HIMSS14: Harnessing analytics to identify HACs, predict readmissions
ORLANDO—Starting Oct. 1, acute care hospitals with the highest number of acquired infections (HACs) will see their Medicare payments reduced by 1 percent. With this reality looming, leaders at Chicago-based Northwestern Memorial Hospital shared their efforts to harness data analytics and workflow tools to more accurately identify and report on safety and quality incidents at the Health Information and Management Systems Society annual conference.
Under the HAC reduction program, CMS selected HACs and patient safety indicators (PSIs) that are high volume, high cost and can be prevented by following evidence-based measures, according to Paul Bradley, PhD, chief data scientist at MethodCare. Six measures from the Agency for Healthcare Research and Quality include pressure ulcer rate; volume of foreign object left in the body; iatrogenic pneumothorax rate; postoperative physiologic and metabolic derangement rate; postoperative pulmonary embolism or deep vein thrombosis rate; and accidental puncture and laceration rate. Two measures from the Centers for Disease Control include central line-associated bloodstream infection and catheter-associated urinary tract infection.
At Northwestern Memorial Hospital, leaders suspected that HAC and patient safety indicator reporting included false positives, said Kristine Green, RN, BSN, the hospital’s manager of clinical documentation and quality utilization.
But analyzing data to ascertain this was challenging. Case identification often relied on coding and billing data that lagged 30 to 90 days from discharge, and the HAC review process implemented with clinical coding and clinical documentation teams was “fragmented and frustrating for all of us involved.” It entailed an inefficient workflow between the teams utilizing email and phone, and relied on memory by clinical coding staff, she said.
The PSI review process was “even less organized,” she said, as knowledge of criteria for qualifying incidents relied on individual experience rather than a systematic process. Also, review of cases was retrospective.
The hospital thus implemented a solution to improve workflow and better identify HACs and PSIs. This solution is a pre-bill analytics tool that flags accounts for HACs and PSIs; incorporates CMS and AHRQ specifications; integrates disparate claim data to run an analytics engine; leverages workflow to review each account; reviews accounts for accuracy to reduce false positives; and leverages reporting to proactively increase patient safety.
The solution proved a success, as the hospital was able to reverse false positives for more than 200 HACs and PSIs, she said. “By automating an extremely complex process of reviews and building workflow around expertise, we are able to get the right case to the right person at the right time.”
The analytics approach also proved beneficial for predicting readmissions, said Bradley. The tool integrated clinical, financial and other data—about 7,000 to 10,000 elements per patient—and applied algorithms to identify correlations with the greatest likelihood to readmit, Bradley said.
In one example, Bradley said the team learned that while most people assume the elderly with multiple conditions are the most likely to succumb to pneumonia, the analytics tool found that in actuality males ages 25 to 25 are the most likely to readmit in the summertime—namely due to “partying,” Bradley said. “That was a hypothesis no one thought about.”
The tool also generated a list of patients at the highest risk for readmission. These names are pushed directly into clinical workflow “where it become actionable.” Appropriate outreach and interventions were tailored for this group, Bradley said.