A multitude of data-driven genealogical surveys have replicated this error, which stems simply from the faulty transcription of one record: Doda (Duxia) is assumed to not have died before giving birth at least to Herleva in 1003.
Hence she died in that year, or conceivably decades after.