Consider a Wald-type confidence interval for a proportion $p$ given by $\hat{p} \pm z_{\alpha/2} \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$. In a clinical trial with $n=20$ and 0 observed successes, what is the primary statistical issue with this interval?