner #2
|
0.21 (0.04)
|
(0.14, 0.28)
|
<0.001
|
(-0.22, 0.64)
|
(-0.32, -0.12)
|
(0.54, 0.75)
|
Mandibular right
central incisor
|
|
|
|
|
|
|
CBCT measurements
from examiner #1
|
0.14 (0.03)
|
(0.08, 0.21)
|
<0.001
|
(-0.26, 0.54)
|
(-0.35, -0.16)
|
(0.45, 0.63)
|
CBCT measurements
from examiner #2
|
0.18 (0.04)
|
(0.11, 0.25)
|
<0.001
|
(-0.25, 0.62)
|
(-0.36, -0.15)
|
(0.52, 0.72)
|
Bias= CBCT – USD (in mm)
|
Table Legends. CBCT: Cone Beam Computed Tomography, CI: Confidence Interval, LOA: Limit of Agreement, USD: Ultrasound Device