Randomised controlled trials: inferring significance of treatment effects based on confidence intervalsBMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g5196 (Published 18 August 2014) Cite this as: BMJ 2014;349:g5196
- Philip Sedgwick, reader in medical statistics and medical education1
- 1Centre for Medical and Healthcare Education, St George’s, University of London, London, UK
Researchers investigated the effects of a food supplement on body mass in wasted adults with HIV who were starting antiretroviral therapy. A randomised controlled superiority trial was performed. Intervention was a fortified spread food supplement delivered for 14 weeks. Control treatment was a corn-soy blend food supplement—the most commonly available supplementary food in food aid programmes. The setting was a large public clinic associated with a referral hospital in Blantyre, Malawi. Participants were adults with a body mass index (BMI) less than 18.5 kg/m2. A total of 491 participants were recruited and randomised to the fortified spread (n=245) or corn-soy blend (n=246) food supplement groups. The primary outcomes were changes in BMI and fat-free body mass from baseline at 14 weeks.1
After 14 weeks of intervention, the mean change in BMI was an increase of 2.2 kg/m2 (95% confidence interval 1.96 to 2.44) for the fortified spread group and 1.7 kg/m2 (95% confidence interval 1.49 to 1.91) for the corn-soy blend group. The mean change in fat-free body mass was an increase of 2.9kg (95% confidence interval 2.50 to 3.30) for the fortified spread group, and 2.2kg (95% confidence interval 1.82 to 2.58) for the corn-soy blend group.
Which of the following statements, if any, are true?
a) The difference in fat-free body mass between treatment groups was not significant at the 5% level because the 95% confidence intervals for the two groups overlapped.
b) The difference in BMI between treatment groups was significant at the …