In the face of a disaster hospitals are expected to be able to continue providing efficient and high-quality care to patients. It is therefore crucial for hospitals to develop business continuity plans (BCPs) that identify their vulnerabilities, and prepare procedures to overcome them. A key aspect of most hospitals' BCPs is creating the backup of the hospital information system (HIS) data at multiple remote sites. However, the need to keep the data confidential dramatically increases the costs of making such backups. Secret sharing is a method to split an original secret message so that individual pieces are meaningless, but putting sufficient number of pieces together reveals the original message. It allows creation of pseudo-redundant arrays of independent disks for privacy-sensitive data over the Internet. We developed a secret sharing environment for StarBED, a large-scale network experiment environment, and evaluated its potential and performance during disaster recovery. Simulation results showed that the entire main HIS database of Kyoto University Hospital could be retrieved within three days even if one of the distributed storage systems crashed during a disaster.