Worst Case Accuracy is effectively the furthest the actual output voltage can be from the ideal at any given code. If the DAC is programmed it would be expected that the output would load to a certain value. Depending on the level of user experience, errors can be accounted for.  However, there may be surprises when the output voltage or current is not where expected. Looking at most DAC datasheets, there are many separate error specifications detailed. These sources of error are offset error, gain error, and linearity errors.  Calculating the overall error specification in the absence of a total unadjusted error specification can be done using the formula shown here. As well as integral non-linearity, this formula also accounts for offset and gain errors or fullscale and zeroscale errors. Fullscale/Offset error and Zeroscale/gain error can be used interchangeably in the formula.