Data Quality Surveillance System (DQSS): a new dimension for high data quality in prospective databases
“Garbage in, garbage out” i.e. a poor quality of data will always lead to a biased result. A reliable, valid, trusted and interpretable output can only be generated if we ensure a proper input and processing of data. Present study was undertaken to identify the possible sources of errors which compromises the data quality and to develop a model (DQSS) to address these issues.
METHODS:
DQSS for internal quality assurance is a system which regularly monitors the database followed by appropriate actions in cases required. Delphi method was used to identify the possible error points in the flow of data from collection till the freezing/locking of dataset. This was followed by the identification of solutions to prevent such biases. Before the discussion among the experts, a systematic literature review was conducted to assess the issues related with prospective data management. Interviews of data collectors, data entry operators and data analyst were also conducted to explore the gaps.
RESULTS:
The DQSS has four dimensions; prevention of wrong data capturing and wrong data entry, prevention of data leakage and appropriate action cued towards compromised data quality. For preventing wrong data capturing; understanding the dynamics of the community, training of the data collectors, healthy relationship and reputation with the informers, review of transcribed data, random domiciliary checks by the supervisors are required. Data collectors must be made to realize the importance of data generated. To prevent wrong data entry; internal quality checks, double data entry, and accountability can play a major role. Inter village along with previous months statistics need to be regularly (after every round of data entry) compared. Annual census (to capture missing information) to be followed by freezing/locking of database. Performances of all the stakeholders involved in data flow to be reflected in their annual confidential report along with regular motivation.
CONCLUSIONS:
Maintenance of high data quality is an important concern in prospective data management. DQSS is being developed and incorporated in the database of Ballabgarh HDSS. Such model can be adopted at other places where prospective database is being maintained.