5 must-haves for effective analytics in clinical decision-making

In healthcare, data is king. Clinicians rely on data to make informed decisions at the point-of-care and ultimately propel patients' health forward. However, the wealth of unstructured data captured by IT systems in hospitals may prove fruitless if physicians can't use it to inform care decisions in real time. Analytics must be straightforward, easily digestible and accessible to help physicians personalize and improve patient care planning.

 

Unfiltered data in the clinical space has given rise to a condition known as "alert fatigue." When a physician's exposure to frequent or excessive information thwarts his or her clinical productivity, that clinician is suffering from alert fatigue.  

 

Physicians battle excessive alerts from EHRs daily. Primary care physicians received 79.6 notifications and spent more than one hour responding to EHR alerts on average per day, according to a 2016 Medscape study. While some alerts are time-sensitive and necessary, physicians consider most notices unwarranted.Clinicians reported ignoring safety notifications between 49 percent and 96 percent of the time, according to a Harvard Medical School study.

These study findings indicate the greater number of notifications a physician receives, the greater difficulty he or she may have prioritizing tasks and making clear clinical decisions. Healthcare experts designing clinical analytics programs are challenged to create initiatives that cut through the noise and deliver data when and where physicians need it most.

What makes analytics actionable for physicians

Data is crucial for delivering high-quality clinical care. But few physicians get the right type of data at the right time. Instead, they get too much data to process it meaningfully.

The most effective and actionable analytics programs share certain characteristics. Discussions with clinical information executives and leaders at hospitals across the U.S. revealed five necessities for analytics to support clinical decision-making and improve patient quality outcomes.

1. Real-time or near-time analytics.
For physicians to incorporate analytics meaningfully into their decision-making processes, analytics must be delivered to caregivers in real-time when they are at the patient's bedside.

"Doctors don't want to get a report on what they could have done days or even hours after treatment," says Darin Vercillo, MD, a hospitalist at DavisHospital and MedicalCenter. 

Retrospective analytics is valuable in helping providers understand past cases. Evaluating past performance enables physicians to see what care tactics were most effective, or areas in which they have room to improve. But retrospective analytics are less useful in helping a physician craft a patient care plan in the moment, before the patient is discharged from the hospital.   

"If I get information three to four weeks after the fact, then the window of opportunity has come and gone," says Dr. Vercillo. Improved visibility into the patient's condition helps physicians make proactive decisions to ward off potential issues at the point of care. "We need more real-time capability ... because managing patients is what we are being graded on more and more, notreactingto issues."

2. Complete and airtight information.
Analytics engines fed poor quality or incomplete information are challenged to produce meaningful or accurate insights for caregivers. In essence, "If you put bad data in, you're going to get bad data out," says Sunil Budhrani, MD, CMO and CMIO at Innovation Health, a joint venture insurance company created by Aetna and Inova Health System in Falls Church, Va.

Many large systems moving to implement big-data analytics programs face technological hurdles, including incomplete data residing in multiple electronic health records that often cannot communicate with one another. Before hospital systems can even implement analytics tools, they often spend months or years purging their insurance claims and medical record databases of incomplete, incorrect and redundant data — a costly and labor-intensive undertaking.

Hospitals and health systems looking to implement clinical decision support systems can save themselves serious time and money by addressing the issue of incomplete data on the front end during clinical documentation. Many healthcare organizations have deployed clinical documentation improvement specialists to help clinicians improve data entry through manual overview. Other progressive institutions are leveraging analytics technology to ensure physicians' documentation accurately represents patients' medical acuity.

"In the health community, we've got to focus on collecting good, accurate information to be sure that the conclusions we make are the right ones that we act upon," Dr. Budhrani says.

Moreover, clinicians are more likely to incorporate analytics into their care planning when they trust the data is valid, without second thought.

"We found physicians are much more apt to take action and change practice patterns when they are able to view the entire data picture, which is best accomplished using multiple data tools simultaneously, like graphs paired with written text," Dr. Teplitz says.

