In such circumstances it can be confusing for readers if the two are inconsistent: for example if the null hypothesis statistical test suggests that the null hypotheses of p=0 cannot be rejected at the 5% level but yet the 95% confidence interval does not contain the value zero.