-
Re:Statistical vs. Clinical Significance
Submit responseWe fully agree with Dr Miller, who draw attention to the fact that statistical significance is not always equal to clinical significance. We also agree that the effect size is the appropriate measure for clinical relevance. For the difference in mean total problems scores on the Child Behavior Checklist (CBCL) between moderately preterm and term-born children the effect size is 0.22 in our study, being a small (but not negligible) effect. However, clinicians in particular take care for those children that have elevated (clinical) CBCL scores, which have been presented in Table 4. In this Table, the effect sizes for total, externalizing, and internalizing problems are 0.34, 0.27, and 0.50, meaning small (0.34 and 0.27) to moderate effects (0.50). An effect size of 0.5 is often the value to be detected in clinical trials. However, we think that the effects are clinically relevant indeed because of the high prevalence of moderate preterm birth, which implies rather large effects on child public health. We thank Dr Miller for giving us the opportunity to provide this additional information on the relevance of our findings.
Conflict of Interest:
None declared
-
Statistical vs. Clinical Significance
Submit responseThe interesting and well-conducted study of Potijk et al reminds us once again (though we probably don't need reminders) of the important difference between statistically versus clinically significant differences in research studies. The authors report that moderately preterm-born children had significantly worse scores on all subscales of the CBCL than did term born children; inspection of the P values in Table 2 shows that, statistically, this is quite correct. What is not discussed in the paper, however, is the clinical significance of these differences. A commonly used metric for evaluating the clinical significance of observed differences is Cohen's effect size coefficient 'd'. This 'd' is the ratio of the mean difference in scores between two groups to the standard deviation in scores of the groups. Most of the differences shown in Table 2 have a 'd' value of 0.2 or less, which by convention would be considered at the lower end of a small effect size. None of this takes away from the finding that there were differences between the groups, but it is important for readers (and authors) to consider the clinical significance of differences. The use of large study samples can make differences that are small and even trivial clinically, appear quite impressive statistically.
Conflict of Interest:
None declared
Latest from Education and Practice
Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of
ADC.
View free sample issue >>
Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.
- Online First : Online First
- Current issue : Current issue | Mobile site
- Archive : Supplements | eLetters | Topic collections | Blog | Podcasts | RSS
-
About the journal
: About the journal | Editorial board | Instructions for authors | Unlocked |
Thank you to our reviewers | Most read articles | Events - Submit a paper : Online submission site | Instructions for authors | Unlocked | Track your article
- Subscribe : Subscribe to the journal | Prices | Email alerts
- Help : Contact us | Feedback form | Reprints | Permissions | Advertising
