Keywords
acquired hand deformities/classification - finger injuries - reproducibility of results
- rupture - tendon injuries
Introduction
Mallet finger is, by definition, the rupture of the terminal extensor mechanism, resulting
in a flexural deformity of the distal phalanx.[1]
Extensor mechanism injuries are very common in hand injuries. The third, fourth and
fifth fingers of the dominant hand are the most affected, and 95% of the lesions are
closed. The general mechanism of injury is a trauma on the fingertip causing sudden
flexion or counter-resistance extension.[1]
[2]
This lesion, if left untreated, could result in an elongated and potentially less
functional tendon. Rupture of the capsule and retinacular ligaments leads to greater
deformity. Importantly, the fibers of the oblique retinacular ligament also have a
role in the extension. Bone injuries by avulsion cause similar deformities due to
disinsertion alongside the fragment of tendon and ligament structures.[2]
[3]
[4]
The present research has as overall objective to quantify the intra- and interobserver
agreement of the Albertoni classification, and to evaluate the importance of using
the goniometer in the lesions of the terminal extensor mechanism of the finger[3]
[5]
[6] ([Figure 1]).
Fig. 1 Scheme demonstrating the classification proposed by Albertoni (We have omitted types
C and D to emphasize angle measurement). Source: Albertoni et al.[7]
Materials and Methods
The study protocol was approved by the institutional ethics committee, under the number
CAAE 57854216300005135.
This is a cross-sectional observational study conducted between August 08 and 22,
2016. A total of 10 cases with mallet finger injuries were selected to be examined
by lateral radiographs, with and without the use of the goniometer. All images were
digitalized and then analyzed by orthopedists and orthopedics residents working in
public and private health services in the city of Belo Horizonte, state of Minas Gerais,
Brazil and its metropolitan region. The questionnaire is divided into two parts, each
containing 10 questions and a heading with information about the professional formation
([Supplementary Material 1]).
Table 1
|
n
|
%
|
|
Hand Surgery Specialist
|
No
|
50
|
86%
|
|
Yes
|
8
|
13%
|
We have omitted types C and D from the Albertoni classification to emphasize angle
measurement and the use of the goniometer. In types C and D, angulation is not essential
for classification (in addition, they are rarer lesions).
In one of the questionnaires, the radiographs should be analyzed without the use of
a goniometer (NG) and in the other with the use of this instrument (WG). The inclusion
criteria of professionals were familiarity with the use of the goniometer and knowledge
of the Albertoni classification for mallet finger. The radiographs were evaluated
by 60 professionals, being 20 orthopedic residents, 32 orthopedists, and 8 hand surgeons.
To describe the angle between the questions and between the question classifications,
measures of central tendency, dispersion and position were used. To verify intra-
and interobserver agreement, the Kappa coefficient was used.[7]
[8]
The software used in the analysis was R, version 3.2.2.
Results
A total of 60 questionnaires were applied. Of these, two questionnaires were excluded
because they were not complete or had angles measured with unreadable letters.
[Table 1] presents the observers considering the formation of “Hand Surgery”. Eight (13%)
of respondents were hand surgery specialists.
[Table 2] presents the description of the questions classification with and without use of
the goniometer. The response agreement percentage was high, and in questions 2 and
8 the majority was classified as “A”, in questions 3, 4 and 5 most were classified
as “B”, in questions 1, 6, 7 and 9 most were classified as "C" and in question 10
most were classified as "D".
