Wrong Conclusion when the Statistical Test is Incorrect

0 downloads 0 Views 44KB Size Report
Regarding original article LA Adinegara and MS Razzak. Does Lifestyle Increase the Incidence of Pregnancy-. Induced Hypertension? Med J. Malaysia 2004; ...
23/WRONG CONCLUSION

30/12/04

2:19 PM

Page 692

LETTER TO EDITOR

Wrong Conclusion when the Statistical Test is Incorrect B Norsa'adah, MComMed, Medical Lecturer Unit of Biostatistics and Research Methodology, School of Medical Sciences, University Sains Malaysia, 16150 Kubang Kerian, Kelantan

Sir, Regarding original article LA Adinegara and MS Razzak. Does Lifestyle Increase the Incidence of PregnancyInduced Hypertension? Med J. Malaysia 2004; 59(1): 39-44. I would like to point that the authors have been using incorrect statistical tests that make the results and conclusions invalid Tables I and III. For the quantitative analysis of Tables I and III, the authors

used student t test. This is a matched case control study and the correct statistical analysis is a paired t test1. In Table II, for the analysis of qualitative data, the authors used chi-square and Fisher's exact tests. The right test is a Mc Nemar test because of the same The authors can also proceed with reasons1. conditional logistic regression after the univariate analysis, thus are able to control confounders in the study.

References

1. Schlesselman JJ. Case-control studies. Design, conduct, analysis. 1982. Oxford University Press: New York.

692

Med J Malaysia Vol 59 No 5 December 2004