• Guy Tsafnat

Warehouses Episode 3: Nobody wants your data extracts

The third mistake we cover in our series on clinical data warehouses, is also the most common one. Creating clinical data warehouses that serve database managers by being an easy way for them to provide data extracts to users, but users don't want data, they want answers to clinical questions.

The total user experience for your clinical data warehouse will determine how fast patient care can improve in your hospital and beyond it. It is tempting for database managers to provide subsets of data in the form of an extract but these only move the problem downstream to the clinicians. If you think that data extracts are what they need to be able to improve care, then think again. Consider:

  • Data extracts are a security risk that is exponentially hard to manage as more and more copies of the data are created.

  • Data extracts may be acceptable to some users who have no other choice but the reality is that the large majority of these data extracts are thrown into the “too hard” basket leading to waste and frustration.

  • Giving data extracts to data scientists to conduct research is slow and inefficient as there are far too many questions and far too few data scientists.

Instead, grant clinicians access to data:

  • via a browser using tools that enforce access restrictions set by the hospital, IT and/or data custodian, and

  • Provide a set of interoperable tools that are flexible enough to ask a variety of questions, but are specific to healthcare and already speak the “language” of clinicians.

Further reading: Pitfalls in developing HealthCare Data Warehouses

#data #bigdata #datawarehouse #clinicaldata #datamodels #research #automation #ohdsi #cdm #medicalresearch #ai #sysadmin #saas #isaas #datacenter #analytics #ml #researchwaste