“Data integration is easier said than done. Data isn’t always clean. Human users don’t always use systems in the way they were designed to be used. There are also discrepancies in the data, the little one-offs that make integration a challenge.”
KRISTI PAYNE
CHIEF OPERATIONS OFFICER
MDABSTRACT
As noted in a Stanford Medicine research report, Harnessing the Power of Data in Health, the ability to integrate disparate sources of data, both clinical and nonclinical, will help fuel medical research, inform value-based care models, improve the patient experience, and define population health initiatives. Proper data integration will remain an integral part of providing the highest quality patient care, ensuring that providers have the data they need to make informed clinical decisions at the point of care. 1
“Data integration is easier said than done,” said Kristi Payne, Chief Operations Officer at MDabstract. “Data isn’t always clean. Human users don’t always use systems in the way they were designed to be used. There are also discrepancies in the data, the little one-offs that make integration a challenge. You need to put a strategy in place that can meet those challenges, day in and day out, because the data never stops coming in.
Payne recommends three best practices for successful data integration:
1. KNOW WHERE YOUR DATA ORIGINATES
With so many different data sources — and new ones added all the time — Payne said it is crucial that healthcare organizations start data-integration planning by taking a comprehensive digital census.
“To start, you need to know exactly where your data lives,” she said. “You need to know how good the data is, whether it’s clean, valid, and current. How is it captured? Which avenues is the data currently coming from? It’s important to be honest with yourself so you can map out a current-state document. Without that, it will be impossible to know where you might find gaps that will impede your
organization’s ability to put all the information you need together into the patient chart.”
2. KNOW HOW YOUR DATA NEEDS TO BE USED
3. ADD A MANUAL COMPONENT
Many times relying solely on electronic strategies for data migration will leave significant gaps, so Payne highly recommends adding manual interventions to your data integration plan. Too often, Payne said, data that is imperative to patient clinical records and the health system is not ported into discrete data fields or template grids inside the EHR. This is where manual data capture can help.
“Take interfaced lab results,” said Payne. “Maybe the primary care provider didn’t order the test, but it contains pertinent data about health conditions the provider manages. The lab results might just be moved to the chart under a generic labs category, or worse, routed to an inbox or work queue to be filed in a chart,” she explained.
“Digging through folders or huge inboxes to find results is really overwhelming for providers and makes data mining and outcome tracking almost impossible. What’s needed is for those results to be correctly attributed to the patient, correctly routed to the appropriate provider, and then for the data points to be discretely entered in a lab grid or clinical context so they can be referenced for patient care,” she said.
Partnering with a manual, discrete data service to close integration gaps, in conjunction with an electronic strategy, can help ensure the right data is getting to the right place without placing an undue burden on physicians or other clinical staff. It is also the only way to accurately move narrative text fields — “those real nuggets of clinical information that help remind providers of ongoing clinical processes” — from system to system, according to Payne.
“A manual data integration method provides a validation strategy that ensures your data is accurate, complete, and your electronic routing works the way it should,” she said. “Manual integration allows the chart to stay cohesive and avoid gaps in clinical data, and makes sure that vital patient data is getting to where it needs to be — at the provider’s fingertips at the point of care.”
“Digging through folders or huge inboxes to find results is really overwhelming for providers and makes data mining and outcome tracking almost impossible. What’s needed is for those results to be correctly attributed to the patient, correctly routed to the appropriate provider, and then for the data points to be discretely entered in a lab grid or clinical context so they can be referenced for patient care.”
KRISTI PAYNE
References:
1 Stanford Medicine. Stanford Medicine 2017 Health Trends Report: Harnessing the Power of Data in Health. June 2017. https://med.stanford.edu/content/dam/sm/smnews/documents/StanfordMedicineHealthTrendsWhitePaper2017.pdf.