BACKGROUND: As of 12/20/15, >1200 Zika virus (ZVD) infections were reported in Florida, including 252 infections acquired locally in Florida. ZVD response activities for travel-associated and locally acquired cases have been extensive; data have been collected for >10,000 persons under investigation (PUIs) and urosurvey participants (UPs). Thirty-two urosurveys were conducted to assess ongoing local transmission. Compared to routine reportable disease surveillance, the enhanced ZVD response posed additional data management challenges.
METHODS: Merlin, Florida’s reportable disease surveillance application, was the primary system for data collection on PUIs and UPs. Staff in each of 67 Florida counties enter data on cases, PUIs, and UPs into Merlin. System functionality was developed as areas for improvement were identified.
RESULTS: Successful existing Merlin functionality included automated processing of electronic laboratory reporting (ELR) results, case creation, and case classification based on clinical and laboratory information; an outbreak module allowing users to connect people to events; and dynamic data collection screens that could be updated in real time. Data collection screens for ZVD cases include many data fields needed for confirmed/probable cases. PUIs were entered using the same screens though only minimal information was needed until they were ruled in or out as a case. Identifying only the key data fields needed for PUIs was cumbersome for users. A “short survey” solution was designed that included only the key fields needed for PUIs. If later determined to be a case, normal case screens displayed retaining all information from the short survey. Auto-processing of ELR results and case creation resulted in duplicate profiles and cases, highlighting the need for better profile matching algorithms, which will be implemented in January 2017. Increased focus on laboratory screening necessitated additional development of auto-processing logic allowing automated handling of negative results while maintaining normal processes for positive results. Data and specimen collection in the field highlighted the need for interoperability between Merlin and the laboratory information management system. Requirements are currently being defined for this project.
CONCLUSIONS: Data management needs during large scale responses differ compared to routine surveillance and from response to response. Data challenges responding to ZVD were different than challenges encountered during Ebola virus disease (EVD) response, though functionality built during EVD supported ZVD response. Defining new business requirements and rapid development of functionality is fundamental to agile data management during and after public health responses. Lessons learned from Florida could be applied proactively in other states.