Blood donation, body iron stores, and risk of myocardial infarctionBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7097.1830 (Published 21 June 1997) Cite this as: BMJ 1997;314:1830
Confidence intervals and possible selection bias call study results into question
- Harri Hemilä, Senior research fellowa,
- Mikko Paunio, Senior research assistanta
- a Department of Public Health, POB 41, University of Helsinki, FIN-00014 Finland
- b Research Institute of Public Health, University of Kuopio, 70211 Kuopio, Finland
Editor—The importance of stating the 95% confidence interval when reporting results has been shown in a recent paper by Tomi-Pekka Tuomainen and colleagues.1 In their paper blood donation reduced the risk of myocardial infarction anywhere between 3% and 98%. An estimate with such imprecision seems to be of little use. Moreover, although the association was only marginally significant to begin with (P=0.047), the authors commented that further adjustments attenuated the association marginally. The confidence interval cannot become considerably wider with further adjustments; even a marginal change at the lower limit of the confidence interval can cause a change in the sign, rendering the association non-significant.
The authors urged that new studies be carried out to confirm their findings. However, although the results of several previous studies were unable to corroborate the hypothesis that raised iron …