A study from the University of Texas Southwestern shed valuable light on the question of whether larger pools of data always provide more answers than smaller pools of data.
As HealthITAnalytics notes, the study was conducted using electronic health record data collected from 33,000 patients in 75 regional hospitals during 2009 and 2010. Of these 33,000 patients, 12.7 percent were readmitted to a hospital within 30 days of being discharged from their first stay.
What the study found was that data collected over the course of his or her hospital stay, such as hospital-acquired infections and the patient's health stability at his or her discharge was not significantly more accurate at predicting 30-day readmission than data collected upon a patient's initial hospital admission, such as his or her health literacy, financial challenges, and behavioral health issues like smoking and dietary habits.
The researchers initially hypothesized that the data collected throughout each patient's hospital stay would better identify whether he or she would return within 30 days because conventional wisdom says that more information gives a more accurate story. When the results showed that this was not the case, they were surprised.
But what else do the results show? How can a limited amount of data enable us to make similar predictions to a larger collection of data on the same topic? The answer is the quality of the data. Bigger is not always better, and when there is too much irrelevant information available, the data that actually matters can become buried and potentially overlooked.
Healthcare providers can use this data to target their most at-risk patients and work with them to reduce their likelihood of readmission. For instance, other studies about patient readmission support the finding that better planning and communication during transitions of care could lead to a decrease in 30-day readmission. A separate study published in the American Journal of Managed Care discussed how a patient's cognitive state during his or her time in the hospital has a direct correlation to whether he or she will be readmitted.
This research carries over to the business world, too. To determine which sets of data to use to create new policies or draw other conclusions, it’s important to learn how to differentiate between the data that has predictive value and the data that is simply “noise.” Using data without determining whether it is truly useful for your purposes can lead to inaccurate and irrelevant conclusions.
In a hospital setting, these conclusions can put patients at risk of not receiving the care they need. In a business setting, it can mean time-consuming, expensive projects that do not end up boosting profits or helping the company in any way.
This may change over time as Big Data analytics tools become more robust, but even then, it’s important to remember as this study shows that more data is not always better. It’s one thing to have the data in the first place—but it’s another thing entirely to know how to use it.