Major events and surgeries are not the only sources of trauma during a hospital encounter. Many small, less invasive events such as shots, line placements, blood draws, and imaging studies happen throughout a patient’s hospital stay. Many of these less traumatic events have the potential to negatively impact patient outcomes by increasing the risk of hospital-acquired infections
through skin invasions and exposure to organisms, reducing the patient experience by causing pain and frustration, increasing cost and causing other complications.
This use case provides insight into how Fusion Consulting helped Loma Linda University Health Systems reduce such events when they are not clinically required. We identified hospital-acquired
infections as a key clinical event that affect clinical outcomes. To address hospital-acquired infections, we created an analytic program called Prevent Pain and Organisms from sKin and catheter Entry and Radiology (POKE-R).
Below we will identify the data points for analysis and the outline some of the technical challenges, solutions and results documented from the analytic program.
• Chief Medical Officer (CMO)
• Chief Nursing Officer
• VP of Patient Care
• Director of Clinical Quality
• Inpatient Clinical Services Director
Every time a line, drain, tube or airway (LDA) is placed in a patient, every time blood is drawn from a patient, every time medication is administered to a patient, there is increased risk for hospitalacquired infection, increased pain to the patient, and increased opportunity for blood loss. This can reduce both clinical outcomes and patient experience. The problem is amplified in neonatal and
pediatric patients where it is more challenging to place a line and where even a small amount of blood loss can cause complications.
Taking further steps, we identified which clinical events are actually POKE-R events. We then monitored patients for the count of POKE-R events and we analyzed upcoming scheduled events. The
goal is to allow the clinician to reduce the number of POKE-R invasions performed.
We have built an enterprise data warehouse using software provided by the EHR combined with custom extensions developed at Loma Linda. This data warehouse allows us to get to a substantial
amount of detailed information which is conformed across the hospital encounter. Next, we built custom reporting tables or views for each of our programs. This enabled simpler reporting and better performance. We faced several challenges in this but worked to overcome.
For POKE-R only, two extractions were added specifically for this project. We added an extension and modified EHR workflow specifically for Lines, Drains, and Airways (LDA) to know how many attempts the Line, Drain, or Airway placement took. Furthermore, physician-performed LDAs such as central lines were documented in a different manner so we created a special extract to get the placement times and attempts. Finally, it was not enough to know when a specimen was taken. We needed to know which procedure orders shared blood draws and which required separate blood draws. If 5 lab draws show the same collection time, it is important to know whether they were separately drawn, or all of the tests used the same blood collection.
With these extensions, all of the data needed to mine the POKE-R information was available in the data warehouse. However, before we could search for the POKE-R events, we had to configure
which events were defined as POKEs. We did not want to hard-code this information and we did not want the information determined or maintained by IT personnel as it is clinical in nature.
Therefore, we established an interface to configure POKE-R.
We needed to define every event which was a POKE-R event and whether it was painful. This needs to be configured using attributes of the data elements. The following attributes were
identified by the clinician as identifying POKEs:
1. Medication Administration: Route and Administration Event
2. Lab Test: Specimen Type and Specimen Source
3. Procedure Order: Type and Code
Additionally, the presence of a line or drain prior to the event can impact whether the event is a POKE and whether it is painful. For example, blood tests and medication administrations are
considered non-painful if they use an existing line. A urine sample is not a POKE at all unless there is a catheter used to obtain the specimen.
We created a simple secure interface for the Patient Safety and Reliability leadership to provide and administer this clinical information. This interface contains the data points listed above
prepopulated from the actual clinical data warehouse. The user can then choose which values for each data point indicate a POKE and can combine data points.
Another thing that was very important was to determine the scheduled POKE-R events. Our goal was to show the clinician the upcoming POKE-R schedule so that treatment could be altered to
reduce the POKEs. To do this we brought in every scheduled medication administration, procedure, surgery, image or lab test.
Through analytics we successfully facilitated various goals such as enabling the clinician to reduce the number of POKE-R invasions performed, and identify which clinical events were POKE-R events to improve patient outcomes. We monitored patients for the count of POKE-R events and we analyzed upcoming scheduled events. Furthermore, we ensured that accurate and meaningful information was readily available to the appropriate personnel, including timely information to clinicians so they can ultimately alter treatment, and retrospective trend analysis to enable and track performance improvement and identify opportunities for additional process improvement.
We have evaluated early progress of the analytics in improving clinician behavior and patient outcomes. The following results vs patient encounters prior to the analytics rollout has been a
successful reduction within the hospital POKE-R events:
POKE-R: Reduction in POKE-R events in the PICU by 8.3% and cost savings of $11.1 million.
POKE-R analytics can be used in an inpatient hospital setting to help facilitate clinical decision making. Most importantly it provides the medical staff with visibility into the patients’ daily care and provides opportunities to reduce the risk of potential infection, reduce unnecessary expenditures/charges to the patient that are not clinically necessary, and improve the overall satisfaction of the patient. The same use case or program has been proven to work in pediatric critical care environments and is best suited for areas of the hospital that provide acute care services to patients.
“Improving Patient Care Through Analytics” – Paper Publication Conference Paper· September 2016
Conference: 2016 4th International Symposium on Computational and Business Intelligence (ISCBI)