Table 2
|
Questions
|
Questions without using the Goniometer
|
Questions using the Goniometer
|
|
A1
|
A2
|
B1
|
B2
|
A1
|
A2
|
B1
|
B2
|
|
n
|
%
|
n
|
%
|
n
|
%
|
n
|
%
|
n
|
%
|
n
|
%
|
n
|
%
|
n
|
%
|
|
Question 1
|
0
|
0.0%
|
0
|
0.0%
|
58
|
100.0%
|
0
|
0.0%
|
0
|
0.0%
|
0
|
0.0%
|
57
|
100.0%
|
0
|
0.0%
|
|
Question 2
|
58
|
100.0%
|
0
|
0.0%
|
0
|
0.0%
|
0
|
0.0%
|
54
|
94.7%
|
3
|
5.3%
|
0
|
0.0%
|
0
|
0.0%
|
|
Question 3
|
0
|
0.0%
|
57
|
98.3%
|
1
|
1.7%
|
0
|
0.0%
|
0
|
0.0%
|
56
|
98.2%
|
0
|
0.0%
|
1
|
1.8%
|
|
Question 4
|
7
|
12.1%
|
46
|
79.3%
|
1
|
1.7%
|
4
|
6.9%
|
4
|
7.0%
|
52
|
91.2%
|
0
|
0.0%
|
1
|
1.8%
|
|
Question 5
|
1
|
1.7%
|
56
|
96.6%
|
1
|
1.7%
|
0
|
0.0%
|
1
|
1.8%
|
56
|
98.2%
|
0
|
0.0%
|
0
|
0.0%
|
|
Question 6
|
1
|
1.7%
|
0
|
0.0%
|
54
|
93.1%
|
3
|
5.2%
|
2
|
3.5%
|
0
|
0.0%
|
53
|
93.0%
|
2
|
3.5%
|
|
Question 7
|
0
|
0.0%
|
1
|
1.7%
|
55
|
94.8%
|
2
|
3.4%
|
1
|
1.8%
|
0
|
0.0%
|
55
|
96.5%
|
1
|
1.8%
|
|
Question 8
|
55
|
94.8%
|
2
|
3.4%
|
1
|
1.7%
|
0
|
0.0%
|
55
|
96.5%
|
2
|
3.5%
|
0
|
0.0%
|
0
|
0.0%
|
|
Question 9
|
0
|
0.0%
|
4
|
6.9%
|
54
|
93.1%
|
0
|
0.0%
|
0
|
0.0%
|
0
|
0.0%
|
56
|
98.2%
|
1
|
1.8%
|
|
Question 10
|
0
|
0.0%
|
0
|
0.0%
|
0
|
0.0%
|
58
|
100.0%
|
0
|
0.0%
|
0
|
0.0%
|
0
|
0.0%
|
57
|
100.0%
|
The following [Figure 2] illustrates the information of the table above.
Fig. 2 Description of using and without using questions classifications.
[Table 3] presents the mean angle measured by the interviewees.
Table 3
|
Angle
|
Mean
|
SD
|
Min.
|
1° Quartile
|
2° Quartile
|
3° Quartile
|
Maximum
|
|
Question 1
|
18.32
|
3.78
|
10.00
|
16.00
|
20.00
|
20.00
|
28.00
|
|
Question 2
|
22.47
|
4.15
|
13.00
|
20.00
|
21.00
|
26.00
|
35.00
|
|
Question 3
|
40.95
|
4.01
|
30.00
|
40.00
|
40.00
|
44.00
|
50.00
|
|
Question 4
|
35.35
|
4.90
|
26.00
|
32.00
|
35.00
|
40.00
|
48.00
|
|
Question 5
|
35.47
|
4.49
|
28.00
|
32.00
|
35.00
|
39.00
|
49.00
|
|
Question 6
|
21.53
|
3.49
|
10.00
|
20.00
|
20.00
|
22.00
|
32.00
|
|
Question 7
|
9.86
|
4.64
|
2.00
|
8.00
|
10.00
|
11.00
|
35.00
|
|
Question 8
|
17.46
|
5.46
|
10.00
|
13.00
|
18.00
|
20.00
|
32.00
|
|
Question 9
|
11.79
|
4.58
|
4.00
|
10.00
|
10.00
|
12.00
|
35.00
|
|
Question 10
|
44.33
|
4.62
|
34.00
|
40.00
|
45.00
|
48.00
|
58.00
|
The following [Figure 3] illustrates graphically the information of the table above.
Fig. 3 Description of the angle between the questions.
