April 1, 2020
Unplanned transfer to the intensive care unit (ICU) is implicated as a harbinger of morbidity and mortality in hospitalized patients. This event has been attributed to inhospital mortality rates of 18−25 percent,1-2 and when compared with patients who were able to avoid unplanned transfer, mortality can be three times greater.1
Transfer to the ICU increases hospital LOS and hospital costs. Hence, members of the operating team need to pay close attention to appropriately identifying patients who require a higher care level on admission and to intervene early in deteriorating patients.
Furthermore, transfer to the ICU increases hospital length of stay (LOS) and hospital costs.1-4 Hence, members of the operating team need to pay close attention to appropriately identifying patients who require a higher care level on admission and to intervene early in deteriorating patients. The Joint Commission has specified this occurrence a target to improve quality.1
Many professional societies, including trauma organizations, have offered ICU admission criteria. In many cases, these guidelines are obviated by the need to deliver specialized care only available in an ICU, such as mechanical ventilation. For other patient cohorts, the decision is more nuanced.3 Previous studies have identified acute respiratory failure, respiratory insufficiency, increasing oxygen requirements, and tachypnea as common diagnoses for transfer.1-5
In addition, many reports have attributed the need for transfer to medical error, including triage error.1,3-4 Rib fracture patients are one such group especially vulnerable to unanticipated transfer because of progressive respiratory failure and difficulty in triaging. Several scoring systems are available to help predict clinical deterioration, but it would seem the most important approaches include appropriate triage on admission and an effective protocol to optimize care.6-8
Geisinger-Community Medical Center’s (CMC) initial Trauma Quality Improvement Program (TQIP) report from spring 2017 indicated that the hospital was a high outlier for unplanned ICU admission with an odds ratio (OR) of 1.83. As a result, we performed an in-depth chart review of all patients transferred to the ICU the prior year. We identified 2 percent of our trauma patients who met this criterion. Of these patients, 80 percent had developed respiratory failure secondary to rib fractures requiring further interventions in the ICU, including 75 percent of those patients requiring intubation. Geisinger had no guideline driving clinical decision making for admission level of care or uniformly addressing patient needs or follow-up once admitted. Leaving the decision to the physician’s discretion resulted in unacceptable rates of transfer and patients unnecessarily exposed to risk of morbidity, mortality, increased LOS, and increased cost.
Geisinger-CMC is a community-based, 300-bed, Level II trauma center that is part of the integrated health network. The trauma team serves six mainly rural counties in northeast Pennsylvania. The trauma service cared for 1,900 patients last year.
There was no direct hospital oversight, but the health system has prioritized “five star” ratings from the Centers for Medicare & Medicaid Services (CMS). As such, support from respiratory therapy and nursing leadership was provided to help decrease LOS and mortality from unplanned ICU transfer.
Several scoring systems are available to help predict clinical deterioration, but it would seem the most important approaches include appropriate triage on admission and an effective protocol to optimize care.
Anecdotally, we had identified the need for intubation as a driving factor in transfers to the ICU. We used the Pennsylvania Trauma Outcomes Summary (PTOS) data to conduct the patient review. Our performance improvement (PI) team identified common factors believed to result in decompensation. These factors included more than three rib fractures, older than age 65, history of smoking, bilateral rib fractures, pulmonary complication (acute traumatic injury, such as pneumothorax, or historical, such as history of chronic obstructive pulmonary disease), and immobility related to other injuries. These factors were targeted based on prior scoring systems and weighted against our population of interest. Each factor was given equal weight (score of 1). This triage system allowed us to accurately predict decompensation in patients with a score of three or higher. Our PI team then internally validated the scoring system using 2015 PTOS data. The scoring system was able to accurately predict more than half the patients with clinical demise.
Triaging patients appropriately would prove to be only part of the equation. Prior evidence indicates that 48–76 percent of patients go on to clinical deterioration.1,3 As such, our physician and advanced practitioner team devised a protocol focusing on multimodal pain control and short interval follow-up with a specific focus on respiratory parameters. We engaged nursing and respiratory therapy to help assess and alert for early physiologic changes, such as poor pain control, worsening vital signs, or decline in incentive spirometry (IS). We used an aggressive education approach and real-time chart review to audit compliance.
