56 / 232
anomalies, GER, intact palate, airway anomalies, and pre-operative
intubation. However, when failure is de
fi
ned as limited improve-
ment in AHI, there are no variables statistically associated with
failure. This suggests that multifactorial or unanalyzed variables are
in
fl
uencing failure in this unanalyzed and previously unreported
sub-population.
As reported previously, laryngomalacia is not associated with
failure of MDO across any of the analyzed variables (
p
<
0.05)
(
Tholpady et al., 2015
). When this supraglottic disease is separated
from other airway anomalies, a clear difference can be seen be-
tween the two variables. Non-laryngomalacia airway anomalies are
associated with failure by tracheostomy and so should still be
approached with the knowledge that MDO will not be successful at
a higher rate.
The analysis of this patient population provides the basis for
score creation, much like the GILLS score (
Rogers et al., 2011;
Abramowicz et al., 2012
). The score is a well-known predictor of
success of tongue-lip adhesion (TLA) in the RS population. It has
identi
fi
ed
G
ER,
I
ntubation pre-operatively,
L
ate operation,
L
ow
birth weight, and
S
yndromic diagnosis as important predictors of
success; fewer than three of these predicts a 100% successful TLA.
Of these variables GER and intubation pre-operatively were iden-
ti
fi
ed as being important in this mandibular distraction study. Low
birth weight was not shown to be signi
fi
cant, but approached sig-
ni
fi
cance in the deceased population (
p
<
0.06). Syndromic status
was not signi
fi
cant.
Interestingly, using paired
t
-test analysis, a breakpoint was
identi
fi
ed between successful and unsuccessful MDO with regard
to the age at performance of distraction. RS patients below 30 days
of age at the time of distraction were more likely to be successful
than children older than 2 months. The reasons for this age dif-
ference could be many, but this is similar to the GILLS score in that
Table 3
Paired
t
-test of numeric variables between mandibular distraction osteogenesis successes and failures.
Failure by tracheostomy
Failure by AHI
Any failure
Any failure
þ
deceases
No
Yes
p
value
No
Yes
p
value
No
Yes
p
value
No
Yes
p
value
n
74
7
e
68
6
e
65
13
-
64
16
e
Birth weight (kg)
2.97
2.46
0.067
3.00
2.69 0.307
3.02
2.61 0.053
3.03
2.50
0.006
Age (days)
29.9
70.5
0.001
32.2
43.0
0.423
30.8
51.7
0.037
29.6
49.3
0.031
Weight (kg)
3.33
3.75
0.417
3.43
3.16 0.419
3.39
3.44 0.898
3.38
3.28
0.707
AHI: apnea-hypopnea index.
Signi
fi
cant values (
p
<
0.05) are listed in bold.
Table 4
Sensitivity, speci
fi
city, positive and negative predictive value for the combinations of the six variables with corresponding receiver operating characteristic curve scores.
Combination of
variables analyzed
a
GAITO
GIANTO
GIT
OAINT
OTIG
TINGO
TING
TONI
Failure by tracheostomy
Speci
fi
city
100
100
85.7
100
100
100
85.7
100
Negative predictive
value
22.6
19.4
54.5
28
31.8
25
31.6
43.8
Positive predictive
value
100
100
98.6
100
100
100
98.4
100
Sensitivity
67.6
60.8
93.2
75.7
79.7
71.6
82.4
87.8
Area under ROC
curve
0.93 (0.86
e
1)
0.94 (0.87
e
1)
0.93 (0.86
e
1)
0.94 (0.88
e
1)
0.95 (0.9
e
1)
0.96 (0.9
e
1)
0.92 (0.82
e
1)
0.96 (0.92
e
0.99)
Failure by AHI
Speci
fi
city
83.3
83.3
33.3
66.7
83.3
83.3
50
50
Negative predictive
value
18.5
16.1
28.6
18.2
27.8
21.7
21.4
23.1
Positive predictive
value
97.9
97.7
94
96.2
98.2
98
95
95.1
Sensitivity
67.6
61.8
92.6
73.5
80.9
73.5
83.8
85.3
Area under ROC
curve
0.75 (0.5
e
1)
0.75 (0.49
e
1)
0.73 (0.53
e
0.94) 0.71 (0.47
e
0.96) 0.77 (0.53
e
1)
0.75 (0.51
e
1)
0.72 (0.5
e
0.95) 0.72 (0.5
e
0.94)
Any failure
Speci
fi
city
84.6
84.6
53.8
76.9
84.6
84.6
61.5
69.2
Negative predictive
value
36.7
32.4
70
40
52.4
42.3
47.1
56.3
Positive predictive
value
95.8
95.5
91.2
94.3
96.5
96.2
91.8
93.5
Sensitivity
70.8
64.6
95.4
76.9
84.6
76.9
86.2
89.2
Area under ROC
curve
0.8 (0.64
e
0.97) 0.8 (0.63
e
0.97) 0.8 (0.66
e
0.95) 0.79 (0.63
e
0.95) 0.83 (0.67
e
0.99) 0.81 (0.65
e
0.98) 0.78 (0.63
e
0.94) 0.81 (0.66
e
0.96)
Any failure including all deaths
Speci
fi
city
75
81.3
43.8
68.8
68.8
68.8
50
56.3
Negative predictive
value
38.7
36.1
63.6
44
50
39.3
42.1
56.3
Positive predictive
value
92
93.3
87.1
91.1
91.5
90.6
87.1
89.2
Sensitivity
70.8
64.6
93.8
78.5
83.1
73.8
83.1
89.2
Area under ROC
curve
0.74 (0.58
e
0.91) 0.77 (0.63
e
0.92) 0.72 (0.57
e
0.87) 0.77 (0.63
e
0.91) 0.73 (0.56
e
0.9) 0.76 (0.61
e
0.9) 0.75 (0.62
e
0.88) 0.75 (0.61
e
0.9)
AHI: apnea-hypopnea index; ROC: receiver operating characteristic.
a
The combinations of variables analyzed are drawn from the following:
G
astroesophageal re
fl
ux;
A
ge
>
30 days;
N
eurologic anomaly; airway anomalies
O
ther than lar-
yngomalacia;
I
ntact palate; and pre-operative intubation.
R.L. Flores et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1614
e
1619
34