Eur J Cardiothorac Surg 2009;35:552-553. doi:10.1016/j.ejcts.2008.11.010
Copyright © 2009, European Association for Cardio-thoracic Surgery. Published by Elsevier. All rights reserved.
Experimental design should be regulated for the study
Hong-Lin Chenb,*,
Qin-Sheng Youa,
Xiao-Dan Wena
a Department of Cardiothoracic Surgery, Affiliated Hospital of Nantong University, Nantong, China
b Nantong University, Qi Xiu Road 19# of Nantong City, 226001, Jiansu, China
Received 27 October 2008;
accepted 12 November 2008.
* Corresponding author. Tel.: +86 513 80501846. (Email: pphss{at}126.com).
Key Words: Pulmonary surgical procedures • Chest tubes • Experimental design
We read with interest the article entitled ‘Comparison of the single or double chest tube applications after pulmonary lobectomies’ by Okur and colleagues [1]. We congratulate the authors for their excellent results.
This is a prospective randomized study, which has strict experimental control. However, we find there are some flaws in the experimental design of this study. It decreases the validness of the conclusion to some extent.
- 1. Randomization is an important principle. Two standards should be included: doctors and patients can not know in advance which group the patient will be assigned to; doctors and patients can not speculate which group the next patient will be assigned to from the former patient. The study assigns patients by entering turns in the study, which obviously break the standard random. This can be referred to as quasi-random [2]. Randomization aims to make uniform distribution of prognostic factors to the study groups, so as to achieve balance and comparability between the groups. Fortunately, from Table 1, we find that the common demographic factors and some preoperative physiological indicators in the groups have reached a balance. However, there may be some unknown prognostic factors between the groups that are not balanced, especially in the relatively small sample size of the study. The sample size of this study is 100.
- 2. Blind is an important method to reduce the bias. The study could not be designed as ‘blind’ because the same surgical team that performed the operations had to manage the patients postoperatively. However, measurements of many indicators for the study will be interfered largely by man-made factors; in particular, using the visual analogue scale (VAS) to assess the degree of pain, intentionally or unintentionally may bias the results. However, blind rules cannot be performed in reality, and it should have taken some quality control methods to reduce the bias at least. But in the text we have not found anything from the method introduced.
- 3. As a RCT study, sample size should be identified in the first. If sample size is too small, differences are not easy to detect; but if samples are excessive, it will cause waste of resources. Sample size can be based on the overall difference estimated between two groups
, the definition of 
, β, as well as the overall level of standard deviation 
. This study does not calculate the sample size. It was stopped when statistically significant differences on pain score were observed between the groups. One hundred consecutive cases were assessed, 50 patients each. In this sense, this design is more similar to the sequential trial. Sequential trial has less rigorous design than RCT, and the conclusions are less valid. As the past study had shown some of the preliminary results [3,4], we believe RCT should be used in this study to further verify the conclusions.
- 4. Also we have different views for surveying the amount of drainage. The drainage pipe itself can stimulate the fluid to occur, resulting in additional loss of body fluids. But to achieve adequate drainage is the most important criterion. Gomez-Caro and colleagues [3] compared residual space and residual pleural effusion in two groups before evaluating the amount of drainage, and this study only compared the residual pleural air space. In our view, respiratory function, such as PO2, PCO2, SPO2, should be compared in the two groups of patients additionally. If patients of the double-tube group with smaller respiratory dysfunction, have even more amount of drainage, we believe it has advantages.
In addition, we find Mann–Whitney U test was used in some items; maybe these data do not obey normal distribution. The data of non-normal distribution should be described by the median and quartile. However, the paper described the data by mean ± standard deviation style.
References
- Okur E, Baysungur V, Tezel C, Halezeroglu S, Sevilgen C, Gokhan E, Mertol G. Comparison of the single or double chest tube applications after pulmonary lobectomies. Eur J Cardiothorac Surg 2009;35:32-36.[Abstract/Free Full Text]
- The Cochrane Collaboration. Cochrane Handbook for Systematic Reviews of Interventions 4.2.5, 2005:191.
- Gomez-Caro A, Roca MJ, Torres J, Cascales P, Terro E, Castaner J, Pinero A, Parilla P. Successful use of a single chest drain postlobectomy instead of two classical drains: a randomized study. Eur J Cardiothorac Surg 2006;29:562-566.[Abstract/Free Full Text]
- Alex J, Ansari J, Bahalkar P, Agarwala S, Rehman MU, Saleh A, Cowen ME. Comparison of the immediate postoperative outcome of using the conventional two drains versus a single drain after lobectomy. Ann Thorac Surg 2003;76:1046-1049.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
E. Okur, V. Baysungur, C. Tezel, and S. Halezeroglu
Reply to chen et Al.
Eur. J. Cardiothorac. Surg.,
March 1, 2009;
35(3):
553 - 553.
[Full Text]
[PDF]
|
|
|
