Clinical decision-making hinges on reliable analytics backups

Many healthcare analytics professionals take backups for granted, not realizing their value until a backup failure cripples clinical decision-making.

Data analytics is a field long associated with data, dashboards, business insight and predictions in many market

sectors. Healthcare is coming around to analytics, too, as many organizations are beginning to see its value on both the clinical and business sides.

Backups, however, aren't often seen as a critical part of analytics systems. They should be.

It is the responsibility of those who manage and govern data to confirm that system or data backups are being administered.

Many, if not most, organizations would be crippled if their analytics capability (and with it their ability for informed decision-making) was to "go dark" for any extended period of time because it would take away information needed to make informed decisions. Despite this, professionals working in the field of healthcare analytics take backups for granted. Many will continue to have this attitude until they're hit with a worst-case scenario and important data and decision-making capability is lost.

Because of the critical role analytics has within healthcare organizations (HCOs), backups for related systems and data should be in the forefront of analytics professionals' minds, and a central component of an analytics strategy. This is because a well-formed backup strategy ensures that HCOs are not simply backing up their data; they are also backing up their capacity for operational and clinical insight, and ultimately, their ability to do business.

Data governance, stewardship and backups

Analytics professionals often hold data quality, governance and/or stewardship positions within their healthcare organizations. Their responsibilities extend beyond day-to-day use and analysis of data. They are responsible for ensuring that sufficient and effective measures are being taken to confirm that an HCO's analytics capability can be restored quickly and completely following a catastrophic loss of systems or data.

In the event of a catastrophic downtime resulting from the loss of a system or data storage device, system recovery does not simply consist of restoring services. All data and related content must be restored as well. A fully protected HCO means that backup policies and procedures must encompass more than just what is considered "data." A complete restoration of system capabilities requires comprehensive backup and restoration procedures that include the major components of analytics: data, content and context.

Data -- Data is the raw material of analytics. Without data, there can be no insight and no evidence for decision-making. Setting up and managing the systems and processes for backing up enormous volumes of data typically found in HCOs is most often the responsibility of dedicated technical specialists. Yet regardless of who is responsible for the technical aspects of data backups, it is the responsibility of those who manage and govern the data to confirm that system or data backups are being administered, and to check the validity of the backups.

I have seen situations where, due to some technical glitch or system setting, data was not being backed up when it was assumed to be. This can be catastrophic if found out after a loss of data. It is also essential to validate that backed up data can be restored, and did not somehow become corrupted during the backup or restoration process. Although these checks can seem like a nuisance to perform, they are a lot more convenient than learning the hard way that data is unrecoverable.

Content -- Although data is usually thought of first when discussing backups, an organization cannot neglect to back up analytics content. This consists of the myriad reports, dashboards and supplemental programming code that make insight via analytics possible. Many thousands of hours go into developing analytics tools and content within an HCO, so losing this work would mean both a loss in capability and a major loss in time and effort and, therefore, money. The backup and recovery plans of HCOs should encompass the full backup of systems such as business intelligence platforms, including both the application itself, and associated content such as reports, agents and dashboards.

What may constitute a bigger risk for loss are the volumes of programming code, such as R, SAS, Python and others, that sit on individual analysts' workstations or on second-tier development and test servers that may not be regularly backed up. Responsibility for backing up this code usually defaults to individual analysts. To address this risk, policies and infrastructure should be in place to ensure that all such content be placed on servers that are secure, protected and regularly backed up. This content should not solely reside on workstations or laptops. This is a prudent step to protect both the intellectual property itself, and financial investment in the development of the content.

Context -- In addition to the data and analytics content, there is likely a plethora of supporting documentation, such as business processes and data dictionaries, that provides context for the data and code used to drive analytics. This context is vital to analytics because without this knowledge and documentation, it can be challenging to reconstruct what specific data points represent in a clinical workflow or business process, or to understand the logic behind business rules. Even though it is important for business continuity, contextual information is perhaps at a greater risk for loss than content because it is often documented or created by business analysts during early stages in a project, and may not end up in a storage location that is regularly backed up.

Without a tested and implemented backup plan, HCOs run the risk of losing the backbone of their ability to make timely and intelligent decisions. The points above illustrate that to ensure full analytical capability is restored in the event of a disaster or major technical issue, regular backups must consist not only of an organization's data, but the analytics content (reports, dashboards and other analytical tools) and context (such as business process documentation). Finally, backup processes should be validated to ensure that all backed up content can be completely restored should that unfortunate circumstance become necessary.

About the author:
Trevor Strome, M.S., PMP, leads the development of informatics and analytics tools that enable evidence-informed decision-making by clinicians and healthcare leaders. His experience spans public, private and startup-phase organizations. A popular speaker, author and blogger, Strome is the founder of HealthcareAnalytics.info, and his book, Healthcare Analytics for Quality and Performance Improvement, was recently published by John Wiley & Sons Inc.

This was first published in March 2014

Dig deeper on Clinical data analytics software and systems

Pro+

Features

Enjoy the benefits of Pro+ membership, learn more and join.