In the healthcare of today, we have a parallel universe of data. Increasingly, everything we do has a data trail, with more and more records to be kept and knowledge to be gleaned. However, some activities still aren’t recorded electronically. For those still-untracked activities, we often spend extensive resources to bring them into the larger data matrix. We believe that more data will catalyze the data-information-knowledge process and knowledge holds the promise of improved outcomes. Healthcare data are among the most complicated in the world. When handled improperly, they can dangerously misdirect patient care and resources. On the other hand, when done well, healthcare analytics (the output needed to drive analysis and insight) opens a window so we can understand and improve performance—and see into the future.
Healthcare data are among the most complicated in the world. When handled improperly, they can dangerously misdirect patient care and resources.
When done well, healthcare analytics (the output needed to drive analysis and insight) opens a window so we can understand and improve performance—and see into the future.
The optimal use of healthcare data flows from identifying data needs, preprocessing data, adding value, and then applying information is illustrated in Figure 11-1. But using data in healthcare is rarely this simple. Until the last few decades, working with healthcare data had been limited to aspiring researchers and their small teams. Now, we have widespread data workers throughout the healthcare system. But data scientists and clinicians do not always understand one another: things sometimes get lost in translation. Not only does the nonclinician have the responsibility to learn clinical operations, but clinicians need to be conversant with the onslaught of technical requirements and terminology required to transform high-quality data into knowledge.
A model of data flow in healthcare.
This figure above shows the flow of data in healthcare: beginning with the patient, who is the focus of the flow; the team specifying needs is guided by a data expert; data experts, connected to the best data, determine what is possible using these data; the data experts then organize, and validate data in a reproducible manner; then the data experts assemble, visualize, and/or model data as information; and then we journey back to the team who, using the information, creates action plans and knowledge to improve care to the patient.
Poorly constructed analytics causes embarrassment, incorrect conclusions, wasted resources, and potentially patient harm. This chapter will cover ways that we can guard ourselves against bad data and what is required for a health system to develop and maintain reliable data analytics.
PART 1: FUNDAMENTAL ISSUES IN HEALTHCARE DATA
Healthcare data and analytics challenges start with a major fundamental problem: healthcare has a very fragmented business model. Consider ...