Year: 2003
Type: letter
Abstract: Stattin and Kaaks's data show that insulin levels increase with tumour stage, which is what we reported. They do not have a large enough sample size (sufficient statistical power) to conclude that there is no difference in the insulin levels of the groups in the table. I suggest that Stattin and Kaaks collect a larger sample after determining the sample size; they need to provide 90% power of detecting a significant ( P =0.05 two-tailed) result. Also, they might want to measure fasting insulin before breakfast to reduce the variability of the data, since insulin levels fluctuate throughout the day and after meals. In the meantime, I refer Stattin, Kaaks and all interested readers to the article by Altman and Bland (1995) , which eloquently explains the pitfalls of concluding clinical insignificance when differences between sets of small sample size are found to be statistically nonsignificant.
Source: British Journal of Cancer
Author: S Lehrer
Language: English
Subfield: Cancer Research
Domain: Life Sciences
Open Access status: hybrid
Estimated APC paid: $5,190
OpenAlex ID: W2013089294