Routine auditing helps medical researchers foster good data management practices

Data auditing efforts help medical science researchers improve their research data management (RDM) practices, allowing them to objectively assess their current performance, propose corrective actions for lapses, and spread awareness of data management policies, according to a recent Singapore study.

“Overall, we believe that routine data auditing has good potential in reinforcing research integrity and can be adopted by other medical institutions,” the authors said.

Through surveys with 31 researchers and 25 principal investigators (PIs) from the Nanyang Technological University in Singapore, the authors assessed the effect of routine data auditing on RDM practices. Surveys were given before and after the implementation of audits and included Likert scale items regarding their awareness of RDM, compliance with depositing data into the mandated central repository, and reception to the required data management plan (DMP).

Response rate for PIs was 100 percent. Significant differences in five of 12 items were detected after the implementation of the data auditing programme. In particular, there was a significant increase in the total score between the pre- and postauditing periods (p=0.003). [BMC Med Ethics 2021;22:103]

The impact of the routine auditing programme on researchers was less pronounced, with only one item registering a significant change. Nevertheless, total scores for researchers increased significantly (p=0.021).

Looking at the individual items, the authors noted that among PIs, the auditing programme led to significant improvements in the agreement that improper data management could cause lack of reproducibility in science, as well as in the awareness of proper RDM practices.

Moreover, compliance with the DMP likewise improved with the implementation of the audit, as did the strength of RDM mechanisms in their respective labs.

Notably, both PIs and researchers saw significant improvements in terms of depositing all data into the mandated repository.

However, when they looked at the data repositories of five audited laboratories (vs five control labs), the authors found no significant main group effect (p=0.158), suggesting that the rate of data deposition was not significantly changed by the audit. The main temporal effect was likewise not significant (p=0.642), indicating that data input had also not changed over time.

Though both researchers and their PIs felt that they had become more compliant to the data deposition mandate, “[d]ata auditing did not affect data deposition rates significantly, which could be a consequence of small sample sizes and data production patterns,” the authors said.

“For future research, we can evaluate the effectiveness of data auditing on a larger scientific research community and explore other aspects of compliance, such as developing metrics for RDM,” the authors said. “It would be useful to disseminate findings from this study to research integrity policymakers and researchers to increase recognition of data audit as a tool to promote good RDM.”