Table 3.

Sample size assumptions and minimum detectable relative reductions for the coprimary aims

	Original	Revised
Target sample size, mother–infant cohort	33 600*	49 600

Aim 1A. Effect of intervention on composite outcome of stillbirths and infant deaths from birth to 12 months
Baseline combined stillbirth and infant mortality rate	105/1000†	74/1000†
Minimum detectable relative mortality reduction, comparing azithromycin to placebo treatment arms
80% power	9.5%	8.7%
90% power	10.9%	10.1%

Aim 1B. Effect of intervention on infant deaths from first dose (approx. 6 weeks) to 12 months
Infant mortality rate	28/1000‡	9/1000‡
Minimum detectable relative mortality reduction, comparing azithromycin to placebo treatment arms
80% power	18.6%	24.7%
90% power	21.4%	28.3%
With addition of infants from the infant-only cohort§
80% power	16.6%	22.3%
90% power	19.1%	25.6%

*In the original sample size calculation, 15% anticipated loss to follow-up (LTFU) meant that 28 560 infants were expected to have analysable endpoints.

†Original assumed composite baseline stillbirth and infant mortality was 105 deaths per 1000 births, calculated from 33 stillbirths/1000 births⁵⁵ and 74.8 infant deaths per 1000 live births (Sikasso data, Mali 2012 Demographic and Health Survey [DHS] report⁵⁶), assuming that 6–11 months mortality data are available for 90% of infants and subtracting stillbirths from the live birth denominator. Revised observed composite stillbirth and infant mortality rate in the study population is 74 deaths per 1000 births.

‡Original baseline infant mortality rate after 6 weeks of age was 28 per 1000 live births (Sikasso data, 2012 Mali DHS report) assuming that 6–11 months mortality data are available for 90% of infants; revised infant mortality rate after 6 weeks of age was updated to 9 deaths per 1000 live births based on observed mortality in the study population.

§Original target sample size for infant-only cohort was 8500, assuming 7650 analysable (10% LTFU); revised target sample size 12 000.