[Table 4] presents the intraobserver and interobserver agreement for each question. From this,
it can be highlighted that the agreement was high for all questions, and:
Table 4
|
Question
|
Intraobserver
|
Interobserver
|
|
Without using
|
Using
|
|
n
|
%
|
n
|
%
|
n
|
%
|
|
Question 1
|
57
|
100.0%
|
58
|
100.0%
|
57
|
100.0%
|
|
Question 2
|
54
|
94.7%
|
58
|
100.0%
|
54
|
94.6%
|
|
Question 3
|
55
|
96.5%
|
57
|
98.2%
|
56
|
98.2%
|
|
Question 4
|
45
|
78.9%
|
46
|
79.0%
|
52
|
91.1%
|
|
Question 5
|
54
|
94.7%
|
56
|
96.5%
|
56
|
98.2%
|
|
Question 6
|
53
|
93.0%
|
54
|
93.0%
|
53
|
92.9%
|
|
Question 7
|
53
|
93.0%
|
55
|
94.7%
|
55
|
96.4%
|
|
Question 8
|
52
|
91.2%
|
55
|
94.7%
|
55
|
96.4%
|
|
Question 9
|
52
|
91.2%
|
54
|
93.0%
|
56
|
98.2%
|
|
Question 10
|
57
|
100.0%
|
58
|
100.0%
|
57
|
100.0%
|
|
Kappa
|
0.906
|
0.829
|
0.865
|
|
P-Value
|
0.000
|
0.000
|
0.000
|
-
Regarding the intraobserver agreement (“with goniometer” versus “no goniometer”),
questions 1 and 10 presented 100% agreement, while question 4 presented the lowest
agreement. The Kappa coefficient presented a value of 0.906, which indicates a high
agreement (p = 0.000).
Discussion
The importance and efficiency of a classification is based on simplicity and reproducibility.
It should also allow the comparison between scientific studies and present high agreement
between examiners.[9]
[10] In assessing the reliability and agreement among observers, there is a need to consider
chance in the evaluation,[11]
[12] and for that the Kappa method of comparison is used. The use of the goniometer and
its influence on the agreement in this classification was also incorporated in this
work.[11]
[13]
[14]
[Table 2] shows the distribution of answers for each question without and with the use of
the goniometer. We see that the questions 1, 2, 3, 5, 6, 7, 8, 9 and 10 presented
low intra- and interobserver variation. Question 4 presented greater disagreement
for the same answer. This is because this question demonstrates a mallet finger injury
with borderline measurement.[15]
[16]
[17]
There was uniformity between the responses, demonstrated by a standard deviation with
little variation between them (between 3.49° and 5.46°). In question 8, the distal
phalanx presented a deformity, which may have affected these isolated results obtained.
In the boxplot chart, it is shown that the variation of angles between quartiles of
each question was small, with a mean of 5.1° (smaller difference: 2° / greater difference:
8°). No work was found as a reference for comparison with this data. However, we consider
the values obtained as adequate.[18]
[19]
The results of the intra- and interobserver evaluation were excellent, as shown by
[Table 4], following the Kappa Intraclass Coefficient Scale ([Table 5]).
Table 5
|
Kappa
|
Interpretation
|
|
< 0
|
No Agreement
|
|
0 - 0.19
|
Poor
|
|
0.20 - 0.39
|
Fair
|
|
0.40 - 0.59
|
Moderate
|
|
0.60 - 0.79
|
Good
|
|
0.80 - 1
|
Excellent
|
There was no statistically significant difference between the evaluators considering
the formation (resident, orthopedist or hand surgeon). Noting that residents who did
not know the classification, or who reported not knowing how to use the goniometer,
were not included.
The interobserver agreement for each question without using the goniometer with Kappa
index 0.829 (excellent) suggests that even without the instrument, the classification
is reproducible for different observers. The use of the goniometer generated a slightly
higher kappa of 0.865, which shows that the goniometer can increase agreement, but
without statistical significance.[20]
[21]
[22]
The intraobserver measure of agreement, by comparing results obtained with and without
the use of the goniometer, showed a Kappa index of 0.906. For this work, it was established
that the classification does not require the use of the goniometer and that the assessment
by a professional who is aware of the classification (orthopedic resident, orthopedist
or hand surgeon) is sufficient.
Conclusion
We conclude that the Albertoni classification is simple, reproducible and with good
intra- and interobserver agreement, both with and without the goniometer. The use
of the goniometer slightly increased the interobserver agreement, but without statistical
significance, suggesting that it is a dispensable instrument for Mallet Finger assessment.
