"Tennis is not dangerous for the spinal column while growing:

results from a cross-sectional study "

​

( Zaina F  et al. Tennis is not dangerous for the spine during growth: results of a cross-sectional study. Eur Spine J  25 (9): 2938-44, 2016 )

Criticalities found

In the text (excerpt, introduction and methods) it is reported that the aim of this work is to verify the prevalence of spinal deformities and back pain in adolescents who play tennis at a competitive level, compared with healthy controls. A cross-sectional study design was used to achieve this goal.

We observe several critical issues in the choice of study design and its consequences.

1. The cross designs have strengths and weaknesses. Indeed:

- Collect information on exposure to risk factors and their outcomes (onset of the disease) at the same time and on the same patient. They can be thought of as a snapshot of a disease in a population at a particular time. However, since exposure and disease status are measured at the same instant in time, this is why they are often used only in the first instance, to then carry out case-control or cohort studies;

- They offer immediate results and are not economically relevant as they do not require the use of resources, time and personnel for long periods (as opposed to case-control or even more cohort studies);

- They are useful for exploring the distribution of a disease at time "0", the association between disease and random factor; they can be considered as a first phase of a study to be deepened with other types of designs (case-control or cohort);

- It is not clear why a control group was taken, as it is not included in this type of study. Furthermore, in the text they are defined as "healthy controls", when instead it is expressly stated that for the choice of this group no restrictions were placed on the type of sport practiced, nor on a previous diagnosis of disease. By definition, a control group is a group of subjects who, apart from the fact that they have not developed the disease, are identical in all other characteristics to the case group.

2. In the section dedicated to "Statistical Analysis" the software used to carry out the analysis has not been reported, nor whether or not the normality of the data has been tested. Furthermore, it was not specified which type of descriptive statistic was used in the following tables. In fact we observe for continuous variables a xy.z ± xy.z (x, y, z are digits from 0 to 9), but it is not clear what the quantity to the right of the "±" represents, but only to guess that it could be a deviation standard or a standard error being the quantity on the left of the "±" symbol an average.

3. In the “Methods” the use of the Chi-square test is described. To use this method, the total number should be greater than 200 and no observed frequency should be less than 5. If the first hypothesis is violated, we should resort to Yates' continuity correction, in the second case, use Fisher's exact test. In the present work, in tables 1 and 2 we observe the lack of both requirements, but the Chi-square test is still used, probably leading to misleading results.

4. In the case of non-statistically significant results, a "NS" was simply reported, without having the magnitude of non-significance (it would have been more appropriate also in these cases to report the p-value).

​

5. In the section "Results" it is reported that "the mean of ATR does not differ between groups, both in males and in females" (5): this is not true at all; in fact, the means differ between groups, both in males and females, but these differences are not statistically significant. It would therefore have been appropriate to specify that the mean ATR does not differ statistically significantly between groups in both males and females, an approach followed elsewhere in the results for other variables.

6. The “Extract” (Results) shows values for ATR that are not found in the text: for females, tennis group 3.2 ° ± 1 ° when table 10.1 shows 3.2 ° ± 2.1 °; for the "School" group the extract shows 2.8 ° ± 1 °, while in table 10.1, 2.8 ° ± 1.7 °. For the male group a similar thing happens: tennis group 2.8 ° ± 1 ° when in table 10.2 2.8 ° ± 1.5 ° is reported; for the "School" group the extract shows 2.6 ° ± 1 °, while in table 10.2 2.5 ° ± 1.4 °. In this last case the p value was also wrong, as it is reported in the extract "p <0.05" and in the "NS" table: this generates a lot of confusion.

7. In table 3 the row of male and female subjects appears to have been inverted: it appears that the male subjects who play tennis have an average age of 12.0 ± 0.8 while from table 2 the value reported is 12.0 ± 1.0. Same thing for female subjects: they have an average age of 12.0 ± 1.0 while from table 10.1 the reported value is 12.0 ± 0.8: the values have probably been reversed. As regards the ATR variable, the same thing is observed, making the relationship between the values shown in the table and the respective number of groups: the value 28% is for female subjects, 26% for males . Furthermore, it is not clear why in table 3 it was considered in ATR≥5 ° and ATR≥7 °, while in table 1 and 2 ATR> 5 ° and ATR> 7 °: perhaps it means that there is no subject with ATR = 5 and ATR = 7? This is misleading to the understanding and comparison that we want to make in table 3 by exploiting the data of this work.

8. In the "Discussions" section it is stated that, from the results achieved, the idea that tennis is traditionally considered as a sport that can damage the spine and increase the risk of spinal deformities is rejected, as a similar prevalence is observed in the two groups under comparison. This type of conclusion cannot be reached with the study design employed, but with a case-control or cohort. At the same time, it is rightly said later that, due to the type of design employed, we cannot establish a cause-effect relationship, but simply a correlation, awaiting future studies. In the "Conclusions" the lack of association between Tennis and LBP is rightly reiterated.

9. Given the correct conclusions, and the way of interpreting the results, the title of this paper is misleading as it seems to establish a cause-effect relationship, asserting that tennis is not dangerous for pathologies related to the spinal column during growth. In the opinion of the writer, a title such as: "Tennis is not correlated with pathologies related to the spine during growth: results from a cross-sectional study" would have been more correct (even if less captivating and / or mediatically expressive).

BIBLIOGRAPHY

1 Becker TJ. Scoliosis in swimmers. Clin Sports Med 5: 149–158, 1986.

2 Zaina F et al. Swimming and spinal deformities: a cross-sectional study. J Pediatr 166: 163–167, 2015.

3 Warren MP et al. Scoliosis and fractures in young ballet dancers. Relation to delayed menarche and secondary amenorrhea. N Engl J Med 314: 1348–1353, 1986.

4 Sato T et al. Back pain in adolescents with idiopathic scoliosis: epidemiological study for 43,630 pupils in Niigata City, Japan. Eur Spine J 20: 274-27, 2011.

5 Zaina F et al. Tennis is not dangerous for the spine during growth: results of a cross-sectional study. Eur Spine J  25 (9): 2938-44, 2016.

​