J Knee Surg 2022; 35(04): 466-468
DOI: 10.1055/s-0040-1715131
Letter to the Editor

Comment on “Robot-Assisted versus Conventional Total and Unicompartmental Knee Arthroplasty: A Meta-analysis of Radiological and Functional Outcomes”

Kai Lei
1   Center for Joint Surgery, Third Military Medical University, Southwest Hospital, Chongqing, China
,
2   Minimally Invasive Gastrointestinal Surgery Center, Third Military Medical University, Southwest Hospital, Chongqing, China
› Author Affiliations

We read with interest the article “Robot-Assisted versus Conventional Total and Unicompartmental Knee Arthroplasty: A Meta-analysis of Radiological and Functional Outcomes” by Chin et al[1] and appreciate the authors for their work; however, we found that the analysis of some data in this article is not appropriate.

The authors used meta-analysis to synthesize the mechanical axis deviation and component angle deviations, which is very innovative, but there are statistical errors. When synthesizing continuous variables into weighted mean difference (WMD), it is not accurate to directly use the difference between an angle and its target value, but to include the absolute value of the difference between this angle and its target value in the calculation.

Take the coronal tibial component angle as an example. Assuming that its target value is 90 degrees, and the original data of groups A and B conform to the normal distribution ([Table 1]), then group A is often expressed as 90.2 ± 3.83, and group B is often expressed as 91.8 ± 0.84. If the WMD is directly synthesized using deviations, that is, 0.2 ± 3.83 for group A and 1.8 ± 0.84 for group B are used for synthesizing the data, it will be concluded that group A has better control of the coronal tibial component angle. However, the correct method should be as follows: 1). to obtain the difference between each original data and its target value; 2). to obtain its absolute value of the difference; 3). to include the absolute deviation in the calculation. In this way, 3.0 ± 1.87 for group A and 1.8 ± 0.84 for group B are used for synthesizing the data, and we could reach the conclusion that group B is actually better ([Fig. 1]).

Table 1

The coronal tibial component angle of groups A and B

Raw data

Deviations

Absolute deviations

Group A

 a1

86.0

86.0–90

|86.0–90|

 a2

87.0

87.0–90

|87.0–90|

 a3

90.0

90.0–90

|90.0–90|

 a4

93.0

93.0–90

|93.0–90|

 a5

95.0

95.0–90

|95.0–90|

Mean ± SD

90.2 ± 3.83

0.2 ± 3.83

3.0 ± 1.87

Group B

 b1

90.5

90.5–90

|90.5–90|

 b2

91.5

91.5–90

|91.5–90|

 b3

92.0

92.0–90

|92.0–90|

 b4

92.5

92.5–90

|92.5–90|

 b5

92.5

92.5–90

|92.5–90|

Mean ± SD

91.8 ± 0.84

1.8 ± 0.84

1.8 ± 0.84

Abbreviation: SD, standard deviation.


Zoom Image
Fig. 1 The forest plot of coronal tibial component angle.

By reading the article and the references it cited, we found that the authors used the above-mentioned direct synthesizing method, and the subsequent results and conclusions are based on this inappropriate statistical method, which needs to be further verified.



Publication History

Article published online:
24 August 2020

© 2020. Thieme. All rights reserved.

Thieme Medical Publishers, Inc.
333 Seventh Avenue, 18th Floor, New York, NY 10001, USA