Table 2 Lessons learned by the GRAPH network on data management

Data Collection

− Quality control at the data cleaning step is primordial to preventing downstream problems.

− Quantitative evaluation and constructive feedback to data providers improves data quality.

Data Storage

− Data files should systematically be archived, and file names standardized, searchable, and dated.

− Analysts need to be trained to foresee data infrastructure issues, such as how to avoid problems with large data file formats.

− Clear separation between original data, cleaned data and analytical results is key for reproducibility.

Data Use

− Data utilisation approaches must involve and be agreed to by the data provider

− Studies must be designed with clear reasoning for how it serves the community from which the data originate.

Results Dissemination

− Dissemination of data and/or results of analyses requires large levels of mutual trust and meaningful collaboration between the Network and the data providers.

Data Analysis

− Data analysis and interpretation must be tailored to the context of each country or region’s specific situation.

− Source code versioning and review are key tools for the development of correct and well documented code.