With healthcare long having gone digital and away from paper and microfilm, hospitals have no shortage of data on hand. But as we recently noted, engineering good quality data sets is difficult and time consuming, so many hospitals lack actionable data that is ready for analytics.
To understand why, let’s talk about problems gathering good data.
Simply extracting data is difficult for most users. If you’re, say, a director with access privileges, then you might be able to get itemized revenue and usage reports from your automated dispensing cabinets or pharmacy information system that you can use to track billing units and revenue codes. But that’s usually the extent of the kinds of data available. And anyone lower on the organizational rung is shut out.
Depending on which EHR system you use, getting other kinds data is very difficult — especially if you want to see data on things like readmissions or length of stay. You may need someone to create a workbench report or learn how to use a tool such as Epic’s SlicerDicer. Alternately, when the EHR doesn’t have reporting functionality, or multiple EHR systems are in use, you have to request information from a third-party cost accounting tool. Usually, data must be retrieved from multiple sources — a time-consuming process that we’ll come back to in a moment — and only then might you find something that’s actually useful.
Also complicating things is the unique nature of the pharmacy, which rarely follows the larger business trend of the hospital. For example, the hospital could see a big decrease in the number of inpatients or outpatients on any given day. But if those patients have more medications, dispensation volumes could climb despite lower patient volumes, making the pharmacy a busy outlier in a relatively quiet organization. That means that hospital pharmacies are sometimes managed based on data that doesn’t apply to their service line.
Another problem: Good data relies on careful and accurate input from doctors and nurses. And sometimes, that doesn’t happen — often because EHR systems are designed with user-friendly enhancements that increase the chances of inaccuracies, leading to incomplete or missing data. It could also be flawed if the person entering the data doesn’t account for a filter that’s been activated. Or maybe a physician didn’t make note of taking a two-week vacation, making it difficult to explain the corresponding skewed data from the drop in patient volume.
What makes data good?
That said, let’s pivot to the characteristics of good data.
- It should be consistent and easily verifiable from multiple sources. For example, let’s say there’s a discrepancy between dispensation data from your pharmacy and the revenue and usage report that’s used to bill patients. Further investigation reveals that someone changed the billing unit of an antibiotic from 1 gram to 1,000 units at 1 milligram each. If you don’t catch that in your system and continue to bill it at 1 gram, the payer is only reimbursing you for 1 milligram — meaning one one-thousandth of what you should be getting paid, a discrepancy that becomes a bigger and bigger problem the longer it goes unnoticed. Being able to easily cross-reference your data is crucial.
- It should be timely. Having to pull data from multiple sources means delays — and potentially outdated data that don’t reflect what’s happening in the hospital right now. With real-time data, you get the most complete picture possible about how a drug might affect not only the hospital, but potentially thousands of patients, and pharmacy and therapeutics committee decision-making can be expedited. That adds up to time and money saved and improved medication regimens.
- It depends on accurate input by clinical staff, as explained above.
- It depends on accurate mapping, meaning there must be effective and accurate linkages across systems.
- It must have robust IT security to ensure its integrity.
What good data can enable
Once you’ve established standards for producing high-quality data, then you can put it to use for your organization. Good data can tell you where to look for problems, identify them and suggest ways to tackle them.
For example, Agilum recently had a beta site for its CRCA P&T platform, and the director of pharmacy was spending a lot of time going back and forth with anesthesiologists and surgeons about the appropriateness of a certain high-cost drug. Using Agilum’s POP-BUILDER Rx tool, we were able to identify a pattern of improved patient outcomes using the anesthetic and resolve the pharmacy director’s concerns. That saved the hospital a lot of time, resources, effort and energy.
Sometimes, as this example shows, good data can also demonstrate when it’s time to stop looking into an issue and move on to other matters.