3. Visual or graphic data representation.
Physicians are accustomed to using multiple forms of information to solve problems due to their medical training. Students are trained to analyze visual data, such as X-rays and scans, listen for auditory information, such as wheezing or coughing, feel for kinesthetic information, like swollen lymph nodes, and comprehend statistics and writing, like lab results and clinical studies.

Clinicians digest information through all types of sensory modes, and physicians often digest and glean insights faster from data that is represented visually than in its raw or statistical form. Clinical analytics programs that use visual dashboards can help relieve physician's cognitive burden and drive new insights through visual formats.

"A combination of multiple forms of data — tables, graphs, bar charts — combined with written descriptions of opportunities for improvement works best [in our clinical analytics programs]," says Donald Teplitz, MD, senior vice president of medical affairs and CMO at Good Samaritan Hospital Medical Center in West Islip, N.Y.

For example, Good Samaritan Hospital Medical Center uses an analytics program that displays data on hospital acquired infections and sepsis as tables and control charts, featured alongside best practice recommendations. The dashboard presents individualized data to each physician illustrating his or her compliance with each component in the hospital's sepsis bundle. Performance information is represented visually to show physicians how they compare to hospital, regional and national benchmarks. Written recommendations to improve individual performance metrics are included.

Since implementing the program, Good Samaritan has seen a 50 percent increase in usage order sets for sepsis by physicians in the ED, Dr. Teplitz said. This shows physicians with a more in-depth, multifaceted understanding of performance patterns are better motivated to make real changes in their behavior and treatment patterns.

Physicians are comfortable with, and may prefer using, multiple forms of information to make care decisions. But some data modes are more valuable to physicians when medical acuity or time constraints force them to make immediate decisions. Visual data is oftentimes an easier way to represent and understand information quickly than lines of black and white statistics. This rings particularly true when data is complex, rare or novel to the physician.

"Analytics becometruly valuablewhen they correctly and consistently identify problems that need to be addressed without you having to pick through mounds of information," Dr. Vercillo says.

4. Evidence-based care recommendations.
Digesting patient information through multiple data modes gives physicians better insight into the patient's condition. When using that data to build a patient care plan, physicians can benefit from clinical support tools that also offer best-practice recommendations based on evidence in the medical chart.

Analytics that predict the risk or likelihood of certain events are undeniably valuable. But knowing the likelihood of a certain event doesn't necessarily help a physician make an appropriate or effective change in their treatment plan.

For example, an analytics system may inform a rounding physician the patient before him has a 60 percent chance of readmission. Recognizing the elevated readmission risk is valuable, but it does not help the physician determine the most effective treatment response, like whether to send the patient to a short-term rehabilitation facility or whether to discharge him at all.

Analytics tools that are truly effective can support physician decision making by also providing recommendations for action based on behavior patterns and best practices, Dr. Vercillo says.

"When the results of good analytics can be coupled with evidenced or compliance-based recommendations, or even standardized order sets, then value goes to the next level," Dr. Vercillo says.

5. Include physician input.
Getting physicians to engage with new IT tools or clinical improvement programs can be challenging if they don't find it valuable. Asking physicians to guide analytics development programs is one way organizations can ensure the final product is meaningful and actionable for the caregivers who will use them the most.

MemorialCare Health System provides its physicians with analytics tools allowing them to drive their own clinical improvement initiatives. MemorialCare Physician Society consists of more than 2,000 physicians and more than a dozen physician-led Best Practice Teams who work to drive clinical outcome and performance improvement. The hospital system gives Best Practice Teams high-performance analytics solutions — data marts, dashboards, report and processing capabilities in data exploration workspaces — to support their research and development.

"Involving clinicians in meaningful ways from the beginning of the development process through to the deployment phase ensures the final product is trusted by the clinicians who will use it," says Daniel Exley, vice president of information services at MemorialCare Health System in Fountain Valley, Calif. "Putting timely, trusted analytics into the hands of physicians and their care team partners enables fact-based decision making to improve quality, reduce waste and improve the patient experience."

More articles on health IT: 

How the right data analytics diminish administrative burden on clinicians
A closer look at 5 health systems' interesting, data-driven population health initiatives
AHRQ drops 2 grants for CDS research

 

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months