Small numbers, big decisions: guidance for sample size in pilot trials