To Adjust or Not to Adjust? When a “Confounder” Is Only Measured After Exposure