As a result of data compiled to define the problem, we focused on three key parameters: appropriate triage, effective multimodal pain control, and restoration and maintenance of normal physiology. A new rib fracture protocol was introduced. The trauma director and senior advanced practitioner were responsible for educating team members. PI staff who were engaged in the development of the scoring system also were instrumental in ensuring understanding and compliance with the new triage criteria. Nursing leadership was included in the education, as patients would now be more aggressively triaged to a higher level of care based on risk factors.
Once admitted, patients were prescribed a multimodal pain regimen that included acetaminophen, ibuprofen (if clinically appropriate), local anesthetic (dermal lidocaine), muscle relaxant, and short-acting narcotics. Adjunctive agents, including gabapentin, clonidine, or others, were used selectively based on patient reports of pain. Patients were prescribed nebulizers to optimize pulmonary hygiene and were given respiratory therapy-directed instruction on IS use. Patients were expected to achieve 80 percent of predicted IS target based on age and height per the IS nomogram provided. Patients who fell below this goal were triaged to more aggressive pulmonary interventions prior to transfer. Attention was paid to more invasive pain control, including IV narcotics, IV lidocaine, or epidural catheter placement. Patients who still fell short of this threshold were considered for ICU transfer and/or operative rib fixation if appropriate.
With respect to pulmonary hygiene and physiology, we quickly identified challenges with respect to IS understanding and use. We enlisted the help of nursing and respiratory therapy to design an educational tool to share with patients to understand and track the appropriate use of the device. Nursing documentation of IS in the electronic health record (EHR) was expected every four hours despite more frequent use. Common pitfalls were that the spirometer often was moved outside arm’s reach of the patient, and that new staff were unfamiliar with how to identify a predicted IS goal.
Protocol development and education included three physicians and two advanced practitioners. Two PI integration specialists were instrumental in identifying risk factors to inform the scoring system. The team included one respiratory therapy lead and two PI case managers to help with training staff and real-time chart review. No additional costs were incurred.
Protocol development and implementation took approximately 12 months. We introduced each step incrementally, first focusing on appropriate triage, next on multimodal pain control, and finally on pulmonary hygiene and improved documentation. Our primary outcome was rate of unplanned ICU admission. In one year, unplanned OR transfer decreased from 1.83 to 1.21 (fall 2018), with incremental improvement each quarter (see Figure 1).
Figure 1. Odds ratio for unanticipated transfer to ICU
Because of the increase in required EHR documentation, rates of documentation compliance initially were low. Real-time chart review and engagement from clinical leaders on our trauma floor helped with improved documentation.
We observed a 66 percent reduction in odds of ICU admission—roughly equivalent to 10 fewer ICU admissions. We previously identified that 75 percent of our transfers would need mechanical ventilation, thus we would anticipate that four to five patients would require mechanical ventilation. Using ICU cost data from Dasta and colleagues,9 we estimated cost of care preintervention averaged $430,612 for the ICU phase of care. Postintervention, we estimated mean cost to be $170,801—a savings of $259,811.
Unplanned ICU admission is multifactorial. Geisinger-CMC found our unplanned transfers almost exclusively were the result of respiratory decompensation in patients with rib fractures. Other centers may find other factors are the main cause. We encourage other hospitals to drill down on each event to understand contributing factors in each unique population.
To sustain this momentum, we have incorporated this review into our primary and secondary PI meetings. In our primary PI review (daily), we review the rib score to determine appropriate level of care on admission and ensure multimodal pain control. If patients are inappropriately admitted to a level of care, we resolve that situation immediately with early interventions or planned transfers to optimize care prior to decompensation.
Ongoing education of patients and staff is the foundation for ongoing health care improvements.
We use the secondary PI review to ensure ongoing documentation compliance. If staff fail to comply with documentation, we have an escalation policy, including direct nursing communication to rectify the issue, secondary communication to nurse manager, and final communication to senior nursing leadership. Many issues are resolved at the primary level, and most remain at the nurse manager level.
Ongoing education of patients and staff is the foundation for ongoing health care improvements. We have incorporated lectures every three weeks for nursing staff in our high-acuity units to reinforce the importance of these initiatives and any changes that may result from ongoing reviews.
The authors would like to acknowledge and thank Richard Lopez, DO; Fred Leri, PharmD; Jess Sheridan, BSN; and Wayne Williams, RRT, for their instrumental work in chart review, protocol creation, and education.